Morphological Assessment of MACI Grafts in Patients with Revision Surgery and Total Joint Arthroplasty

Abstract

Objective

To compare the histological and immunohistochemical characteristics of matrix-assisted chondrocyte implantation (MACI) grafts between patients with revision surgery and patients with total joint arthroplasty.

Methods

Biopsies of MACI grafts from patients with revision and total joint arthroplasty. The graft tissue characteristics and subchondral bone were examined by qualitative histology, ICRS (International Cartilage Repair Society) II scoring and semiquantitative immunohistochemistry using antibodies specific to type I and type II collagen.

Results

A total of 31 biopsies were available, 10 undergoing total knee arthroplasty (TKA) and 21 patients undergoing revision surgery. Patients in the clinically failed group were significantly older (46.3 years) than patients in the revision group (36.6 years) (P = 0.007). Histologically, the predominant tissue in both groups was of fibrocartilaginous nature, although a higher percentage of specimens in the revision group contained a hyaline-like repair tissue. The percentages of type I collagen (52.9% and 61.0%) and type II collagen (66.3% and 42.2%) were not significantly different between clinically failed and revised MACI, respectively. The talar dome contained the best and patella the worst repair tissue. Subchondral bone pathology was present in all clinically failed patients and consisted of bone marrow lesions, including edema, necrosis and fibrosis, intralesional osteophyte formation, subchondral bone plate elevation, intralesional osteophyte formation, subchondral bone cyst formation, or combinations thereof.

Conclusions

MACI grafts in patients with revision and total joint arthroplasty were predominantly fibrocartilage in repair type, did not differ in composition and were histologically dissimilar to healthy cartilage. Clinically failed cases showed evidence of osteochondral unit failure, rather than merely cartilage repair tissue failure. The role of the subchondral bone in relation to pain and failure and the pathogenesis warrants further investigation.

Introduction

Articular cartilage of diarthrodial joints is supported by the subchondral bone. Between the hyaline articular cartilage and the subchondral bone, are several structures connecting these two different tissues to form a tight functional association. Together these structures have been commonly referred to as the osteochondral unit. ¹ Isolated articular cartilage defects without subchondral bone changes are common^2,3 and their lack of intrinsic healing warrants treatment when functionally symptomatic. Isolated cartilage defects not only cause morbidity comparable to advanced osteoarthritis ⁴ but can also initiate secondary osteoarthritic changes.^5,6 Moreover, cartilage defects have a propensity to progress when clinical osteoarthritis has been established. ⁷

The contribution of the subchondral bone in the development and progression of osteoarthritis has long been recognized,^1,8-10 but its role in cartilage repair techniques has only recently received more attention.^11,12

Autologous chondrocyte implantation (ACI) was first reported in 1994, ¹³ with the initial technique consisting of a chondrocyte suspension injected under a periosteal flap sutured to the surrounding healthy cartilage, known as periosteum-covered ACI (PACI). ¹³ Subsequent advancements in biomaterials allowed the replacement of the periosteum with fibrin glue–stabilized collagen and hyaluronan-based scaffolds, known as matrix-assisted chondrocyte implantation (MACI) technique. ¹⁴

While the scientific literature increasingly supports the clinical outcomes and usage of PACI and MACI techniques,^15-19 far less attention has been paid to the failures of this technique. A recent systematic review reported an overall failure rate of 14.9% in 4294 patients undergoing ACI, most commonly within the first 5 years following implantation. ²⁰ LaPrade et al. ²¹ illustrated that failed ACI is predominantly composed of fibrous tissue and fibrocartilage with variable positivity for both type I and type II collagen. This study evaluated osteochondritis dissecans lesions repaired with PACI, not MACI. Histological differences in repair tissue could be expected between these 2 techniques, as the former utilizes a cell solution covered with living tissue (periosteum) to initiate repair, whereas the latter initiates repair with a cell-scaffold construct adhered to the subchondral bone.^13,22

Subchondral bone alterations following ACI techniques have been observed,^11,12,23-26 although their clinical significance remain subject of debate. Recognizing articular cartilage and the subchondral bone as one functional unit, we investigated the histological aspects of the subchondral bone.

The aim of our study was therefore to examine the histological and immunohistological features of MACI grafts in patients with revision and total joint arthroplasty and to investigate the subchondral bone status in patients with total joint arthroplasty.

Material and Methods

Patients

Between 2000 and 2006, our institution served as the Australian national reference center for examination of biopsy tissues obtained from MACI grafts. Included biopsies were obtained either from (1) MACI grafts requiring revision arthroscopy due to graft failure or another reason requiring arthroscopy or from (2) knees that had further degenerated and had progressed toward total knee arthroplasty (TKA). The decision as to whether the graft had clinically failed and/or revision arthroscopy was indicated, or whether joint arthroplasty was indeed now the best surgical option for the patient, was undertaken by the treating surgeons.

