Systematic review of the concurrent and predictive validity of MRI biomarkers in OA

SUMMARY

Objective

To summarize literature on the concurrent and predictive validity of MRI-based measures of osteoarthritis (OA) structural change.

Methods

An online literature search was conducted of the OVID, EMBASE, CINAHL, PsychInfo and Cochrane databases of articles published up to the time of the search, April 2009. 1338 abstracts obtained with this search were preliminarily screened for relevance by two reviewers. Of these, 243 were selected for data extraction for this analysis on validity as well as separate reviews on discriminate validity and diagnostic performance. Of these 142 manuscripts included data pertinent to concurrent validity and 61 manuscripts for the predictive validity review. For this analysis we extracted data on criterion (concurrent and predictive) validity from both longitudinal and cross-sectional studies for all synovial joint tissues as it relates to MRI measurement in OA.

Results

Concurrent validity of MRI in OA has been examined compared to symptoms, radiography, histology/pathology, arthroscopy, CT, and alignment. The relation of bone marrow lesions, synovitis and effusion to pain was moderate to strong. There was a weak or no relation of cartilage morphology or meniscal tears to pain. The relation of cartilage morphology to radiographic OA and radiographic joint space was inconsistent. There was a higher frequency of meniscal tears, synovitis and other features in persons with radiographic OA. The relation of cartilage to other constructs including histology and arthroscopy was stronger. Predictive validity of MRI in OA has been examined for ability to predict total knee replacement (TKR), change in symptoms, radiographic progression as well as MRI progression. Quantitative cartilage volume change and presence of cartilage defects or bone marrow lesions are potential predictors of TKR.

Conclusion

MRI has inherent strengths and unique advantages in its ability to visualize multiple individual tissue pathologies relating to pain and also predict clinical outcome. The complex disease of OA which involves an array of tissue abnormalities is best imaged using this imaging tool.

Introduction

Magnetic Resonance Imaging (MRI) is being developed as a method to assess joint morphology in osteoarthritis (OA), with the goal of providing a sensitive non-invasive tool for the study of healthy and diseased states, and a means of assessing the effectiveness of interventions for osteoarthritis. Traditionally structural assessment of OA has relied upon the plain radiograph which has capacity to image the joint space and osteophytes¹. MRI has many advantages in visualizing the joint, and recent efforts are yielding a variety of approaches that offer the potential for monitoring this prevalent synovial joint disease². Because OA is a disease of the whole synovial joint, not just the cartilage, measurements of structure need to be seen broadly and capture important anatomic features, including osteophytes, effusions, meniscal tears, subchondral bone architectural changes or ligamentous instability, in addition to cartilage loss². There is an abundant literature describing the concurrent validity of MRI as it relates to comparable constructs such as histology and radiography but little if any effort has been made to systematically summarize this literature.

Similarly the merits of any OA structural assessment will undoubtedly be assessed for their clinical relevance. There are multiple determinants of pain and functional limitation in OA and there may be many more unknown³. Many studies have examined whether the loss of structural integrity is in some way the physical correlate of these symptoms. Traditionally most epidemiologic studies have relied upon plain radiography to define disease. The major limitation of this method is that measures of symptoms correlate poorly with x-ray features. Less than 50% of people with evidence of OA on plain radiographs have symptoms related to these findings⁴. Uncertainty as to whether measurements of MRI structure alone will adequately reflect what structure connotes, or whether other metrics of structure should also be considered, need to be systematically evaluated. The relationships between structure and pain and/or function and between structure and future outcomes (e.g., arthroplasty) are critical in determining the clinical relevance of MRI.

In psychometrics, validity refers to the degree to which a study accurately reflects or assesses the specific concept that the researcher is attempting to measure. There are many types of validity of which one, criterion validity, is used to demonstrate the accuracy of a measure or procedure by comparing it with another measure or procedure which has been demonstrated to be valid. There is a contention in the OA field about the validity of a number of biomarkers and clinical endpoints and their inclusion here is in an effort to be comprehensive and does not diminish the credible concerns about the lack of well validated clinical endpoints⁵. If the test data and criterion data are collected at the same time, this is referred to as concurrent validity evidence. If the test data is collected first in order to predict criterion data collected at a later point in time, then this is referred to as predictive validity evidence. The purpose of this systematic review was to summarize the OA MRI literature with regards to both concurrent and predictive validity.

Material and methods

Systematic literature search details

An online literature search was conducted using the OVID MEDLINE (1945–), EMBASE (1980–) and Cochrane databases (1998–) to identify the articles published up to April 2009, with the search entries “MRI”, and “osteoarthritis”, “osteoarthritides”, “osteoarthrosis”, “osteoarthroses”, “degenerative arthritis”, “degenerative arthritides”, or “osteoarthritis deformans”. The abstracts of the 1330 citations received with this search were then preliminarily screened for relevance by two reviewers (KH and DJH). For this preliminary search, all articles which used MRI, in some form, on patients with osteoarthritis of the knee, hip, or hand were included. Although review articles were not included (see Inclusion/exclusion criteria), citations found in any review articles which were not already included in our preliminary search were screened for possible inclusion in this study. This added 7 more relevant studies to our search. One further article was added, before publication, by one of authors of this meta-analysis bringing the preliminary total to 1338.

Inclusion/exclusion criteria

Only studies published in English were included. Studies presenting non-original data were excluded, such as reviews, editorials, opinion papers, or letters to the editor. Studies with questionable clinical relevance and those using non-human subjects or specimens were excluded. Studies in which rheumatoid, inflammatory, or other forms of arthritis were included in the OA datasets were excluded, as well as general joint-pertinent MRI studies not focused on OA. Studies with no extractable, numerical data were excluded. Only those articles which had some measure of diagnostic performance were included. Any duplicates which came up in the preliminary search were excluded. Of the preliminary 1338 abstracts, 243 were selected for data extraction (Fig. 1).

Fig. 1.

Flow chart of the screening process for articles included in the systematic review.

Data abstraction

Two reviewers (KH and LM) independently abstracted the following data: (1) patient demographics; (2) MRI make, sequences and techniques used, tissue types viewed; (3) study type and funding source; (4) details on rigor of study design to construct the Downs methodological quality score⁶; (5) MRI reliability/reproducibility data; (6) MRI diagnostic measures and performance; (7) gold standard measures against which the MRI measure was evaluated; (8) treatment and MRI measures (when appropriate).

The Downs methodological quality score⁶ collects a profile of scores for both randomized trials and observational studies in terms of quality of reporting, internal validity (bias and confounding), power, external validity so that the overall study quality score reflects all of these elements. Answers were scored 0 (No) or 1 (Yes), except for one item in the Reporting subscale, which scored 0–2 and the single item on power, which was scored 0–5. The possible range is from 0 to 27 where 0 represents poor quality and 27 optimal quality.

We used a data abstraction tool constructed in EpiData (Entry version 2.0 Odense, Denmark) and more than one reviewer undertook the data abstraction. The data collection forms were designed to target the objectives of the review, and were piloted prior to conducting the study.

The outcomes for psychometric properties on MRI were examined using the OMERACT filter^7,8. The specific focus of this review is upon the truth domain: is the measure truthful, does it measure what it intends to measure? More specifically we were interested in criterion validity; for both the concurrent [Does it agree (by independent and blind comparison) with a measure that reflects the same concept] and predictive [Does it predict (by independent and blind comparison) a future ‘gold standard’] validity of MRI in OA. If the test data and criterion data are collected at the same time, this is referred to as concurrent validity evidence. If the test data is collected first in order to predict criterion data collected at a later point in time, then this is referred to as predictive validity evidence.

It is critical to delineate what we mean by the various terms used, as current usage is often incorrect, and this ambiguity may stem from an incorrect understanding of appropriate definitions. Whilst there are several definitions that have been proposed^9–13, the brief synthesis of some working definitions is as follows:

1. biological marker (biomarker)— a characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes or pharmacologic responses to a therapeutic agent.

2. clinical endpoint—a clinically meaningful measure of how a patient feels, functions, or survives.

For the purposes of this analysis, MRI (the biomarker) is directly compared both to clinical endpoints (symptoms, total knee replacement (TKR)) as well as other biomarkers (including radiography, CT, histology, arthroscopy, alignment). The presentation of the data in the results reflects presentation of clinical endpoints before comparison with other biomarkers.

