Knee Surgery, Sports Traumatology, Arthroscopy

, Volume 18, Issue 4, pp 519–527

Matrix-induced autologous chondrocyte implantation versus microfracture in the treatment of cartilage defects of the knee: a 2-year randomised study

Authors

    • Clinic and Polyclinic for Orthopaedic SurgeryUniversity Hospital Giessen-Marburg GmbH
  • Bernd Ishaque
    • Clinic and Polyclinic for Orthopaedic SurgeryUniversity Hospital Giessen-Marburg GmbH
  • Georg Bachmann
    • Department of Diagnostic RadiologyKerckhoff Clinic
  • Henning Stürz
    • Clinic and Polyclinic for Orthopaedic SurgeryUniversity Hospital Giessen-Marburg GmbH
  • Jürgen Steinmeyer
    • Clinic and Polyclinic for Orthopaedic SurgeryUniversity Hospital Giessen-Marburg GmbH
Knee

DOI: 10.1007/s00167-009-1028-1

Cite this article as:
Basad, E., Ishaque, B., Bachmann, G. et al. Knee Surg Sports Traumatol Arthrosc (2010) 18: 519. doi:10.1007/s00167-009-1028-1

Abstract

Cartilage defects occur in approximately 12% of the population and can result in significant function impairment and reduction in quality of life. Evidence for the variety of surgical treatments available is inconclusive. This study aimed to compare the clinical outcomes of patients with symptomatic cartilage defects treated with matrix-induced autologous chondrocyte implantation (MACI™ or microfracture (MF). Included patients were ≥18 and ≤50 years of age with symptomatic, post-traumatic, single, isolated chondral defects (4–10 cm2) and were randomised to receive MACI™ or MF. Patients were followed up 8–12, 22–26 and 50–54 weeks post-operatively for efficacy and safety evaluation. Outcome measures were the Tegner, Lysholm and ICRS scores. Sixty patients were included in a randomised study (40 MACI™, 20 MF). The difference between baseline and 24 months post-operatively for both treatment groups was significant for the Lysholm, Tegner, patient ICRS and surgeon ICRS scores (all P < 0.0001). However, MACI™ was significantly more effective over time (24 months versus baseline) than MF according to the Lysholm (P = 0.005), Tegner (P = 0.04), ICRS patient (P = 0.03) and ICRS surgeon (P = 0.02) scores. There were no safety issues related to MACI™ or MF during the study. MACI™ is superior to MF in the treatment of articular defects over 2 years. MACI™ and MF are complementary procedures, depending on the size of the defect and symptom recurrence. The MACI™ technique represents a significant advance over both first and second generation chondrocyte-based cartilage repair techniques for surgeons, patients, health care institutions and payers in terms of reproducibility, safety, intraoperative time, surgical simplicity and reduced invasiveness.

Keywords

KneeCartilageAutologous transplantationArthroscopic microfractureMatrix induced chondrocyte implantation (MACI)

Introduction

The exact incidence of cartilage defects in the knee is not known, but a prevalence of between 11 and 63% has been reported in patients undergoing arthroscopic knee surgery [2, 10, 16], and approximately 12% in the overall population [29]. The clinical impact of these defects, and therefore the need for clinical intervention, varies as the presence of defect(s) is not necessarily linked to symptoms. However, patients commonly present with knee pain, swelling, locking and catching of the joint, resulting in a significant loss of function and reduction in quality of life. Patients with articular cartilage defects are also predisposed to developing osteoarthritis with its associated disabilities and socioeconomic impact [11].

Articular cartilage defects may be secondary to sports injury or other trauma, or primary lesions resulting from diseases such as osteochondritis dissecans [29]. The avascular nature of articular cartilage means that it has a limited ability to self-repair and regenerate if damaged. Many treatment modalities have been employed with the goal of restoring function and reducing pain. These operative treatments can be divided into replacement, bone marrow stimulation, and regeneration techniques.

Replacement is most commonly achieved with osteochondral autologous grafts (OATS) [15] but is limited by the quantity of harvesting possible without causing donor site morbidity. Bone marrow stimulation techniques (abrasion, drilling and microfracture) aim to stimulate a healing response by exposing the subchondral bone marrow and creating a blood clot that allows the migration of mesenchymal stem cells. Microfracture (MF) was developed by Steadman et al. [31] and, since the initial publication on the technique, it has become the most frequently utilised procedure in this indication [13].