The tissues obtained from patients undergoing repeat arthroscopy were either cartilage repair tissue biopsies or osteochondral biopsies (3 mm diameter) taken as part of a chondroplasty of symptomatic hypertrophic tissue at the graft site of patients with the primary complaint of “pain, clicking, and catching” of the graft. For clinically failed patients, osteochondral slabs including the graft area were collected as part of a TKA procedure. Patients undergoing TKA presented with recurring and worsening pain following MACI, which failed to respond to conservative treatment, with evidence of degenerative joint disease (and progression thereof) such that TKA was considered the most appropriate treatment. Patient consent for treatment and use of the biopsy for research purposes was obtained and ethical approval was given by the Human Research Ethics Committee of our institution.

Histology

Routine histology was performed to evaluate and characterize the repair tissue. Samples were placed in 4% paraformaldehyde for 48 hours prior to processing. Samples containing mineralized tissue were decalcified in 10% EDTA (ethylenediaminetetraacetic acid), pH 7. For osteochondral slabs from arthroplasty patients, multiple sections 3 mm apart through the entire graft were performed prior to paraffin embedding. Samples were sectioned to 4-µm-thick slices and routinely stained with hematoxylin and eosin, toluidine blue/fast green, and safranin O/fast green stain.

All stained samples were evaluated by bright-field microscopy to characterize tissue as hyaline-like, fibrocartilage, or fibrous tissue based on cellularity (density, shape, and presence of lacunae) and matrix appearance as previously reported by Zheng et al. ²² For osteochondral samples, qualitative analysis of the subchondral bone was performed as described by Zanetti et al. ²⁷ Bone marrow edema was defined as presence of swollen fat cells surrounded by eosinophilic staining, marrow fibrosis as replacement of fat cells by fibrous tissue, and marrow necrosis as formation of foam cells and presence of swollen fat cells with loss of nuclei. Trabecular bone structure was evaluated for pathology, consisting of necrosis (characterized as loss of nuclei), sclerosis, and repeated remodeling (characterized by bone formation with reversal lines and bone resorption with increased osteoclastic activity). Alterations of the subchondral bone, recognized as clinical conditions by Orth et al., ²⁸ were reported accordingly. The grading system as described by Aho et al. ²⁹ was applied on the subchondral bone below and distant to the repair tissue in normal tissue as control. In brief, grade 0 is characterized by a thin subchondral bone plate (SBP) with direct connections between the bone marrow and the articular cartilage via open fenestrae, grade 1 by some subchondral bone sclerosis with remaining open fenestrae, grade 2 by a distinct increase in SBP thickness, absence of fenestrae, and presence of SBP fibrillation and grade 3 by severe sclerosis and flattening of the SBP. Additionally, ICRS (International Cartilage Repair Society) II scoring was performed as previously described. ³⁰ As blinding was impossible due to obvious differences in the nature of the sample, they were analyzed twice by the same investigator, 6 months apart. The average score was taken if the difference between the 2 time points was less than 10 units. If the difference between the 2 scores was greater than 10 units, the sample was evaluated by a board-certified pathologist and consensus between the 2 evaluators was reached.

Immunohistochemistry

Sections deparaffinized in xylene, rehydrated with decreasing ethanol solutions, and rinsed in distilled water. Enzymatic antigen retrieval (Pepsin, Abcam ab128991) was performed at 37°C for 15 minutes and endogenous peroxidase activity was blocked using 3% H₂O₂ in 96% ethanol for 10 minutes at room temperature. Nonspecific binding sites were blocked with 10% fetal bovine serum (FBS) in 0.1% Triton X100 in TBS (Tris-buffered saline) for 1 hour at room temperature. Samples were then incubated at 4°C overnight with primary antibodies against rabbit anti-human collagen type I (Abcam, ab34710, dilution 1:500) and mouse anti-chicken collagen type II (Developmental Studies Hybridoma Bank, Iowa, strain II-II6B3, dilution to 2 µg/ml). Antibody was detected using anti-rabbit (Sigma A0545) and anti-mouse antibodies (Sigma A9917) diluted at 1:1000 for Col I and II, respectively. Peroxidase staining was done following the instructions of the manufacturer (Liquid DAB+ substrate chromogen system, DAKO). Sections were counterstained using Gill’s hematoxylin and saturated in lithium carbonate until they turned blue, dehydrated in ethanol, cleared in xylene, and mounted. Negative control samples were incubated in 5% FBS instead of the primary antibodies. Positive control samples consisted of an osteochondral section from a nonarthritic joint.

For evaluation of the immunostained sections, the total area of cartilage was measured using Bioquant Osteo software (v13.2.6, Bioquant, Nashville, TN). Areas of positive staining were measured using semiautomatic thresholding and the percentage of positively stained tissue, regardless of the staining intensity, was calculated.