There is some overlap in the manuscripts for which data is extracted for these two types of validity. The large majority of studies for concurrent validity were cross-sectional studies although some longitudinal studies reported cross-sectional results and thus are included in the concurrent validity data. There is no attempt made to create summary estimates as the validity effect measures [i.e., odds ratio (OR), Beta coefficient, r, P-value of difference] used in this literature are very heterogeneous.

Results

Concurrent validity (Table I)

Table I.

Summary table of studies reporting data on concurrent validity of MRI in OA

Reference: Author, Journal, Year, PMID	Whole sample size	No. of cases	No. of controls	Age, yrs, mean(SD), range	No. (%) of females	Quantitative cartilage	Compositional techniques	Semi-quantitative	Cartilage	Synovium	Bone	Bone marrow lesions	Meniscus	Ligament	Study design	Score of methodological quality
Chan WP; American Journal of Roentgenology; 1991; 189204036	20	20	0	58(Range: 42–73)	11	No	No	Yes	Yes	No	No	No	Yes	Yes	Cross-sectional	6
McAlindon TE; Annals of the Rheumatic Diseases; 1991; 199486190	12					No	No	Yes	Yes	Yes	Yes	No	Yes	No	Case control	3
Li KC; Magnetic Resonance Imaging; 1988; 339872891	10	10	0	(Range: 33–78)	9(90%)	No	No	Yes	Yes	No	No	No	No	No	Longitudinal Prospective	4
Fernandez-Madrid F; Magnetic Resonance Imaging; 1994; 793465692	92	52	40	Controls: 49(15), (Rang: 22–78); OA patients: 55(14), (Range: 25–86)	60	No	No	Yes	Yes	Yes	No	No	Yes	No	Cross-sectional	11
Karvonen RL; Journal of Rheumatology; 1994; 796607527	92	52	40	Reference: 49(15), (Range: 22–78); All OA patients: 55(14), (Range: 25 –86); Bilateral OA: 53(13), (Range: 25–73)	60	Yes	No	No	Yes	No	Yes	No	No	No	Case control	11
Peterfy CG; Radiology; 1994; 802942093	8	5	3	62(Range: 45–82)	4(50%)	Yes	No	No	Yes	No	No	No	No	No	Cross-sectional	4
Blackburn WD Jr; Journal of Rheumatology; 1994; 803539237	33	33	0	62.7(9.1), (Range: 44–79)	17	No	No	Yes	Yes	No	No	No	No	No	Cross-sectional	6
Broderick LS; American Journal of Roentgenology; 1994; 827370061	23	13	10			No	No	Yes	Yes	No	No	No	No	No	Cross-sectional	4
Miller TT; Radiology; 1996; 881655294	384			47(Range: 14–88)		No	No	Yes	Yes	No	No	No	Yes	Yes	Cross-sectional	8
Dupuy DE; Academic Radiology; 1996; 895918157	7			TKA patients: (Range: 64–75); Asymptomatic: 35(Range: 25–35)	3	Yes	No	No	Yes	No	No	No	No	No	Other	6
Kenny C; Clinical Orthopaedics & Related Research; 1997; 918621595	136					No	No	Yes	No	No	No	No	Yes	No	Case control	6
Breitenseher MJ; Acta Radiologica; 1997; 933224896	60	12	48	37(14.3), (Range: 15–68)	30(50%)	No	No	Yes	No	No	No	No	Yes	No	Cross-sectional	5
Ostergaard M; British Journal of Rheumatology; 1997; 940286097	46	14	47	70(Range: 24–85)		No	No	No	No	Yes	No	No	No	No	Cross-sectional	7
Trattnig S; Journal of Computer Assisted Tomography; 1998; 944875498	20	20	0	72.2(Range: 62–82)	18	No	No	Yes	Yes	No	No	No	No	No	Other	8
Kawahara Y; Acta Radiologica; 1998; 952944062	72			58(Range: 41–74)	46	No	No	Yes	Yes	No	No	No	No	No	Other	6
Drape JL; Radiology; 1998; 964679263	43	43	0	63(Range: 53–78)	30	No	No	Yes	Yes	No	No	No	No	No	Cross-sectional	5
Eckstein F; Clinical Orthopaedics & Related Research; 1998; 967804256	8	0	8	50.6(Range: 39–64)		Yes	No	No	Yes	No	No	No	No	No	Other	7
Uhl M; European Radiology; 1998; 972442358	22			(Range: 50–72)		No	No	Yes	Yes	No	No	No	No	No	Cross-sectional	5
Boegard T; Acta Radiologica - Supplementum; 1998; 975912199	61					No	No	Yes	Yes	No	No	No	Yes	No	Longitudinal Prospective	5
Bachmann GF; European Radiology; 1999; 993339964	320			29.3(8.7), (Range: 13–56)	122	No	No	Yes	Yes	No	No	No	Yes	No	Cross-sectional	7
Cicuttini F; Osteoarthritis & Cartilage; 1999; 10329301100	28			Males: 40.4(Range: 42–58); Females: 31.2(8.6);	11	Yes	No	No	Yes	No	No	No	No	No	Cross-sectional	7
Boegard T; Annals of the Rheumatic Diseases; 1999; 10343536101	58			Women: 40.4(Range: 42–58); Men: 57(49.5), (Range: 41–57)	29	No	No	Yes	Yes	No	No	No	No	No	Cross-sectional	6
Adams JG; Clinical Radiology; 1999; 1048421644	62	32	30			No	No	Yes	Yes	No	No	No	Yes	No	Case control	8
Pham XV; Revue du Rhumatisme; 1999; 10526380102	10	10	10	67.2(7.34), (Range: 57–80)	6	No	No	Yes	No	No	No	No	No	Yes	Cross-sectional	13
Gale DR; Osteoarthritis & Cartilage; 1999; 1055885043	291	233	58			No	No	No	No	No	No	No	Yes	No	Case control	10
Kladny B; International Orthopaedics; 1999; 1065329059	26					Yes	No	No	Yes	No	No	No	No	No	Cross-sectional	6
Zanetti M; Radiology; 2000; 10831707103	16	16	0	67(Range: 43–79)	15	Yes	No	No	Yes	No	Yes	Yes	No	No	Cross-sectional	6
Jones G; Arthritis & Rheumatism; 2000; 11083279104	92	92	0	Boys: 12.8(2.7); Girls: 12.6(2.9)	43	Yes	No	No	Yes	No	Yes	No	No	No	Cross-sectional	13
McCauley TR; American Journal of Roentgenology; 2001; 11159074105	193			40(Range: 11–86)	83	No	No	Yes	Yes	No	No	No	Yes	Yes	Cross-sectional	8
Wluka AE; Annals of the Rheumatic Diseases; 2001; 11247861106	81	42	39	Cases: 58(6.1); Controls: 56(5.4)	81(100%)	Yes	No	Yes	Yes	No	No	No	No	No	Case control	16
Felson DT; Annals of Internal Medicine; 2001; 1128173614	401	401	0	66.8		No	No	Yes	No	No	No	Yes	No	No	Cross-sectional	13
Hill CL; Journal of Rheumatology; 2001; 1140912715	458	433	25	67	(34%)	No	No	Yes	No	Yes	No	No	No	No	Case control	13
Kawahara Y; Journal of Computer Assisted Tomography; 2001; 11584226107	35			57(Range: 33–70)	23	No	No	Yes	Yes	No	No	No	Yes	No	Cross-sectional	8
Arokoski JP; Annals of the Rheumatic Diseases; 2002; 11796401108	57	27	30	Cases: 56.2(4.9), (Range: 47–64); Controls: 56.3(4.5), (Range: 47–64)	0	Yes	No	No	No	No	No	No	No	No	Case control	8
Bergin D; Skeletal Radiology; 2002; 11807587109	60	30	30	Cases: 50; Controls: 57		No	No	Yes	No	No	No	No	Yes	Yes	Case control	9
Beuf O; Arthritis & Rheumatism; 2002; 11840441110	46	18	28	Mild OA: 68(9.1); Severe OA: 70(6.3)	17	Yes	No	No	No	No	No	No	No	No	Case control	5
Arokoski MH; Journal of Rheumatology; 2002; 12375331111	57	27	30	Cases: 56.2(4.9), (Range: 47–64); Controls: 56.3(4.5), (Range: 47–64)	0	Yes	No	No	Yes	No	No	No	No	No	Case control	8
Bhattacharyya T; Journal of Bone & Joint Surgery - American Volume; 2003; 1253356516	203	154	49	Cases: 65; Controls: 67		No	No	Yes	No	No	No	No	Yes	No	Case control	9
Link TM; Radiology; 2003; 1256312817	50	50	0	63.7(11.5), (Range: 43–81)	30	No	No	Yes	Yes	No	No	No	Yes	Yes	Cross-sectional	6
Tiderius CJ; Magnetic Resonance in Medicine; 2003; 12594751112	17			50(Range: 35–70)	4	No	Yes	No	Yes	No	No	No	No	No	Cross-sectional	6
Cicuttini FM; Arthritis & Rheumatism; 2003; 1263242128	252			60.2(10)	157962%)	Yes	No	No	Yes	No	Yes	No	No	No	Cross-sectional	9
Cicuttini FM; Clinical & Experimental Rheumatology; 2003; 12673893113	81	42	39	ERT: 58(6.1); Controls: 56(5.4)	81(100%)	Yes	No	No	Yes	No	Yes	No	No	No	Case control	12
Sowers MF; Osteoarthritis & Cartilage; 2003; 1280147818	120	60	60	no OAK, no Pain: 45(0.8); OAK, no Pain: 46(0.6); No OAK, Pain: 47(0.8); OAK and Pain: 47(0.7)	(100%)	No	No	Yes	Yes	No	No	Yes	No	No	Case control	11
McGibbon CA; Osteoarthritis & Cartilage; 2003; 1281461160	4					No	No	Yes	Yes	No	No	No	No	No	Other	5
Cicuttini FM; Clinical & Experimental Rheumatology; 2003; 1284605046	157	157	0	62(10)	(62%)	Yes	No	No	Yes	No	No	No	No	No	Cross-sectional	10