Regenerative techniques aim to stimulate the production of repair tissue, which shares the structure and biomechanical properties of endogenous hyaline cartilage. Autologous chondrocyte implantation (ACI) was first described in the literature by Brittberg et al. [8], and more than 12,000 patients have since undergone treatment [14].

First generation ACI required the harvesting of an autologous periosteal patch, which was used to secure the chondrocytes in situ. This requirement led to complications such as intra-articular adhesions, periosteal hypertrophy and delamination of the defect [26]. Second generation ACI techniques replaced the need for a periosteal patch with a resorbable collagen membrane [12].

Matrix-induced autologous chondrocyte implantation (MACI™, Genzyme Biosurgery, Cambridge, MA) is a third generation ACI product. The chondrocytes are supplied seeded onto a type I/III collagen scaffold, which is simply secured into the lesion with fibrin glue. MACI™ has been shown to deliver comparative clinical results to traditional ACI [5, 26] but offers several advantages including reduced operative time [19], reduced tourniquet time [4, 24] and the ability to perform the implantation via minimally invasive methods such as mini-arthrotomy or arthroscopy [4].

The variety of surgical techniques employed, the range of products available on the market and the diversity of methodological approaches in study design mean that the evidence for all surgical interventions is, at best, inconclusive [20]. Two recent studies compared ACI and MF: Knutsen et al. [21, 22] concluded that MF should be preferred over ACI, whilst Saris et al. [28] concluded that ACI should be preferred over MF. Such conflicting evidence makes it difficult for clinicians to formulate evidence-based decisions about which technique to use in their practices.

Microfracture is a widely available, simple, minimally invasive, arthroscopic technique associated with reasonable costs. More invasive and expensive cell-based techniques must be shown to have clinical outcomes at least comparable to MF.

This randomised, controlled study was based on the hypothesis that MACI™ would deliver better clinical outcomes than MF in the treatment of cartilage defects.

Materials and methods

Study design

The study was conducted at the principal author’s clinic in Germany between 2000 and 2005 and was open to patients of either gender aged ≥18 and ≤50 with post-traumatic, single, isolated, symptomatic chondral defects (4–10 cm2) of the femoral condyle or patella. Exclusion criteria included the presence of chronic inflammatory arthritis, instability of the knee joint, prior or planned meniscectomy (>30% of the meniscus), BMI > 30, varus or valgus abnormality, osteonecrosis, osteoarthritis and chondrocalcinosis.

At baseline all patients underwent symptomatic evaluation were assessed for their eligibility for inclusion and gave their written informed consent. Patients were allocated consecutive numbers in the order of their study entry and then randomised to receive either MACI™ or MF via a computer-generated randomisation list.

The original study protocol allowed for comparison between three parallel treatment groups, two MACI™ groups and one MF group. Patients in both of the MACI™ groups were to receive the same treatment, but the chondrocytes would be seeded onto two different collagen matrices [one supplied from one manufacturing site (MACI™ A), and one from another (MACI™ B)]. However, both scaffolds were validated by the manufacturers as identical, so the two MACI™ groups were combined for the purposes of statistical analysis.

Treatment of concomitant cartilage or meniscus lesions during treatment was permitted. Preoperative MRI scanning alone is not adequate in determining the extent and severity of cartilage lesions, thus an initial arthroscopy was conducted to assess the fulfilment of arthroscopic inclusion criteria (isolated defect >4 cm2).

Patients in both treatment groups were followed up 8–12, 22–26 and 50–54 weeks after surgery for the evaluation of efficacy and safety. Outcome measures were the Tegner (activity levels) [32], Lysholm (pain, stability, gait, clinical symptoms) [23] and ICRS [18] scores.

MRI scans were taken 1 week post-operatively to check for delamination and graft hypertrophy.

The efficacy population was defined as those patients who provided data from at least one follow-up visit ≥6 months post-operatively. Completers were defined as those patients providing 2 years follow-up data.

Safety during the study was assessed by the continuous monitoring of adverse event (AE) reports, which included description, severity, cause, action(s) taken (if any), outcome and relation (if any) to treatment in all patients enrolled.