Micro-Computed Tomography

If performed, osteochondral blocks underwent micro-computed tomography (micro-CT) scanning prior to decalcification. Samples were scanned in 70% ethanol in a micro-CT scanner (SkyScan 1176; Bruker-microCT, Kontich, Belgium) at 50 kV, 500 µA, 240 ms exposure time, rotation step of 0.3, 360° degree of rotation, and an isotropic resolution of 17.78 μm, using a 0.5 mm-thick Al filter.

Images of each specimen were reconstructed using the manufacturer’s software (NRecon v 1.7.1.0; Bruker-microCT, Kontich, Belgium). For evaluation of the trabecular bone microarchitecture, cylinder-shaped volumes of interest (VOI) with a diameter of 7 mm were selected semiautomatically to mimic core-extraction samples in the grafted area and in an area distant to the graft, excluding cysts manually, if present. The height of the cylinders did not exceed 4 mm. Bone morphometric parameters (bone volume fraction [BV/TV], BS/BV, trabecular thickness, and trabecular separation) were calculated using CTAnalyzer software (Bruker-microCT, Kontich, Belgium).

Statistics

Results are expressed as mean ± SD. Data were not normally distributed, and therefore nonparametric analysis was performed (Kruskal Wallis and Mann-Whitney U test). The Fisher exact test was used to investigate relations between demographic data. Differences were considered significant at P ≤ 0.05. All calculations were performed with SPSS (version 22.0; IBM Corp.). Post hoc power analysis was performed using the G*Power computer program. ³¹

Results

Demographics

Biopsies were available from 31 patients, including 10 undergoing total knee arthroplasty (TKA) and 21 patients undergoing revision surgery. Seventeen of the revision surgeries were performed on knee joints and 4 on ankle joints. Demographic characteristics are described in Table 1 . Micro-CT was performed on 2 clinically failed patients who underwent TKA.

Table 1.

Patient Demographics for Patients Undergoing Total Knee Arthroplasty (TKA) and Revision Surgery.

Characteristic	TKA (n = 10)	Revision (n = 21)
Age, years, mean (SD)	46.3 (5.00)	36.6 (9.43)
Graft survival time, months, mean (SD)	26.8 (25.54)	17.8 (10.3)
Sex, n (%)
Male	3 (30)	9 (43)
Female	7 (70)	12 (57)
Defect size, cm2, mean (SD)	4.3 (2.00)	4.8 (4.00)
Lesion location, n (%)
Knee	10 (100)	17 (81)
Medial femoral condyle	7 (70)	10 (59)
Patella	2 (20)	3 (17)
Trochlea	1 (10)	2 (12)
Unknown		2 (12)
Ankle		4 (19)

The average age at MACI implantation was older in the clinically failed group who underwent subsequent TKA (46.3 years) than the group who underwent revision (36.6 years, P = 0.007).

The mean graft survival time was 26.8 months for those who underwent knee replacement, which was not significantly longer (P = 0.94) than survival time for the revision patients (17.8 months). Defect sizes were on average 4.3 and 4.8 cm² for TKA and revision patients, respectively, and were not significantly different (P = 0.68). There was no significant association between gender and classification group (P = 0.697) and lesion location and classification group (P = 0.464).

Histological and Histopathological Observations

Representative histological features of the TKA patients are illustrated in Figure 1 . The predominant cartilage type in TKA was fibrocartilage (70%), only 10% had a predominantly hyaline-like cartilage and 20% had a predominantly fibrous nature. The repair tissue morphology often consisted of a mixture of areas with randomly organized fibers and areas with less birefringence. Increased birefringence coincided with more elongated cell types, and thus more fibrous nature of the tissue, although vascularization was only occasionally seen.

Figure 1.

(A) Clinically failed matrix-assisted chondrocyte implantation (MACI) graft of the medial femoral condyle (hematoxylin and eosin [H&E] stain; scale bar = 4 mm). Arrows indicate the margins of the MACI graft. Asterisk indicates the original cement line and calcified cartilage layer with the area of subchondral bone plate elevation above this line, including grafted and nongrafted area. Trabecular bone sclerosis and subchondral bone cyst formation are present. Inset 1: Cells within the deeper layers of the MACI graft were generally rounder, representing a more chondrocyte phenotype. Inset 2: Cells within the superficial layers of the MACI graft were generally more elongated, representing a more fibroblast phenotype. Inset 3: Incomplete lateral integration between the graft and the remaining cartilage. Inset 4: Incomplete integration of the graft with the underlying subchondral bone plate with fibrous tissue within the separation (arrowheads). Inset 5: Marked bone remodeling and osteoclast-mediated (arrows) subchondral bone cyst formation. Inset 6: Edge of the subchondral bone plate elevation characterized by multiple small blood vessels extending into the calcified and noncalcified cartilage. Insets 7 and 8: More extensive subchondral bone plate fibrillation and thickening within the grafted area (Inset 7) compared with the nongrafted and nonelevated area (Inset 8). All insets: scale bar = 200 μm.