Felson DT; Annals of Internal Medicine; 2003; 1296594151	256	256	0	Followed: 66.2(9.4); Not followed: 67.8(9.6)	(38.3%)	No	No	Yes	No	No	No	Yes	No	No	Longitudinal prospective	11
Tarhan S; Clinical Rheumatology; 2003; 14505208114	74	58	16	OA Patients: 57.4(8.5), (Range: 45–75); Healthy controls: 59.1(5.8), (Range: 46–77)	60	Yes	No	Yes	Yes	Yes	No	No	No	No	Case control	8
Hill CL; Arthritis & Rheumatism; 2003; 1455808919	451	427		Knee pain/ROA/Male: 68.3; Knee pain/ROA/Female: 65; No knee pain/ROA/Male: 66.8; No knee pain/ROA/Female: 66.1		No	No	Yes	No	No	No	Yes	No	No	Cross-sectional	10
Kim YJ; Journal of Bone & Joint Surgery - American Volume; 2003; 14563809115	43			30(Range: 11–47); Median = 31	40	No	Yes	No	Yes	No	No	No	No	No	Other	5
Lindsey CT; Osteoarthritis & Cartilage; 2004; 1472386829	74	33	21	Controls: 34.2(12.5); OA1 (KL1/2): 62.7(10.9); OA2(KL3/4): 66.6(11.6)	39	Yes	No	No	Yes	No	Yes	No	No	No	Case control	8
Jones G; Osteoarthritis & Cartilage; 2004; 1472387647	372	186	186	45(Range: 26–61)		Yes	No	No	Yes	No	Yes	No	No	No	Case control	9
Raynauld JP; Arthritis & Rheumatism; 2004; 1487249048	32	32	0	62.9(8.2)	(74%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	10
Wluka AE; Annals of the Rheumatic Diseases; 2004; 1496296020	132	132	0	63.1(Range: 41–86)	71(54%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	10
Cicuttini F; Rheumatology; 2004; 1496320152	117	117	0	67(10.6)	(58%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	12
Peterfy CG; Osteoarthritis & Cartilage; 2004; 14972335116	19	19	0	61(8)	4	No	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Other	5
Graichen H; Arthritis & Rheumatism; 2004; 15022323117	21	21	0	70.6(7.7), (Range: 58–86)	17	Yes	No	No	Yes	No	Yes	No	No	No	Cross-sectional	6
Dashti M; Scandinavian Journal of Rheumatology; 2004; 15163109118	174	117	57	61.6(9.5)	123(70.7%)	Yes	No	No	Yes	No	No	No	No	No	Case control	11
Arokoski JP; Journal of Clinical Densitometry; 2004; 15181262119	57	27	30	Cases: 56.2(4.9), Range: (47–64); Controls: 56.3(4.5), (Range: 47–64)	0	No	Yes	No	No	No	No	No	No	No	Case control	9
Dunn TC; Radiology; 2004; 15215540120	55	48	7	Healthy: 38(Range: 22–71); Mild OA: 63(Range: 46–81); Severe OA: 67 (Range: 43–88)	30	No	Yes	No	Yes	No	No	No	No	No	Case control	8
Regatte RR; Academic Radiology; 2004; 15217591121	14	6	8	Asymptomatic: 33.5(Range: 22–45); Symptomatic: 45.5(Range: 28–63)	2	No	Yes	No	Yes	No	No	No	No	No	Case control	7
Baysal O; Swiss Medical Weekly; 2004; 15243849122	65	65	0	53.1(7), (Range: 45–75)	65(100%)	Yes	No	Yes	Yes	No	Yes	No	No	No	Cross-sectional	7
Lerer DB; Skeletal Radiology; 2004; 15316679123	205			46.5(Range: 15–88); Median = 46	113	No	No	Yes	Yes	No	No	No	Yes	No	Cross-sectional	6
Berthiaume MJ; Annals of the Rheumatic Diseases; 2005; 1537485578	32					Yes	No	Yes	Yes	No	No	No	Yes	No	Longitudinal Prospective	10
King KB; Magnetic Resonance Imaging; 2004; 15527998124	16	16	0	Males: Median = 58.5, (11.3), (Range: 43–76); Females: Median = 70 (14.4), (Range: 46–88)	8(50%)	Yes	Yes	No	Yes	No	No	No	No	No	Cross-sectional	7
Carbone LD; Arthritis & Rheumatism; 2004; 15529367125	818			Non-users: 74.8(2.94); Antiresportive users: 74.8(2.9)	818(100%)	No	No	Yes	Yes	Yes	Yes	No	No	No	Cross-sectional	11
Cicuttini F; Journal of Rheumatology; 2004; 15570649126	123					Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	6
Wluka AE; Annals of the Rheumatic Diseases; 2005; 1560174238	149	68	81	Normal: 57(5.8); OA: 63(10.3)	1499(100%)	No	No	No	No	No	Yes	No	No	No	Longitudinal Prospective	13
Ding C; Osteoarthritis & Cartilage; 2005; 1572788549	372	162	210	No cartilage defects: 43.6(7.1); Any cartilage defect: 47(6.1)	(56.5%)	Yes	No	Yes	Yes	No	No	No	No	No	Case control	9
Hill CL; Arthritis & Rheumatism; 2005; 1575106445	433	360	73	Cases males: 68.2; Cases females: 65; Control males: 66.8; Control females: 65.8	143	No	No	Yes	No	No	No	No	No	Yes	Case control	12
Kornaat PR; European Radiology; 2005; 15754163127	205	205	0	Median = 60; (Range: 43–77)	163(80%)	No	No	Yes	Yes	No	No	Yes	Yes	No	Cross-sectional	8
Zhai G; Arthritis & Rheumatism; 2005; 15818695128	151	23	128	Men: 64(8.1); Women: 62(7.7)	72	Yes	No	No	Yes	No	No	No	No	No	Cross-sectional	8
Cicuttini F; Osteoarthritis & Cartilage; 2005; 1592263450	28	28	0	62.8(9.8)	(57%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	10
Blankenbaker DG; Skeletal Radiology; 2005; 15940487129	247	74	173	44	126	No	No	Yes	Yes	No	No	No	Yes	Yes	Cross-sectional	6
Huh YM; Korean Journal of Radiology; 2005; 15968151130	94	73	21	RA group: 49.2 (Range: 37–76), Median = 48; OA group: 57.8 (Range: 40–80), Median = 58	73	No	No	Yes	No	Yes	No	No	No	No	Longitudinal Retrospective	7
von Eisenhart-Roth; Annals of the Rheumatic Diseases; 2006; 15975965131	26	26	0	70.4(7.6), (Range: 58–86)	20	Yes	No	No	Yes	No	No	No	No	No	Cross-sectional	5
Tan AL; Arthritis & Rheumatism; 2005; 16052535132	58	40	18	Early OA: 56 (Range: 49–69); Chronic OA: 60 (Range: 51–68); Hand OA: 60 (Range: 46–72);	44	No	No	Yes	Yes	Yes	Yes	Yes	No	Yes	Cross-sectional	7
Lo GH; Arthritis & Rheumatism; 2005; 16145676133	268	80	188	No BMLs: 64.8(8.5); Medial BMLs: 68.3(7); Lateral BMLs: 66.6(9.5)	(59%)	No	No	Yes	No	No	No	Yes	No	No	Cross-sectional	10
Li X; Magnetic Resonance in Medicine; 2005; 16155867134	19	9	10	Cases: Median = 52, (Range: 18–72); Controls: Median = 30, (Range: 22–74)	8	No	Yes	No	Yes	No	No	No	No	No	Case control	7
Rhodes LA; Rheumatology; 2005; 16188949135	35	35	0	Median = 63; (Range: 49–77)	23	No	No	Yes	No	Yes	No	No	No	No	Cross-sectional	9
Williams A; Arthritis & Rheumatism; 2005; 1625502432	31	31	0	67(10.4), 9 (Range: 45–86)	24(77%)	No	Yes	No	Yes	No	No	No	No	No	Cross-sectional	9
Loeuille D; Arthritis & Rheumatism; 2005; 16255041136	39	39	0	56.4(12.71)	(56.4%)	No	No	Yes	No	Yes	No	No	No	No	Cross-sectional	10
Roos EM; Arthritis & Rheumatism; 2005; 16258919137	30			45.8(3.3)	10(33.3%)	No	Yes	No	Yes	No	No	No	No	No	Randomized controlled trial	17
Hunter DJ; Journal of Rheumatology; 2005; 1626570253	132	162	0	33.5(9.7)	(44.2%)	No	No	Yes	Yes	No	Yes	No	Yes	Yes	Cross-sectional	8
Nojiri T; Knee Surgery, Sports Traumatology, Arthroscopy; 2006; 1639556433	28	9	21	40.3(Range: 16–74)	17	No	Yes	No	Yes	No	No	No	No	No	Cross-sectional	7
Kimelman T; Invest Radiol; 2006; 16428993138	7	4	3	Healthy controls: 23; OA cases: 56	4	No	Yes	No	Yes	No	No	No	No	No	Other	6
Sengupta M; Osteoarthritis & Cartilage; 2006; 16442316139	217	217	0	67.3(9.1)	(30%)	No	No	Yes	Yes	Yes	Yes	Yes	No	No	Cross-sectional	7
Hunter DJ; Arthritis & Rheumatism; 2006; 1650893081	257	257	0	66.6(9.2), (Range: 47–93)	(41.6%)	No	No	Yes	Yes	No	No	No	Yes	No	Longitudinal Prospective	10
Hunter DJ; Arthritis & Rheumatism; 2006; 1664603783	217	217	0	66.4(9.4)	(44%)	No	No	Yes	Yes	No	No	Yes	No	No	Longitudinal Prospective	10
Grainger AJ; European Radiology; 2007; 16685505140	43	43	0	64(Range: 48–75)	19	No	No	Yes	No	Yes	No	No	Yes	No	Cross-sectional	8
Cashman PM; IEEE Transactions on Nanobioscience; 2002; 16689221141	27	10	17	OA patients: (Range: 45–73); Similar age controls: (Range: 50–65); Young healthy controls: (Range: 21–32);	8(29.6%)	Yes	No	No	Yes	No	No	No	No	No	Other	6
Torres L; Osteoarthritis & Cartilage; 2006; 1671331021	143	143	0	70(10)	(78%)	No	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Cross-sectional	9