The original study protocol required an arthroscopic biopsy of each treated defect (requiring additional informed patient consent), 1 year after treatment. Patients were reluctant to consent to this procedure when they were asymptomatic. In two patients who presented at 1 year with symptoms (1× suspected meniscal tear, 1× sub-chondral oedema), arthroscopic biopsy was performed. MRI evaluations following these cylindrical, full thickness biopsies showed a persistent cylindrical defect at the biopsy site with inflow of synovial fluid into the subchondral bone. The decision was therefore taken not to continue with this protocol requirement.

The study protocol was approved by the local ethics committee and was conducted according to the Declaration of Helsinki and Good Clinical Practice.

Surgical technique

For consistency of quality, all surgical procedures were performed at the same clinic by the same surgeon.

All patients underwent a first arthroscopic surgery to assess suitability for study entry and study treatment, and fulfilment of arthroscopic criteria (isolated defect, >4 cm2). A cartilage biopsy was harvested from patients in the MACI™ group and sent for culturing, and patients in the MF group received the study treatment. Four to six weeks later, the patients in the MACI™ group underwent a second intervention for implantation of the chondrocyte seeded collagen scaffold. All implantations were performed via mini-arthrotomy.

During surgery, the defect was prepared by removing all fibrous tissue, sclerotic bone and cartilage remnants until the calcified cartilage layer had been exposed. The minimum possible amount of healthy cartilage was removed, and care was taken not to penetrate the subchondral bone. Once sharp, vertical walls of healthy cartilage of normal thickness surrounded the defect the subchondral bone plate was checked to ensure that it was completely intact and free of fibrous tissue and bleeding.

A template of the defect was made using sterile aluminium foil and used to cut the MACI™ membrane to size. A thin layer of fibrin sealant was applied to the defect and the MACI™ implant placed into the defect, cell side down, facing the subchondral bone. Gentle pressure was applied until the membrane was secured. The treated knee was moved through a full range of motion to ensure the security of the implant prior to closure of the mini-arthrotomy.

Rehabilitation

Patients were required to follow a post-operative rehabilitation programme appropriate to either MACI™ or MF.

Patients in the MF group underwent rehabilitation in line with the recommendations made by Steadman et al. [30], which include 6 weeks of partial weight-bearing (10 kg) on crutches, continuous passive motion (CPM) and physiotherapy. From 6 weeks post-operatively, patients progressed gradually to full weight-bearing.

Rehabilitation for patients in the MACI™ group included a dorsal plaster cast (10° flexion) for 2 days post-operatively to prevent delamination of the graft, CPM and physiotherapy, followed by 8 weeks of partial weight-bearing (10 kg) on crutches.

Patients in both groups received anti-thrombotic prophylaxis with low-molecular heparin certoparin-natrium (Monoembolex s.c. 1 per day) for the entire period of partial weight-bearing.

Statistical analysis

For inferential analyses, the significance level was set to α = 5%. All significance tests used two-sided alternatives. For continuous baseline variables, means and standard deviations in each group are given, accompanied by their medians, 25th and 75th percentiles as useful measures of location and variability for asymmetrical distributions. Frequency tables summarised categorical baseline variables. For each of the five longitudinally observed outcome measures of central interest, two questions were considered:
  1. (1)

    Is there a time trend in the score values post-operatively (i.e. a main effect of time)?

     
  2. (2)

    Is there a difference between the two treatment groups in the time trends of the score values (i.e. an interaction effect between treatment and time)?

     

To answer these questions a fully non-parametric rank method was used to analyse the time courses of each of the scores, as described by Brunner and Langer (2000) [9]. This allowed a unified longitudinal inferential analysis of the ordinal outcome measures (patient’s and surgeon’s ICRS scores and Tegner score) and for the (quasi-) metric Lysholm score. All available data (not just those from patients with complete records) were used as this analysis is able to deal with incomplete records as long as missing values can be considered “missing completely at random”, which was the case.

All statistical analyses were performed using the statistics software R, version 2.8.0 (2008-10-20) [27] including the Matrix [7] and MASS [33] packages.