(B) Collagen type II immunohistochemistry (scale bar = 4 mm). Collagen type II is distributed throughout the basal layers of the graft, except for the surface. (C) Collagen type I immunohistochemistry (scale bar = 4 mm). Collagen type I is present throughout most part of the graft, but less toward the base of the graft.

The tissue surface often showed several degrees of fibrillation ( Fig. 1 ), cleft formation, poor organization of collagen fibers, and elongated cells ( Fig. 1 ). Deeper layers similarly showed poor organization, but cells were often rounder ( Fig. 1 ) and tidemark formation was rarely observed. Integration of the repair tissue with the subchondral bone was highly variable and often a thin layer of fibrous tissue could be observed if integration was incomplete ( Fig. 1 ), separating the repair tissue from the subchondral bone. Abnormal mineralization within the basal layers of the defect was frequently observed and could more specifically be described as intralesional osteophyte formation (30%) and elevation of the subchondral bone plate (40%) ( Fig. 1 ) according to Orth et al. ²⁸ Subchondral bone cyst formation was observed in 2 patients (20%) ( Fig. 1 ). Using the grading system recently developed by Aho et al., ²⁹ the subchondral bone plate below the graft was grade 2 or 3 in all specimens and was always higher than the grade of the perigraft area (grade 0 or 1 in all cases).

In comparison with TKA patients, the different cartilage types in revision patients were more evenly distributed with hyaline-like cartilage counting for 28.5%, fibrocartilage 43%, and fibrous tissue 28.5%. Tissue characteristics were similar to TKA patients, although generally the tissue surface, basal integration of the tissue scored better and less vascularization and abnormal mineralization was seen (representative image in Fig. 2 ). From the observed differences, ICRS II scores for abnormal calcification within the cartilage defect area and vascularization were significantly different between the 2 groups ( Table 2 ), thus indicating more abnormal mineralization within the basal layers of the repair tissue in TKA patients. Excluding ankle cases retained the same significant differences, but additionally showed a significantly better basal integration in knee revision surgery compared with TKA (P = 0.014).

Figure 2.

Matrix-assisted chondrocyte implantation (MACI) knee revision surgery specimen (left to right: hematoxylin and eosin [H&E] stain, toluidine blue–safranin O stain, and Col type II stain, scale bar = 4 mm). The original calcified cartilage is clearly visible on the toluidine blue stain (asterisk) with newly formed bone above. The repair tissue contains collagen type II throughout. Insets 1 and 2: Cells are primarily round-shaped in different layers of the graft, but the matrix is poorly organized. Inset 3: the subchondral bone marrow shows bone marrow edema and mild fibrosis. All insets: scale bar = 200 μm.

Table 2.

ICRS II Scores and Percentage Col Type II and Col Type I for TKA and Revision Surgery.

	TKA (n = 10), Mean (± SD)	Revision Surgery (n = 21), Mean (± SD)	P
Tissue morphology	48.3 (± 24.2)	48.8 (± 25.6)	0.69
Matrix staining	50.0 (± 27.8)	46.5 (± 31.7)	0.82
Cell morphology	49.0 (± 28.1)	43.2 (± 31.3)	0.62
Chondrocyte clustering	60.3 (± 31.2)	73.3 (± 28.5)	0.28
Surface architecture	30.3 (± 37.4)	58.6 (± 37.6)	0.14
Basal integration	70.9 (± 32.0)	90.0 (± 21.2)	0.09
Formation tidemark	24.7 (± 27.5)	16.9 (± 24.8)	0.32
Subchondral bone abnormalities	49.7 (± 35.8)	65.3 (± 30.7)	0.37
Abnormal calcification/ossification	35.3 (± 44.5)	91.5 (± 22.3)	<0.01*
Vascularization	56.8 (± 37.3)	81.3 (± 33.4)	0.04*
Surface/superficial assessment	30.0 (± 30.5)	53.9 (± 34.5)	0.12
Mid/deep zone assessment	53.3 (± 27.2)	63.8 (± 23.9)	0.31
Overall assessment	47.0 (± 25.4)	55.8 (± 24.5)	0.35
Col type II	61.0 (± 33.4)	44.2 (± 37.9)	0.37
Col type I	52.8 (± 37.5)	60.238 (± 35.6)	0.69

ICRS II = International Cartilage Repair Society II; TKA = total knee arthroplasty.

^*Significant differences.

Interestingly, the average percentage of collagen type 2 was higher in the TKR group (61.0%) compared with the revision group (44.2%), although the difference was not significant ( Table 2 ). The reverse was similarly true for collagen type I. For TKA patients, 70% had a ratio Col II %/Col I % higher >1, compared with 38% in revision patients.