Kornaat PR; Radiology; 2006; 1671446322	205	97	103	60 (Range: 43–77)	163(80%)	No	No	Yes	Yes	Yes	Yes	Yes	Yes	No	Cross-sectional	9
Bamac B; Saudi Medical Journal; 2006; 16758050142	46	36	10	Cases: 41.9 (Range: 20–67); Controls: 39.7 (Range: 21–66)	25	No	No	No	No	No	No	No	Yes	No	Case control	8
Boks SS; American Journal of Sports Medicine; 2006; 16861575143	134	136	132	40.8(Range: 18.8–63.8)		No	No	Yes	Yes	No	No	No	Yes	Yes	Cross-sectional	7
Koff MF; Osteoarthritis & Cartilage; 2007; 1694931334	113	113	0	56(11), (Range: 33–82)	84	No	Yes	No	Yes	No	No	No	No	No	Cross-sectional	8
Nakamura M; Magnetic Resonance Imaging; 2006; 1707133639	63			51.8 (Range: 40–59)	42	No	No	Yes	No	No	No	No	Yes	No	Cross-sectional	6
Folkesson J; IEEE Transactions on Medical Imaging; 2007; 17243589144	139			56(Range: 22–79)	(59%)	Yes	No	No	Yes	No	No	No	No	No	Other	7
Li X; Osteoarthritis & Cartilage; 2007; 1730736535	26	10	16	Healthy: 41.3 (Range: 22–74); OA patients: 55.9 (37–72)	11	No	Yes	Yes	Yes	No	Yes	Yes	Yes	Yes	Case control	7
Iwasaki J; Clinical Rheumatology; 2007; 17322963145	26	26	0	63.8(Rang: 49–82)	18	No	No	No	No	No	No	No	No	No	Cross-sectional	5
Dam EB; Osteoarthritis & Cartilage; 2007; 1735313231	139			Evaluation set: 55(Range: 21–78); Scan-rescan set: 61 (Range: 26–75)	(54.5%)	Yes	No	No	Yes	No	No	No	No	No	Other	9
Tiderius CJ; Magnetic Resonance in Medicine; 2007; 17390362146	18	10	8	Controls: 28(Range: 20–47); Cases: 39 (Range: 25–58)		No	Yes	No	Yes	No	No	No	No	No	Case control	6
Baranyay FJ; Seminars in Arthritis & Rheumatism; 2007; 17391738147	297		297	58(5.5)	(63%)	Yes	No	No	Yes	No	No	Yes	No	No	Cross-sectional	16
Issa SN; Arthritis & Rheumatism; 2007; 1739422554	146	146	0	70	109	No	No	Yes	Yes	No	Yes	Yes	Yes	No	Cross-sectional	8
Hanna F; Menopause; 2007; 17413649148	176	0	176	52.3(6.6), (Range: 40–67)	176(100%)	Yes	No	No	Yes	No	No	No	No	No	Cross-sectional	13
Hunter DJ; Annals of the Rheumatic Diseases; 2008; 1747299523	71			67.9(9.3)	(28.2%)	No	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Other	8
Hill CL; Annals of the Rheumatic Diseases; 2007; 1749109624	270	270	0	66.7(9.2)	112	No	No	Yes	Yes	Yes	No	No	No	No	Longitudinal Prospective	9
Qazi AA; Osteoarthritis & Cartilage; 2007; 17493841149	71					Yes	No	No	Yes	No	No	No	No	No	Cross-sectional	8
Lammentausta E; Osteoarthritis & Cartilage; 2007; 17502160150	14			55(18)	2	No	Yes	No	Yes	No	Yes	No	No	No	Other	5
Guymer E; Osteoarthritis & Cartilage; 2007; 17560134151	176	0	176	52.3(6.6)	176(100%)	Yes	No	Yes	Yes	No	Yes	Yes	No	No	Cross-sectional	11
Nishii T; Osteoarthritis & Cartilage; 2008; 17644363152	33	23	10	Volunteers: 34(Range: 23–51); Patients: 40(Range: 22–69)	33(1005)	No	Yes	No	Yes	No	No	No	No	No	Case control	8
Janakiramanan N; Journal of Orthopaedic Research; 2008; 1776345155	202	74	128	61(9)	(73%)	No	No	Yes	Yes	No	No	No	No	No	Cross-sectional	11
Lo GH; Osteoarthritis & Cartilage; 2008; 17825586153	845	170		63.6(8.8)	(58%)	No	No	Yes	No	No	No	No	Yes	No	Cross-sectional	10
Davies-Tuck M; Osteoarthritis & Cartilage; 2008; 17869546154	100	100	0	63.3(10.2)	61(61%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	11
Qazi AA; Academic Radiology; 2007; 17889338155	159			(Range: 21–81)		No	No	Yes	Yes	No	No	No	No	No	Other	8
Folkesson J; Academic Radiology; 2007; 17889339156	71			56(Range: 22–79)	(59%)	No	No	No	No	No	No	No	No	No	Other	7
Englund M; Arthritis & Rheumatism; 2007; 1805020140	310	102	208	Cases: 62.9(8.3); Controls: 61.2(8.3)	211(68%)	No	No	Yes	No	No	No	No	Yes	No	Case control	15
Kamei G; Magnetic Resonance Imaging; 2008; 18083319157	37	27	0	Cartilage defect: 51.6(Range: 42–61); No cartilage defect: 54.5(Range: 45–61)	20	No	No	Yes	Yes	No	No	No	Yes	No	Case control	7
Li W; Journal of Magnetic Resonance Imaging; 2008; 18183573158	29	19	10	OA subjects: 61.7(Range: 40–86); Controls: 31 (Range: 18–40)	19	No	Yes	No	Yes	No	No	No	No	No	Cross-sectional	5
Amin S; Osteoarthritis & Cartilage; 2008; 1820362986	265	265		67(9)	(43%)	No	No	Yes	Yes	No	No	No	Yes	Yes	Longitudinal Prospective	11
Taljanovic MS; Skeletal Radiology; 2008; 18274742159	19	19	0	66	8	No	Yes	No	No	No	No	No	No	No	Case control	8
Oda H; Journal of Orthopaedic Science; 2008; 18274849160	161			58.5(Range: 11–85)	98	No	No	Yes	No	Yes	No	No	Yes	Yes	Cross-sectional	8
Hanna FS; Arthritis Research & Therapy; 2008; 18312679161	176			52.3(6.6)	(100%)	Yes	No	No	Yes	No	No	No	No	No	Cross-sectional	10
Reichenbach S; Osteoarthritis & Cartilage; 2008; 1836741541	964	217	747	63.3	(57%)	No	No	Yes	Yes	No	Yes	No	No	No	Cross-sectional	8
Petterson SC; Medicine & Science in Sports & Exercise; 2008; 18379202162	123	123	0	64.9(8.5)	67	No	No	No	No	No	No	No	No	No	Case control	11
Bolbos RI; Osteoarthritis & Cartilage; 2008; 18387828163	32	16	16	Cases: 47.2(11.54), (Range: 29–72); Controls: 36.3(10.54), (Range: 27–56)	14	Yes	Yes	No	Yes	No	Yes	No	No	No	Case control	7
Quaia E; Skeletal Radiology; 2008; 18404267164	35	35	0	42(17), (Range: 22–67)	14	No	Yes	No	Yes	No	No	No	No	No	Other	6
Folkesson J; Magnetic Resonance in Medicine; 2008; 1850684542	245		143	KL0: 48(Range: 21–78); KL1: 62(Range: 37–81); KL2: 67(Range: 47–78); KL3&4: 68(Range: 58–78)		No	No	No	Yes	No	No	No	No	No	Other	12
Mills PM; Osteoarthritis & Cartilage; 2008; 18515157165	49	25	24	APMM: 46.8(5.3); Controls: 43.6(6.6)	18(36.7%)	Yes	No	Yes	Yes	No	No	No	No	No	Case control	12
Dore D; Osteoarthritis & Cartilage; 2008; 18515160166	50	50		64.5(7.1)	23	Yes	No	Yes	Yes	No	Yes	No	No	No	Cross-sectional	9
Mutimer J; Journal of Hand Surgery; 2008; 18562375167	20	20	0	47 (Range: 26–69)	9	No	No	Yes	Yes	No	No	No	No	No	Cross-sectional	6
Amin S; Journal of Rheumatology; 2008; 18597397168	192	192		69(9)	0.	No	No	Yes	Yes	No	No	No	No	No	Cross-sectional	10
Li X; Journal of Magnetic Resonance Imaging; 2008; 18666183169	38	13	25	Healthy: 28.5 (Range: 20–34); Knee OA or injury: 37.4 (Range: 20–66)	10	Yes	No	No	Yes	No	No	Yes	No	No	Other	7
Pelletier JP; Osteoarthritis & Cartilage; 2008; 1867238625	27	1		64.1(9.6)	14	No	No	Yes	Yes	Yes	Yes	Yes	Yes	No	Other	9
Stahl R; European Radiology; 2009; 18709373170	37	17	20	Mild OA: 54(9.98); Healthy control: 33.6(9.44)	19	No	Yes	Yes	Yes	No	No	No	No	No	Case control	10
Brem MH; Acta Radiologica; 2008; 18720084171	23	23	0	55.5(10.3)	8	No	No	Yes	No	No	No	Yes	No	No	Other	6
Lancianese SL; Bone; 2008; 18755303172	4			80(14)	3	No	No	No	No	No	Yes	No	No	No	Cross-sectional	5
Englund M; New England Journal of Medicine; 2008; 1878410026	991	171		62.3(8.6), (Range: 50.1–90.5)	565(57%)	No	No	Yes	No	No	No	No	Yes	No	Cross-sectional	10
Mamisch TC; Magnetic Resonance in Medicine; 2008; 18816842173	26					No	Yes	No	Yes	No	No	No	No	No	Cross-sectional	7
Rauscher I; Radiology; 2008; 18936315174	60	37	23	Healthy controls: 34.1(10); Mild OA: 52.5(10); Severe OA: 61.6(11.6)	32	No	Yes	No	Yes	No	No	No	Yes	No	Case control	9
Li W; Journal of Magnetic Resonance Imaging; 2009; 19161210175	31	17	14	OA patients: 61.8(Range: 40–86); Healthy controls: 29.2(Range: 18–40)	21	No	Yes	No	Yes	No	No	No	No	No	Case control	7
Choi JW; Journal of Computer Assisted Tomography; 2009; 19188805176	36			39.7(Range: 8–69)	21	No	No	Yes	Yes	No	No	No	Yes	No	Longitudinal Retrospective	7
Chen YH; Journal of Computer Assisted Tomography; 2008; 19204464177	96	25	71	OA patients: 56; Non-OA: 46		No	No	Yes	Yes	No	No	No	Yes	No	Case control	8