Results

Between 2000 and 2005, 60 patients were included in the study. Demographic data and patient disposition are shown in Table 1. The MACI™ group was twice the size of the MF group due to the combining of the two MACI™ groups planned in the original protocol. None of the observed differences between the MF group and the MACI™ group was significant, except for the difference in symptom duration, which was 0.3 years longer in the MF group.
Table 1

Patient demographics and disposition

 

MACI™

MF

Total

N

40

20

60

Gender

 Male N (%)

25 (63)

17 (85)

42 (71)

 Female N (%)

15 (38)

3 (15)

17 (29)

Mean age (years)

33.00

37.50

34.20

Mean BMI kg/m2 (range)

25.3 (20–34)

27.3 (24–35)

26.3 (20–35)

Defect cause (%)*

 Accident

8 (20)

3 (15)

11 (22)

 Sport

19 (48)

8 (40)

27 (45)

 Work

1 (3)

1 (5)

2 (3)

 Daily activities

4 (10)

3 (15)

7 (12)

 Unknown

8 (20)

5 (25)

13 (18)

Symptom duration (years)

2.2

2.5

2.3

Symptom onset*

 Acute (<6 weeks post trauma)

14 (35)

6 (30)

21 (35)

 Gradual (>6 weeks post trauma)

26 (65)

14 (70)

39 (65)

Defect location (%)*

 Condylar

29 (73)

16 (80)

45 (75)

 Patellar-trochlear

11 (28)

4 (20)

15 (25)

* Efficacy population

Patients diagnosed with osteochondral defects were withdrawn from the study as they were unable to receive solely MACI™ or MF. These patients were treated with MACI™ combined with bone grafting [6]. The efficacy population was 56 patients (39 MACI™, 17 MF). By August 2006, 48 patients (33 (84.6%) MACI™, 15 (88.2%) MF) had completed 2 years follow-up.

The MF group had three dropouts. One patient became pregnant 6 months after treatment, one was an early treatment failure and received OATS after 10 months, and one discontinued without giving a reason. One patient discontinued from the MACI™ group without stating a reason. There were several missing values in both groups due to patients failing to attend for follow-up.

Concomitant lesions treated during the study were ACL lesions (one patient, MF group) and smaller meniscal lesions (two patients in the MACI™ group, three patients in the MF group).

The mean Lysholm score in the MACI™ group improved from 52 at baseline to 95 at 12 months. This improvement was broadly maintained at 24 months (mean score 92). In the MF group, these scores improved from 55 at baseline to 81 at 12 months but then declined to 69 at 24 months (Table 2).
Table 2

Mean and median lysholm scores over time

 

Lysholm score

Baseline

3 months

6 months

12 months

18 months

24 months

Mean MACI™

52 ± 26

77 ± 17

87 ± 17

92 ± 11

91 ± 13

92 ± 9

Median MACI™

58

81

93

95

93

94

N MACI™

39

39

39

38

32

33

Mean MF

55 ± 25

66 ± 20

82 ± 18

82 ± 22

80 ± 22

69 ± 26

Median MF

56

70

88

90

90

70

N MF

17

17

17

17

17

15

Individual patients’ Lysholm scores showed maintained improvement over 2 years in the patients treated with MACI™, but patients treated with MF showed a much broader scattering of results. The difference between baseline and 24 months post-operatively for both treatment groups was significant (P < 0.0001), but MACI™ was significantly more effective over time than MF (P = 0.005).

The median Tegner score improved from level 2 at baseline to level 4 at 12 months in the MACI™ group, and this improvement was maintained at 24 months. The median Tegner scores improved from level 2 at baseline to level 3 at 12 months in the MF group, and this improvement was maintained at 24 months (Table 3). The difference between baseline and 24 months post-operatively for both treatment groups was significant (P < 0.0001), but MACI™ was significantly more effective over time than MF (P = 0.04).
Table 3

Disposition of Tegner score levels and medians over time

 

MACI™

MF

Baseline

6 months

12 months

24 months

Baseline

6 months

12 months

24 months

N

39

39

37

37

20

18

18

17

Level 10

0

0

0

0

0

0

0

0

Level 9

0

0

0

0

0

0

0

0

Level 8

1

0

0

0

0

0

0

0

Level 7

0

1

1

0

0

0

0

0

Level 6

1

0

3

2

0

0

0

0

Level 5

1

1

5

7

0

1

1

1

Level 4

1

10

11

16

1

3

6

6

Level 3

11

24

15

11

6

13

9

8

Level 2

12

2

2

1

5

1

1

1

Level 1

12

1

0

0

8

0

1

1

Level 0

0

0

0

0

0

0

0

0

Median

Level 2

Level 3

Level 4

Level 4

Level 2

Level 3

Level 3

Level 3

The number of completers by ICRS score, by therapy, over time is shown in Table 4. The difference between ICRS patient scores at baseline and 24 months post-operatively was significant for both treatment groups (P < 0.0001), but MACI™ was significantly more effective over time than MF (P = 0.03).
Table 4