Using the qualitative evaluation criteria of the subchondral bone as outlined by Zanetti et al., ²⁷ all TKA patients had various extents of subchondral bone marrow edema, necrosis, and fibrosis, albeit minimal to mild in most cases. Trabecular abnormalities were also frequently present with predominantly trabecular necrosis, increased bone remodeling as evidenced by new bone formation, reversal lines, and osteoclastic activity. All revision surgery patients had bone marrow fibrosis, which varied from minimal to marked and only a small number had some bone marrow edema and necrosis. Trabecular bone changes consisted primarily of trabecular necrosis, and there was only minimal bone remodeling in a few patients.

Comparison of MACI Histology between Anatomical Sites

The prevalence of graft tissue type, regardless of undergoing arthroplasty or revision surgery, in the different anatomic locations is illustrated in Figure 3 and the ICRS II tissue morphology and overall scores, and percentage collagen types I and II in Figure 4 .

Figure 3.

Repair tissue type distribution within different anatomical locations (MFC = medial femoral condyle). All patellar grafts consisted of primarily fibrous tissue whereas none of the ankles contained fibrous tissue. The majority of MFC specimens consisted of fibrocartilaginous tissue.

Figure 4.

Average (±SD) ICRS II scores for tissue morphology and overall assessment and percentage collagen type II and collagen type I in 4 different anatomic locations (MFC = medial femoral condyle). Tissue morphology and overall assessment scores were highest for ankle grafts and lowest for patellar grafts and were significantly different from each other. There was significantly more collagen type II in MFC graft tissue compared to patellar graft tissue.

Knee Joints

Within the knee joints, patients with lesions on the patella scored significantly worse than patients with lesions in the medial femoral condyle (MFC) for tissue morphology, fill type, deep zone and overall assessment. Fibrous tissue was present in all patients (5/5) undergoing MACI treatment for patella lesions (Figs. 3-5), but only 2 out of 5 patients underwent joint replacement surgery. Collagen type I accounted for more than 95% of the repair tissue in all patella samples and less than 5% was collagen type II. Subchondral bone pathology was rarely present ( Fig. 5 ).

Figure 5.

Clinically failed patella matrix-assisted chondrocyte implantation (MACI). (A) Hematoxylin and eosin (H&E) stain, overview (scale bar = 4 mm). Arrows indicate the margins of the MACI graft. Marked surface fibrillation and clefting and poor lateral integration of the graft with the surrounding cartilage are present. Besides a small subchondral bone plate erosion (asterisk) and intralesional osteophyte (arrow head) at the junction of the graft tissue with the surrounding cartilage tissue, there are minimal changes within the trabecular bone. Inset 1 (scale bar = 200 μm): Graft tissue consisted primarily of high-cellular fibrous tissue with an abundance of blood vessels throughout. (B) Scale bar = 4 mm. Collagen type II immunohistochemistry. Minimal collagen type II is present and is primarily localized within the base of the repair tissue. (C) Scale bar = 4 mm. Collagen type I is present within all layers of the graft tissue.

Ankle Joints

Ankle revisions contained overall better quality of tissue with none of the 4 patients receiving MACI treatment developing fibrous repair tissue. Ankle revisions scored significantly better in cartilage parameters (tissue morphology, P = 0.018; deep zone assessment, P = 0.047; and overall assessment, P = 0.018) compared with knee revision surgeries, but worse in basal integration (P = 0.005) ( Fig. 6 ) and subchondral bone health (P = 0.039) ( Fig. 6 ). The subchondral bone pathology in ankle patients was characterized primarily by marked bone marrow fibrosis ( Fig. 6 ).

Figure 6.

Revision ankle matrix-assisted chondrocyte implantation (MACI) graft. Hematoxylin and eosin (H&E) stain, scale bar = 3 mm. The graft surface is smooth and cells are primarily round shaped throughout the entire graft (Insets 1 and 2). Cell clustering is present at the surface. Inset 3 (toluidine blue–fast green staining) Clefting at the base of graft tissue (asterisk) near the tidemark. The subchondral bone plate shows marked proliferative islands within the calcified cartilage layer (arrows). Scale bar insets = 200 μm.

Age at implantation was negatively associated with the score for chondrocyte clustering (ρ = −0.622, P = 0.001) and abnormal calcification (ρ = −0.511, P = 0.03) when all samples were included. A negative association between age at implantation and chondrocyte clustering (ρ = −0.577, P = 0.010) persisted when TKA patients were excluded. MACI survival time was positively associated with the formation of a tidemark (ρ = 0.542, P = 0.037) when all samples were included. For revision patients subchondral bone health was positively associated with chondrocyte clustering (ρ = 0.790, P = 0.020) and basal integration (ρ = 0.713, P = 0.031).