The analysis included data from 142 manuscripts. The mean Downs criteria score for these manuscripts was 8.3 (range 3–17). What follows below are important excerpts from this data pertaining to different aspects of concurrent validity. The data is further summarized in Table II to discretely identify the associations examined and those where a significant association was found.

Table II.

Summary of Concurrent Validity of MRI in OA

Outcome of interest	Number of studies examining this outcome	Number of studies finding significant associations (P < .05)
Symptoms	21 studies	13 of 21 (62%)
Radiographic features	43 studies	39 of 43 (90%)
Radiographic joint space	9 studies	9 of 9 (100%)
Alignment	10 studies	9 of 10 (90%)
CT	4 studies	4 of 4 (100%)
Histology/Pathology	5 studies	3 of 5 (60%)
Arthroscopy	7 studies	5 of 7 (71%)

Relation to symptoms

21 studies examined the concurrent relation of MRI findings in OA to symptoms. Of these, 62% demonstrated a statistically significant association, defined as P < 0.05. Bone marrow lesions were found in 272 of 351 (77.5%) persons with painful knees compared with 15 of 50 (30%) persons with no knee pain (P < 0.001). Large lesions were present almost exclusively in persons with knee pain (35.9% vs 2%; P < 0.001). After adjustment for severity of radiographic disease, effusion, age, and sex, lesions and large lesions remained associated with the occurrence of knee pain [odds ratio, 3.31 (95% confidence interval (CI), 1.54–7.41)]. Using the same analytical approach, large lesions were also strongly associated with the presence of pain [odds ratio, 5.78 (CI, 1.04–111.11)]. Among persons with knee pain, bone marrow lesions were not associated with pain severity¹⁴.

• After adjusting for the severity of radiographic OA, there was a difference between those with and without knee pain in prevalence of moderate or larger effusions (P < 0.001) and synovial thickening, independent of effusion (P < 0.001). Among those with small (grade 1) or no knee (grade 0) effusion, those with knee pain had a prevalence of synovial thickening of 73.6% compared to 21.4% of those without knee pain (P < 0.001). There was a significant difference in visual analogue scale (VAS) pain scores in those with synovial thickening compared to those without synovial thickening, after adjustment for radiographic severity, size of effusion, age, sex, and BMI. The mean pain score in those with synovial thickening after adjustment for radiographic severity and size of effusion was 47.2 mm [standard error (SE) 6.0], compared to 28.2 mm (SE 2.8) in those without synovial thickening (P = 0.006)¹⁵.