Disposition of ICRS scores, by therapy over time

ICRS score

Therapy

Baseline

3 months

6 months

12 months

18 months

24 months

I

MACI ™

2

5

7

18

11

14

 

MF

 

1

 

4

2

2

II

MACI ™

4

17

22

14

15

14

 

MF

3

5

12

9

8

4

III

MACI ™

21

13

6

2

3

2

 

MF

10

9

3

3

2

3

IV

MACI ™

8

 

 

 

 

 

 

MF

2

 

 

 

1

1

For the surgeon’s ICRS scores, the difference between baseline and 24 months post-operatively was significant for both treatment groups (P < 0.0001). MACI™ was significantly more effective over time than MF (P = 0.02).

There were no treatment-related safety issues during the study. Any irritation experienced on increased weight-bearing was eased by treatment with non-steroidal anti-inflammatory drugs (NSAID’s) and returning to partial weight-bearing for an additional week. One patient in the MACI™ group had persistent pain after 12 months. A second-look arthroscopy showed an even and firm regenerated tissue surface with good bonding to the surrounding tissue. Persistent subchondral oedema led us to perform retrograde bone grafting that finally achieved pain relief.

Discussion

The primary finding of this study is that MACI™ is superior to MF in the treatment of larger (>4 cm2), symptomatic articular defects over 2 years. Microfracture, despite its minimally invasive nature (Fig. 1), did not produce better clinical results than MACI™, probably due to the limited durability of the regenerative tissue. Second-look arthroscopies after MF commonly show a rough, fibrocartilaginous surface, which has a tendency to deteriorate, is soft in texture and is poorly bonded to the surrounding tissue (Fig. 2). Bone ingrowth from the subchondral plate after MF may also cause a negative biomechanical influence on the regenerative tissue, and studies suggest that such ingrowth may be prejudicial to future therapies [25]. However, compared to the higher invasiveness and costs of cell-based techniques, MF presents an indispensable technique for smaller (<3 cm2) cartilage defects.
https://static-content.springer.com/image/art%3A10.1007%2Fs00167-009-1028-1/MediaObjects/167_2009_1028_Fig1_HTML.jpg
Fig. 1

Intraoperative view of MF

https://static-content.springer.com/image/art%3A10.1007%2Fs00167-009-1028-1/MediaObjects/167_2009_1028_Fig2_HTML.jpg
Fig. 2

Arthroscopic visualisation 6 months after MF showing rough regenerative tissue and incomplete bonding to the surrounding tissues

ACI has been shown to produce good clinical outcomes for ≥9 years, and the resultant repair tissue resembles hyaline cartilage rather than fibrocartilage [26]. A multicenter study [22] comparing ACI and MF showed that ACI was not superior to MF in defects up to 4 cm2. However, interpretation of these results is difficult as the invasiveness and the surgical skills required for ACI and MF are very different. Comparing different cartilage repair publications shows that relatively small, heterogeneous study populations, study centre experience and the effects of different rehabilitation protocols may confound results.

The collagen matrix as a cell carrier made the implantation of chondrocytes easier and less time consuming. Thus, MACI™, a third generation technique, is a simplified surgical procedure that can be performed via mini-arthrotomy or arthroscopically [1]. The seeded chondrocytes (1 million per cm2) cannot leak through the matrix, which acts as a cell carrier. Unlike a periosteal flap, the MACI™ artificial porcine collagen I/III matrix provides consistent quality, flexibility and easy handling.

A prospective, randomised study demonstrated comparable clinical, arthroscopic and histological results for first and third generation chondrocyte implantation techniques [5]. However, procedures where the scaffold only has to be glued into the defect remain more attractive to the surgeon. Few studies compare the effectiveness of chondrocyte implantation to MF due to methodological challenges, with blinding being practically impossible. Osteochondral autograft transfer, another established cartilage repair technique, has been compared to ACI [17] and showed comparable results over a relatively short follow-up period. However, OATS can cause donor site morbidity, especially if used in cartilage defects larger than 4 cm2.