Subchondral Bone Structure of MACI with TKA

Patients who underwent TKA had specimens of subchondral bone tissue, which enabled the assessment of the subchondral bone structure. One specimen was characterized by marked subchondral bone plate elevation and subchondral bone cyst formation in the grafted area ( Fig. 7 ) and the other one by subchondral bone plate erosion ( Fig. 7 ).

Figure 7.

Micro-computed tomography (micro-CT) images (scale bar = 4 mm). (A) Marked subchondral bone overgrowth and 3 subchondral bone cysts are present. (B) The grafted area contains subchondral bone plate erosion with a steep sharp osseous rim (osteophyte) protruding toward the cartilage.

Bone morphometric parameters showed a 2.2-fold increase in bone volume percentage in the grafted area, characterized by increased trabecular thickness and decreased trabecular separation ( Table 3 ). The Structure Model Index was closer to 0 in the grafted area indicative of trabeculae to be more plate-like than rod-like ( Table 3 ).

Table 3.

Bone Morphometric Parameters for the Trabecular Bone in the Graft and Perigraft Area.

Specimen	BV/TV (%)	BS/BV (1/mm)	Tb.Th (mm)	Tb.Sp (mm)	Tb.N (1/mm)	SMI
MFC, 55 year old female, 72 months post-MACI graft area	34.4	15.7	0.22	0.44	1.60	0.6
MFC, 55 year old female, 72 months post-MACI peri-graft area	15.5	25.0	0.14	0.65	1.11	1.2
MFC, 46 year old male, 43 months post-MACI peri-graft area	22.3	20.9	0.17	0.50	1.30	1.2

BV/TV = percentage bone volume; BS/BV = bone surface/bone volume ratio; Tb.Th = trabecular thickness; Tb.Sp = trabecular separation; Tb.N = trabecular number; SMI = Structure Model Index.

Power Calculation

A 1-sided post hoc power analysis using the G*Power computer program ³¹ indicated that a total of 102 specimens would be needed to detect medium effects (d = 0.51) with 80% power using a Mann-Whitney U test with α at 0.05.

Discussion

The data of the present study show that there was no significant difference in histological features of MACI grafts between patients with revised surgery and arthroplasty. The majority of cases in both groups contained fibrocartilaginous repair tissue, although hyaline-like tissue was more frequently observed in the revision group. Clinically failed TKA patients were significantly older, had more abnormal calcification and vascularization within the graft tissue compared with patients undergoing revision surgery. The observed subchondral bone pathology of differing degrees was present in all TKA patients, and hence we propose that the mechanism of failure is of the entire osteochondral unit and joint, rather than just failure of the cartilage repair tissue. The MACI grafting procedure succeeded in generating and maintaining repair tissue within symptomatic cartilage defects as evidenced by histology but failed to preserve joint function in patients therefore proceeding to TKA.

Other studies have observed similar graft tissue to this study following failed cartilage restoration procedures. For ACI procedures, LaPrade et al. evaluated repair tissue of 6 periosteal-ACI failures for the treatment of osteochondral dissecans. ²¹ All specimens were characterized as primarily fibrocartilaginous tissue with variable amounts of type II and type I collagen. The failure mode however differed, with patients experiencing pain following dislodged or partially dislodged graft tissue.

In contrast to our study, LaPrade et al. ²¹ reported a ratio of Coll II %/Col I % of <1 in all ACI patients, whereas we found this ratio <1 in only 30% of the cases. A minimal degree of graft integration with the subchondral bone may be needed to form or maintain type II collagen as nondislodged tissue in the microfracture-treated specimens in the study by LaPrade et al. ²¹ formed more collagen type II compared with collagen type I. This is further supported by findings of Nehrer et al. ³² who similarly demonstrated fibrous tissue in detached repair tissue compared to more hyaline-like tissue where the PACI graft had bonded with the subchondral bone. Nehrer et al. ³² reported fibrous tissue as the most common type in PACI, accounting for 60% of the cross-sectional area. ³² None of the patients in our study were evaluated because of graft detachment, a failure mechanism now largely avoided by improved fixation technique with fibrin glue and collagen matrix. Differences in ACI technique (MACI vs. PACI) may account for different failure mechanisms, but tissue characteristics do not seem to differ. Tissue characteristics in “failed cases” may also not depend on treatment method as Kaul et al. similarly found fibrocartilaginous tissue following bone marrow stimulation procedures in patients with early osteoarthritis. ³³

We found fibrous tissue in all patients undergoing MACI treatment for patella lesions and although considered inferior quality tissue, it was interesting to note that only 2 out of 5 patients underwent joint replacement surgery. While a certain degree of poorly regenerated tissue thus seems to be acceptable in the patellofemoral joint, it is also probable that surgeons are less likely to offer TKA for patellofemoral lesions, a selection bias evident in this study.