• A medial or lateral meniscal tear was a very common finding in the asymptomatic subjects (prevalence, 76%) but was more common in the patients with symptomatic osteoarthritis (91%) (P < 0.005). There was no significant difference with regard to the pain or WOMAC score between the patients with and those without a medial or lateral meniscal tear in the osteoarthritic group (P = 0.8 to 0.9 for all comparisons)¹⁶.

• Significant differences between WOMAC scores were found for the grades of cartilage lesions (P < 0.05) but not bone marrow edema pattern, and ligamentous and meniscal lesions¹⁷.

• Bone marrow lesions >1 cm were more frequent (OR = 5.0; 95% CI = 1.4, 10.5) in the painful knee OA group than all other groups. While the frequency of BME lesions was similar in the painless OA and painful OA groups, there were more lesions, >1 cm, in the painful OA group. Full-thickness cartilage defects occurred frequently in painful OA. Women with radiographic OA, full-thickness articular cartilage defects, and adjacent subchondral cortical bone defects were significantly more likely to have painful knee OA than other groups (OR = 3.2; 95% CI = 1.3, 7.6)¹⁸.

• Peripatellar lesions (prepatellar or superficial infrapatellar) were present in 12.1% of the patients with knee pain and ROA, in 20.5% of the patients with ROA and no knee pain, and in 0% of subjects with neither ROA nor knee pain (P = 0.116). However, other periarticular lesions (including bursitis and iliotibial band syndrome) were present in 14.9% of patients with both ROA and knee pain, in only 3.9% of patients with ROA but no knee pain, and in 0% of the group with no knee pain and no ROA (P = 0.004)¹⁹.

• More severe symptoms relating to knee OA (pain, stiffness, and function) are weakly inversely related to tibial cartilage volume. Patients with lower cartilage volume had more severe symptoms of knee OA than those with higher cartilage volume²⁰.

• The increase in median pain from median quantile regression, adjusting for age and BMI, was significant for bone attrition (1.91, 95% CI 0.68, 3.13), bone marrow lesions (3.72, 95% CI 1.76, 5.68), meniscal tears (1.99, 95% CI 0.60, 3.38), and grade 2 or 3 synovitis/effusion vs grade 0 (9.82, 95% CI 0.38, 19.27). The relationship with pain severity was of borderline significance for osteophytes and cartilage morphology and was not significant for bone cysts or meniscal subluxation. When compared to the pain severity in knees with high scores for both bone attrition and bone marrow lesions (median pain severity 40 mm), knees with high attrition alone (30 mm) were not significantly different, but knees with high bone marrow lesion without high attrition scores (15 mm) were significantly less painful²¹.

• A large joint effusion was associated with pain (OR, 9.99; 99% CI: 1.28, 149) and stiffness (OR, 4.67; 99% CI: 1.26, 26.1). The presence of an osteophyte in the patellofemoral compartment (OR, 2.25; 99% CI: 1.06, 4.77) was associated with pain. All other imaging findings, including focal or diffuse cartilaginous abnormalities, subchondral cysts, bone marrow edema, subluxation of the meniscus, meniscal tears, or Baker cysts, were not associated with symptoms²².

• Maximal bone marrow lesion (BML) size on the Boston Leeds Osteoarthritis Score (BLOKS) scale had a positive linear relation with VAS pain (P for linear trend = 0.04)²³.

• No correlation of baseline synovitis with baseline pain score (r = 0.09, P = 0.17)²⁴.

• No relation between baseline synovitis score and VAS pain score (r = 0.11, P = 0.60)²⁵.

• In the group of persons with radiographic evidence of osteoarthritis (Kellgren–Lawrence grade 2 or higher), the prevalence of a meniscal tear was 63% among those with knee pain, aching, or stiffness on most days and 60% among those without these symptoms (P = 0.75); the corresponding prevalences in the group without radiographic evidence of osteoarthritis were 32% and 23% (P = 0.02). The majority of the meniscal tears – 180 of 297 (61%) were in subjects who had not had any pain, aching, or stiffness in the previous month²⁶.

Relation to radiographic features

43 studies examined the concurrent relation of MRI findings in OA to radiographic features. Of these, 90% demonstrated a statistically significant association, defined as P < 0.05.

Relation of quantitative cartilage morphometry measures to radiographic abnormalities

• Significant differences in lateral and medial femorotibial cartilage thickness were found between those with and without radiographic OA. Significant cartilage thinning could be detected by MRI in patients with OA, even when the joint space was normal radiographically²⁷.

• For every increase in grade of lateral tibiofemoral osteophytes the lateral tibial cartilage volume was significantly reduced by 255 mm³, after adjustment. There was a reduction of 77 mm³ in medial tibial cartilage volume for every increase in grade of medial tibiofemoral osteophytes, but this finding was only of borderline statistical significance²⁸.

• Cartilage volume and thickness were less in patients with OA compared to normal controls (P < 0.1)²⁹.

• Kellgren and Lawrence (KLG)2 participants displayed, on average, thicker cartilage than healthy controls in the medial femorotibial compartment [particularly anterior subregion of the medial tibia (MT) and peripheral (external, internal) subregions of the medial femur], and in the lateral femur. KLG3 participants displayed significantly thinner cartilage than KLG0 participants in the medial weight-bearing femur (central subregion), in the external subregion of the MT, and in the internal subregion of the lateral tibia³⁰.

• Mean cartilage signal intensity provided a clear separation of healthy from KLG1 (P = 0.0009). Quantification of cartilage homogeneity by entropy was able to clearly 11 separate healthy from OA subjects (P = 0.0003). Furthermo121re, entropy was also able to separate healthy from KL 1 subjects (P = 0.0004)³¹.

Relation of other MRI measures to radiographic abnormalities

• Significant difference (P = 0.002) in the average T(1rho) within patellar and femoral cartilage between controls (45.04 ± 2.59 ms) and osteoarthritis patients (53.06 ± 4.60 ms). A significant correlation was found between T(1rho) and T(2); however, the difference of T(2) was not statistically significant between controls and osteoarthritis patients³¹.

• Trend toward a lower dGEMRIC index with increasing KLG; the spared compartments of knees with a KLG grade 2 had a higher dGEMRIC index than those of knees with a KLG grade 4 (mean 425 msec vs 371 msec; P < 0.05)³².

• All cases demonstrating decreased T1 values on dGEMRIC, showed abnormal arthroscopic or direct viewing findings. The diagnosis of damage in articular cartilage was possible in all 16 cases with radiographic KLG 1 on dGEMRIC, while the intensity changes were not found in 10 of 16 cases on Proton density Weighted Image (PDWI)³³.

• No differences of T2 values were found across the stages of OA (P = 0.25), but the factor of BMI did have a significant effect P < 0.0001) on T2 value³⁴.

• Average T(1rho) and T(2) values were significantly increased in OA patients compared with controls [52.04 ± 2.97 ms vs 45.53 ± 3.28 ms with P = 0.0002 for T(1rho), and 39.63 ± 2.69 ms vs 34.74 ± 2.48 ms with P = 0.001 for T(2)]. Increased T(1rho) and T(2) values were correlated with increased severity in radiographic and MR grading of OA. T(1rho) has a larger range and higher effect size than T(2), 3.7 vs 3.0³⁵.

• Statistically significant correlation between radiography and MR cartilage loss in the medial (r 0.7142, P .0001) and lateral compartments (r = 0.4004, P .0136). Significant correlations also found between radiographic assessment of sclerosis and osteophytes and those found on MRI³⁶.

• Patients in whom plain radiographs, MRI, and arthroscopy were compared, the plain radiographs and MRI significantly underestimated the extent of cartilage abnormalities³⁷.

• Presence of synovial thickening was more likely with increasing KLG, from 24.0% in those with KLG 0–78.3% in those with LG 3/4 (P < 0.001)¹⁵.

• Higher KLG was correlated with a higher frequency of meniscal tears (r = 0.26, P < 0.001)¹⁶.

• KLG correlated significantly (P < 0.05) with the grade of cartilage lesions, and a substantially higher percentage of bone marrow and meniscal lesions with higher KLG found on MR images¹⁷.