The variety of chondrocyte seeded scaffolds available on the market, and the differences in their manufacturing processes and materials used, confound attempts to objectively compare clinical studies using different products. Additionally, cartilage repair studies are of heterogeneous design and differ widely in their inclusion criteria, patient demographics and history, consistency of lesion size and location for comparison, clinical scores, follow-up periods and multiple centres with different surgeons and skills.

The fact that most studies, especially those used for regulatory approvals, include only one ACI product, may suggest a conflict of interest. Current studies provide insufficient information about the systematic reporting of adverse effects. The most significant local side effect is complete graft rejection with delamination and consecutive therapy failure, occurring usually within the first week after surgery. Other side effects are swelling, haemorrhage and graft hypertrophy, which may occur with the use of periosteal flaps [26].

No serious adverse events were seen in the current study. Some patients experienced slight swelling and inflammation of the knee during the phase (12th week) after partial weight-bearing. This suggests that the regenerative tissue remains vulnerable during this early rehabilitation time period. Rest, local cryotherapy and NSAID’s generally reduced these symptoms within a few days.

The original study protocol included a requirement for arthroscopic biopsy of treated defects 1 year post-operatively, which caused ethical and practical challenges. Patients were unwilling to undergo a relatively invasive procedure when asymptomatic, and obtaining meaningful samples requires multiple biopsy sites (graft, surrounding cartilage, border zone etc.). Reliability is low in these circumstances, and the biopsy procedure itself may cause tissue damage. Arthroscopic evaluation is highly specific but invasive and only allows the evaluation of the cartilage surface, with no assessment of the subchondral bone possible. Studies therefore lack systematic histological assessments, and there is no information in the literature about patient reluctance to allow biopsies in an asymptomatic knee.

MRI is an alternative, highly effective, non-invasive method to evaluate regenerated tissue surfaces, overall morphology, thickness, volume and signal equalization to the surrounding cartilage [3] (Fig. 3). Unlike biopsy (after which an osteochondral defect remains for months), MRI does not cause harm to endogenous cartilage or grafted tissues and can be repeated many times for the evaluation of the entire treated defect and surrounding cartilage.
https://static-content.springer.com/image/art%3A10.1007%2Fs00167-009-1028-1/MediaObjects/167_2009_1028_Fig3_HTML.jpg
Fig. 3

Equalisation of MRI signal intensity in the regenerative tissue 1 and 6 months after treatment with MACI™

Studies, such as this one, which aim to evaluate cartilage repair techniques can be confounded by the subjective nature of pain reported by patients, and individual responses to pharmaceutical treatments, physical therapies and surgical interventions. Whilst these confounding factors cannot be eliminated, they have been partially mitigated in this study by arthroscopic and MRI visualisation of the treated defects. These visualisations support the fact that the MACI™ implant integrates well with the surrounding endogenous, healthy cartilage and results in the formation of hyaline-like cartilage [34]. Figure 4 shows a cartilage defect following debridement back to the macroscopically healthy tissue. The same patient was re-examined arthroscopically 12 months after MACI™ implantation. A smooth, regenerated surface with tight bonding to the surrounding cartilage was observed (Fig. 5).
https://static-content.springer.com/image/art%3A10.1007%2Fs00167-009-1028-1/MediaObjects/167_2009_1028_Fig4_HTML.jpg
Fig. 4

Debridement of the defect and implantation of the MACI™ membrane via mini-arthrotomy

https://static-content.springer.com/image/art%3A10.1007%2Fs00167-009-1028-1/MediaObjects/167_2009_1028_Fig5_HTML.jpg
Fig. 5

The same defect 1 year after MACI™ treatment

Conclusions

MACI™ is superior to MF in the treatment of larger (>4 cm2), symptomatic articular defects over 2 years. MACI™ and MF are complementary procedures for the treatment of articular cartilage defects, depending on the size of the defect and symptom recurrence. As a third generation technique, MACI™ is not only superior to MF but also improves upon the first and second generation chondrocyte-based cartilage repair techniques in terms of reproducibility, safety, operative time, surgical simplicity and reduced invasiveness.

Acknowledgments

The authors would like to thank Dr. Gerrit Eichner for his assistance with the statistical analysis.

Copyright information

© Springer-Verlag 2010