Of interest, despite the fact that more revision patients contained hyaline like repair tissue, 62% of revision patients had a Col I%/Col II % >1 compared with 30% of TKA patients. In many instances, specimens from revision patients consisted of biopsies from the surface of the repair site. It is unclear whether this hypertrophic surface represents a tissue composition indicative of the repair as a whole, or only the upper portion of the repair. We frequently observed increased levels of type 1 collagen toward the surface of the repair tissue. In our study, patients undergoing TKA were significantly older than patients undergoing revision surgery. The question whether age is a limiting factor for MACI remains controversial. Some studies have found a negative association of age with outcome,^34,35 while others have found no such association.^36-38 Recent guidelines advise on careful selection above the age of 50 years. ³⁹ Our study included 5 patients ≥50 years (3 TKA patients 51, 51, and 52 years and 2 revision patients 50 and 50 years).

The effect of preexisting osteoarthritis on MACI treatment could not be examined in our study due to its design; however, ACI in the presence of osteoarthritis is controversial in the literature.

Minas et al. ⁴⁰ reported an overall failure rate of 12 out of 155 knees in 153 patients (8%) with early osteoarthritic changes undergoing ACI, selected with the specific aim of delaying arthroplasty surgery. Arthroplasty in failed cases was performed at an average of 38 months after ACI, which was longer compared with our study. Interestingly, almost half of the patients receiving ACI in the authors’ series included patients with early osteoarthritic changes (153 out of 328 patients with follow-up over 2 years) indicating a greater tendency to perform ACI in early arthritic knees than in other series. Morphological and qualitative information on the graft tissue was not given. Moreover, Hollander et al. ⁴¹ did not find a negative effect of osteoarthritis on graft repair, but on the contrary osteoarthritis was shown to enhance tissue regeneration. Despite these interesting observations, Filardo et al. ⁴² observed a failure rate of 27.5% in 44 patients with osteoarthritis (Kellgren grade 2 and 3) treated with ACI and suggested that once the osteoarthritic process had started, the potential of ACI is compromised regardless of the severity of the joint condition.

The observation of different degrees of subchondral bone pathology in all clinically failed TKA patients, significantly worse abnormal calcification and vascularization in TKA patients are interesting and could suggest subchondral bone and endochondral ossification may play a pivotal role in clinically failed patients.

In recent years, the importance of the subchondral bone in cartilage repair strategies has gained interest.^11,12,24,43 Despite the frequent observation of subchondral bone alterations and pathology, ²⁴ its clinical significance and correlation with pain remain controversial, while in osteoarthritis the subchondral bone is a well-recognized source of pain.^1,8-10 We observed subchondral bone changes in all patients that underwent TKA due to clinical failure. The changes were often localized and more extensive below the graft tissue, although subchondral bone plate elevation also extended beyond the grafted area. Vasiliadis et al. ²⁴ reported intralesional osteophyte formation in 64% of patients and subchondral bone cyst formation in 39% of patients, 9 to 18 years following ACI treatment, but failed to find significant correlation between clinical outcome and subchondral bone abnormalities. However, they considered its occurrence a negative prognostic factor. ²⁴ Impaired bone and cartilage regeneration processes, altered biomechanical loading, disturbed mechanisms of cartilage–subchondral bone crosstalk, and pathological vascularization or angiogenesis have been proposed as potential causes. ²⁸ Although it has been proposed that overgrowth of the subchondral bone leads to narrowing of the repair tissue, ²⁸ this was not our observation, as the repair tissue was always at least as thick as the nongrafted cartilage, but repair tissue protruded above the normal position of the cartilage surface. More research needs to be undertaken to investigate the pathophysiology and clinical significance of the subchondral bone pathology.

Patella repair tissue had the worst tissue composition. In the original report of the MACI technique, Brittberg et al. ¹³ similarly reported significantly worse clinical and morphological outcomes for patients with patella lesions. Adressing patella malalignment has led to improved outcomes, similar to other anatomic locations. ⁴⁴ A certain degree of patella malalignment could not be ruled out in our cohorts.

Ankle MACI grafts in our study performed overall better and none of the specimens had fibrous repair tissue. Other studies have similarly reported good to excellent outcomes and a negative association with age >40 years.^45,46 Three out of 4 patients in our study were younger than 40 years. Interestingly, Dixon et al. ⁴⁵ observed a high proportion of patients with magnetic resonance imaging evidence of complete defect filling, complete integration of borders, an intact graft surface and homogeneous signal within the graft in patients with persistent pain, implying that failure to relieve pain may not necessarily indicate technical failure of the procedure.

Our data suggest that repair tissues in revised MACI repair and clinically failed MACI are similar. However, this study has several limitations and conclusions need to be approached with caution.