• Women with osteoarthritis had larger medial and lateral tibial plateau bone area [mean (SD): 1850 (240) mm² and 1279 (220) mm², respectively] than healthy women [1670 (200) mm² and 1050 (130) mm²] (P < 0.001 for both differences). For each increase in grade of osteophyte, an increase in bone area was seen of 146 mm² in the medial compartment and 102 mm² in the lateral compartment³⁸.

• Statistically significant correlations were observed between the medial tibial spur classification on X-ray, the medial meniscal displacement rate on MRI and the medial meniscal signal change classification on MRI³⁹.

• Meniscal damage was mostly present in knees with OA and demonstrates a relation to KLG⁴⁰.

• Bone attrition of the tibiofemoral joint, scored >1, was found in 228 MRIs (23.6%) and in 55 radiographs (5.7%). Moderate to strong correlation between MRIs and radiographs for bone attrition of the tibiofemoral joint (r = 0.50, P < 0.001)⁴¹.

• Surface curvature of articular cartilage for both the fine- and coarse-scale estimates were significantly higher in the OA population compared with the healthy population, with P < 0.001 and P < 0.001, respectively⁴².

• The prevalence of meniscal damage was significantly higher among subjects with radiographic evidence of tibiofemoral osteoarthritis (KLG 2 or higher) than among those without such evidence (82% vs 25%, P < 0.001), and the prevalence increased with a higher KLG (P < 0.001 for trend). Among persons with radiographic evidence of severe osteoarthritis (KLG 3 or 4 in their right knee), 95% had meniscal damage²⁶.

Relation to radiographic joint space width

Nine studies examined the concurrent relation of MRI findings in OA to radiographic joint space. Of these, 100% demonstrated a statistically significant association, defined as P < 0.05.

• Strong correlation between the degree of medial meniscal subluxation and the severity of medial joint space narrowing (JSN) (r = 0.56, P = 0.0001)⁴³.

• Meniscal extrusion identified in all 32 patients with JSN (KLG 1–4). Definite thinning or loss of articular cartilage was identified in only 15 of the 32 cases. In 17 patients with radiographic JSN (KLG 1–3) and meniscal extrusion, no loss of articular cartilage was observed. A statistically significant correlation (P < 0.001) was observed between KLG and degree of meniscal extrusion and cartilage thinning on MRI⁴⁴.

• For each increase in grade of JSN, tibial plateau bone area increased by 160 mm² in the medial compartment and 131 mm² in the lateral compartment (significance of regression coefficients all P < 0.001)³⁸.

• Persons with symptomatic knee OA with ACL rupture had more severe radiologic OA (P < 0.0001) and were more likely to have medial JSN (P < 0.0001) than a control sample⁴⁵.

• Compartments of the knee joint without JSN had a higher dGEMRIC index than those with any level of narrowing (mean 408 msec vs 365 msec; P = 0.001). In knees with 1 unnarrowed (spared) and 1 narrowed (diseased) compartment, the dGEM-RIC index was greater in the spared vs the diseased compartment (mean 395 msec vs 369 msec; P = 0.001)³².

• Grade of JSN as measured on skyline and lateral patellofemoral radiographs was inversely associated with patella cartilage volume. After adjusting for age, gender and body mass index, for every increase in grade of skyline JSN (0–3), the patella cartilage volume was reduced by 411 mm³. For every increase in lateral patellofemoral JSN grade (0–3), the adjusted patella cartilage volume was reduced by 125 mm³. The relationship was stronger for patella cartilage volume and skyline JSN (r = −0.54, P < 0.001) than for lateral patellofemoral JSN (r = −0.16, P = 0.015)⁴⁶.

• Grade one medial JSN was associated with substantial reductions in cartilage volume at both the medial and lateral tibial and patellar sites within the knee (adjusted mean difference 11–13%, all P < 0.001)⁴⁷.

• Cartilage volume in the medial compartment and the narrowest JSW obtained by radiography at baseline in 31 knee OA patients, revealed that some level of correlation exists between these two measurements (r = 0.46, P < 0.007)⁴⁸.

• Knee cartilage defects are inconsistently associated with JSN after adjustment for osteophytes but consistently with knee cartilage volume (beta: −0.27 to −0.70/ml; OR: 0.16–0.56/ml, all P < 0.01 except for OR at lateral tibial cartilage site P = 0.06)⁴⁹.

• Moderate, but statistically significant, correlation between JSW and femoral and tibial cartilage volumes in the medial tibiofemoral joint, which was strengthened by adjusting for medial tibial bone size (R = 0.58–0.66, P = 0.001)⁵⁰.

• JSN seen on both medial and lateral radiographs of the tibiofemoral joint was inversely associated with the respective tibial cartilage volume. This inverse relationship was strengthened with adjustment for age, sex, body mass index (BMI), and bone size. After adjustment for these confounders, for every increase in JSN grade (0–3), the medial tibial cartilage volume was reduced by 257 mm³ (95% CI 193–321) and the lateral tibial cartilage volume by 396 mm³ (95% CI 283–509). The relationship between mean cartilage volume and radiologic grade of JSN was linear²⁸.

Relation to alignment

10 studies examined the concurrent relation of MRI findings in OA to alignment. Of these, 90% demonstrated a statistically significant association, defined as P < 0.05.

• Valgus-aligned knees tended to have lower dGEMRIC values laterally, and varus-aligned knees tended to have lower dGEMRIC values medially; as a continuous variable, alignment correlated with the lateral: medial dGEMRIC ratio (Pearson’s R = 0.43, P = 0.02)³².

• Limbs with varus alignment, especially if marked (≥7 degrees), had a remarkably high prevalence of medial lesions compared with limbs that were neutral or valgus (74.3% vs 16.4%; P < 0.001 for relation between alignment and medial lesions). Conversely, limbs that were neutral or valgus had a much higher prevalence of lateral lesions than limbs that were in the most varus group (29.5% vs 8.6%; P = 0.002 for alignment and lateral lesions)⁵¹.

• Medial tibial and femoral cartilage volumes increased as the angle decreased (i.e., was less varus). Similarly, in the lateral compartment there was an inverse association at baseline between tibial and femoral cartilage volumes and the measured knee angle⁵².

• The main univariate determinants of varus alignment in decreasing order of influence were medial bone attrition, medial meniscal degeneration, medial meniscal subluxation, and medial tibiofemoral cartilage loss. Multivariable analysis revealed that medial bone attrition and medial tibiofemoral cartilage loss explained more of the variance in varus malalignment than other variables. The main univariate determinants of valgus malalignment in decreasing order of influence were lateral tibiofemoral cartilage loss, lateral osteophyte score, and lateral meniscal degeneration⁵³.

• Correlation between medial meniscal displacement rate on MRI and the femorotibial angle (r = 0.398)³⁹.

• Worsening in the status of each medial lesion cartilage morphology, subarticular bone marrow lesions, meniscal tear, meniscal subluxation, and bone attrition was associated with greater varus malalignment⁵⁴.

• For every one degree increase in a valgus direction, there was an associated reduced risk of the presence of cartilage defects in the medial compartment of subjects with knee OA (P = 0.02). Moreover, for every one degree increase in a valgus direction, there was an associated increased risk of the presence of lateral cartilage defects in the OA group (P = 0.006)⁵⁵.

Relation to CT

Four studies examined the concurrent relation of MRI findings in OA to CT. Of these, 100% demonstrated a statistically significant association, defined as P < 0.05. MR frequently showed tricompartmental cartilage loss when radiography and CT showed only bicompartmental involvement in the medial and patellofemoral compartments. In the lateral compartment, MR showed a higher prevalence of cartilage loss (60%) than radiography (35%) and CT (25%) did. In the medial compartment, CT and MR showed osteophytes in 100% of the knees, whereas radiography showed osteophytes in only 60%. Notably, radiography often failed to show osteophytes in the posterior medial femoral condyle. On MR images, meniscal degeneration or tears were found in all 20 knees studied. Partial and complete tears of the anterior cruciate ligament were found in three and seven patients, respectively. MR is more sensitive than radiography and CT for assessing the extent and severity of osteoarthritic changes and frequently shows tricompartmental disease in patients in whom radiography and CT show only bicompartmental involvement. MR imaging is unique for evaluating meniscal and ligamentous disease related to osteoarthritis³⁶.

• Strong linear relationship (r = 0.998) between MRI imaging and CT arthrography. The mean absolute volume deviation between magnetic resonance imaging and computed tomography arthrography was 3.3%⁵⁶.