Because of the retrospective nature of the study, the number of specimens in each group could not be controlled for and limited statistical power because of small sample size may have contributed to limiting the significance of some of the statistical comparisons. Specimens from revised MACI grafts contained several biopsies from the surface of the graft site. This tissue may not represent the composition within the entire graft. Similarly, 3-mm diameter osteochondral samples from the center of the defect may also not represent the entire graft. Indeed, similar to Laprade et al., ²¹ we found large variation in tissue composition throughout the entire sample in the knee arthroplasty slabs.

Patients classified as clinical failures may be completely or partially a consequence of other pathologic changes within the joint and thus sources of pain not directly related to the MACI treatment. In other words, clinically failed cases may contain biologically and functionally adequate repair tissue, but the repair tissue failed to prevent degeneration of the entire joint. After all, cartilage does not contain nerve endings and pain can persist despite excellent cartilage defect infill.^45,47

Although the histological and immunohistological evaluation of biopsies yield important data regarding the quality of repair tissue, their correlation with clinical symptoms and function remains unclear. Intuitively, tissue with characteristics closer to native hyaline articular should be superior in function and more durable, but frequently fibrocartilaginous repair tissue seems sufficient. Hollander et al., ⁴¹ for example, collected 2mm diameter biopsies of 23 patients undergoing follow-up arthroscopy for nonclinical reasons following ACI procedures and found that 43% of patients had fibrocartilaginous tissue 6 to 25 months after the ACI procedure. Peterson et al. ⁴⁸ found primarily fibrous tissue in biopsy samples of 4 out of 12 patients with good to excellent clinical grading at a mean follow-up of 54.3 months.

Similarly, Saris et al. ¹⁷ evaluated 116 biopsies in a prospective randomized trial comparing MACI with microfracture. Two years following treatment both treatments showed very good structural repair and similar mean ICRS II overall assessment scores (±63.8). Despite similar histological appearance, MACI-treated patients performed clinically significantly better than microfracture-treated patients. ¹⁷ The mean overall assessment scores in revised MACI in our study (55.8) was lower, although similar to MACI treated patients in the study by Saris et al. ¹⁷ It is clear that, besides cartilage repair tissue composition, other factors must contribute to clinical success or failure.

As a noninvasive technique and with the ability to follow patients over time, magnetic resonance imaging has provided valuable information in regards to cartilage repair and avoids some of the limitations of biopsies, but studies have similarly shown a poor correlation between repair tissue structural parameters and clinical outcome.^24,49,50

Another limitation of the current study is the inability to blind the investigators to the groups, an inherent limitation of the sampling method. Additionally, the different sizes in specimens (blocks for TKA vs. cylinders for clinically intact grafts) and the nature of tissue (osteochondral and cartilage shavings for clinically intact grafts) leads to a comparison of heterogeneous sample groups. Similarly, pathology of the subchondral bone was heterogeneous and certain aspects such as the position of the subchondral bone plate in the osteochondral biopsy samples relative to the surrounding subchondral bone could not be determined.

The decision to offer TKA or revision arthroscopy was undertaken by the treating surgeons. Although specific clinical scores and/or patient-reported outcome measures (PROMs) collected at the time of revision surgery would have been useful for the analysis and interpretation of our data, this information is not routinely collected from patients in a clinical setting and is rarely used as part of the decision-making process for arthroplasty surgery. There are no agreed criteria for patients to have a knee replacement and the decision to proceed with TKA is typically a patient’s preference-based decision. ⁵¹

We used immunohistochemistry to measure the percentage of type I and II collagen within the repair tissue by measuring the total area of positive immunostaining, regardless of staining intensity, within the total grafted area. Although immunohistochemistry provides information about the distribution of the collagen subtypes and an indication of the amount of positively stained tissue, it can be argued that this is not a true quantitative method, ⁵² as results may be affected by potential inconsistencies in methodology (such as thickness of tissue in sections and antigen retrieval protocols). ⁵³ Additionally, the limited association between antigen presence and staining intensity above and below a certain threshold of bound antibody combined with the subjectivity in assessment, make staining intensity an unreliable method. ⁵³

Micro-CT imaging of clinically failed specimens was performed in only 2 out of 10 cases and illustrations may therefore not be representative of all clinically failed MACI grafts. The low number of available micro-CT data was due to delayed recognition of the subchondral bone involved in the pathology and therefore only the most recent samples underwent micro-CT scanning.

In summary, the most prevalent tissue in revised and clinically failed MACI biopsies was of fibrocartilaginous nature and there was no significant difference in composition between the 2 groups. Subchondral bone pathology of differing degrees was present in all clinically failed MACI biopsies and failure of the osteochondral unit, rather than the graft repair tissue is probably more appropriate terminology. Further investigation into the cause of joint pain and tighter definition of failure are needed.