Relation to histology/pathology

Five studies examined the concurrent relation of MRI findings in OA to histology/pathology. Of these, 60% demonstrated a statistically significant association, defined as P < 0.05. Observed measurements of MRI volume of articular cartilage correlated with actual weight and volume displacement measurements with an accuracy of 82%–99% and linear correlation coefficients of 0.99 (P = 2.5e-15) and 0.99 (P = 4.4e-15)⁵⁷.

• The signal behavior of hyaline articular cartilage does not reflect the laminar histologic structure. Osteoarthrosis and cartilage degeneration are visible on MR images as intra-cartilaginous signal changes, superficial erosions, diffuse cartilage thinning, and cartilage ulceration⁵⁸.

• Comparison of data on cartilage thickness measurements with MRI with corresponding histological sections in the middle of each sector revealed a very good magnetic resonance/anatomic correlation (r = 0.88)⁵⁹.

• Correlation between MRI Noyes grading scores and Mankin grading scores of natural lesions was moderately high (r = 0.7) and statistically significant (P = 0.001)⁶⁰.

Relation to arthroscopy

Seven studies examined the concurrent relation of MRI findings in OA to arthroscopy. Of these, 71% demonstrated a statistically significant association, defined as P < 0.05.

• Moderate correlation between imaged cartilage scores and the arthroscopy scores (Pearson correlation coefficient = 0.40)³⁷.

• Spearman rank linear correlation between arthroscopic and MR cartilage grading was highly significant (P < 0.002) for each of the six articular regions evaluated. The MR and arthroscopic grades were the same in 93 (68%) of 137 joint surfaces, they were the same or differed by one grade in 123 surfaces (90%), and they were the same or differed by one or two grades in 129 surfaces (94%)⁶¹.

• The overall sensitivity and specificity of MR in detecting chondral abnormalities were 60.5% (158/261) and 93.7% (89/95) respectively. MR imaging was more sensitive to the higher grade lesions: 31.8% (34/107) in grade 1; 72.4% (71/98) in grade 2; 93.5% (43/46) in grade 3; and 100% (10/10) in grade 4. The MR and arthroscopic grades were the same in 46.9% (167/356), and differed by no more than 1 grade in 90.2% (321/356) and 2 grades in 99.2% (353/356). The correlation between arthroscopic and MR grading scores was highly significant with a correlation coefficient of 0.705 (P < 0.0001)⁶².

• Statistically significant correlation between the SFA-arthroscopic score and the SFA-MR score (r = 0.83) and between the SFA-arthroscopic grade and the SFA-MR grade (weighted kappa = 0.84). The deepest cartilage lesions graded with arthroscopy and MR imaging showed correlation in the medial femoral condyle (weighted kappa = 0.83) and in the medial tibial plateau (weighted kappa = 0.84)⁶³.

• Magnetic resonance imaging was in agreement with arthroscopy in 81% showing more degeneration but less tears of menisci than arthroscopy. Using a global system for grading the total damage of the knee joint into none, mild, moderate, or severe changes, agreement between arthroscopy and MRI was found in 82%⁶⁴.

Predictive validity (Table III)

Table III.

Summary table of studies reporting data on predictive validity of MRI in knee OA

Reference: Author, Journal, Year, PMID	Whole sample size	No. of cases	No. of controls	Age, yrs, Mean(SD), Range	No. (%)of females	Quantitative cartilage	Compositional techniques	Semi-quantitative	Cartilage	Synovium	Bone	Bone marrow lesions	Meniscus	Ligament	Study design	Score of methodological quality
Boegard TL; Osteoarthritis & Cartilage; 2001; 11467896178	47			Women: Median = 50, (Range: 42–57); Men: Median = 50, (Range: 41–57)	25(53.2%)	No	No	Yes	Yes	No	No	No	Yes	No	Longitudinal Prospective	9
Wluka AE; Arthritis & Rheumatism; 2002; 12209510179	123	123	0	63.1(10.6)	71	Yes	No	No	Yes	No	Yes	No	No	No	Longitudinal Prospective	14
Cicuttini FM; Journal of Rheumatology; 2002; 12233892180	21	8	13	Case: 41.3(13.2); Controls: 49.2(17.8)	14(66.7%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Retrospective	13
Biswal S; Arthritis & Rheumatism; 2002; 1242822876	43	4	39	54.4(Range: 17–65)	21	No	No	Yes	Yes	No	No	Yes	Yes	Yes	Longitudinal Retrospective	8
Cicuttini F; Journal of Rheumatology; 2002; 12465162181	110	110	0	63.2(10.2)	66	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	12
Pessis E; Osteoarthritis & Cartilage; 2003; 12744942182	20	20		63.9(9)	13	Yes	No	Yes	Yes	No	Yes	Yes	No	No	Longitudinal Prospective	12
Felson DT; Annals of Internal Medicine; 2003; 1296594151	256	156	0	Followed: 66.2(9.4); Not followed: 67.8(9.6)	(38.3%)	No	No	Yes	No	No	No	Yes	No	No	Longitudinal Prospective	11
Cicuttini FM; Arthritis & Rheumatism; 2004; 1473060477	117	117		63.7(10.2)	(58.1%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	9
Wluka AE; Annals of the Rheumatic Diseases; 2004; 1496296020	132	132	0	63.1(Range: 41–86)	71(54%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	10
Cicuttini F; Rheumatology; 2004; 1496320152	117	117	0	67(10.6)	(58%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	12
Cicuttini FM; Ann Rheum Dis; 2004; 1511571465	123	123	0	Joint replacement: 64.1(9.3); No joint replacement: 63.1(10.3)	65(52.8%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	11
Dashti M; Scandinavian Journal of Rheumatology; 2004; 15163109118	174	117	57	61.6(9.5)	123(70.7%) Yes		No	No	Yes	No	No	No	No	No	Case control	11
Cicuttini FM; Journal of Rheumatology; 2004; 15229959183	102	102	0	63.8(10.1)	(63%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	10
Berthiaume MJ; Annals of the Rheumatic Diseases; 2005; 1537485578	32					Yes	No	Yes	Yes	No	No	No	Yes	No	Longitudinal Prospective	10
Cicuttini F; Journal of Rheumatology; 2004; 15570649126	123					Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	6
Cubukcu D; Clinical Rheumatology; 2005; 15599642184	40	40		HA group: 52.6(7.16); Saline group: 57.6(2.77)	24(60%)	No	No	Yes	Yes	No	Yes	No	Yes	Yes	Randomized controlled trial	15
Ozturk C; Rheumatol Int; 2006;15703953185	47	47	0	HA-only group: 58(7.7); HA&Cortico group: 58.1(10.3)	39(97.5%)	No	No	Yes	Yes	No	No	Yes	No	No	Randomized controlled trial	17
Wang Y; Arthritis Res Ther; 2005; 15899054186	126	126		63.6(10.1)	68	No	No	No	No	No	Yes	No	No	No	Longitudinal Prospective	12
Cicuttini F; Osteoarthritis & Cartilage; 2005; 1592263450	28	28	0	62.8(9.8)	(57%)	Yes	No	No	Yes	No	No	No	No	No	Longitudinal Prospective	10
Wluka AE; Rheumatology; 2005; 1603008466	126	126	0	63.6(10.1)	68(54%)	Yes	No	Yes	Yes	No	Yes	No	No	No	Longitudinal Prospective	14
Garnero P; Arthritis & Rheumatism; 2005; 16145678187	377	377	0	62.5(8.1)	(76%)	No	No	Yes	Yes	No	No	No	No	No	Longitudinal Prospective	11
Wang Y; Rheumatology; 2006; 1618894779	124	124	0	Females: 57.1(5.8); Males: 52.5(13.2)	81(65.3%)	No	No	Yes	Yes	No	Yes	No	No	No	Longitudinal Prospective	11
Phan CM; European Radiology; 2006; 1622253368	40	34	6	57.7(15.6), (Range: 28–81)	16	No	No	Yes	Yes	No	Yes	Yes	Yes	Yes	Longitudinal Prospective	7
Hayes CW; Radiology; 2005; 16251398188	117	117	115	No OA, No Pain: 44.6(10.7); OA, No Pain: 16.2(0.8); No OA, Pain: 47(0.7); OA&Pain: 47.1(0.8)	(100%)	No	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Longitudinal Prospective	13
Wang Y; Journal of Rheumatology; 2005; 16265703189	40	0	40	52.3(13)	0	Yes	No	No	Yes	No	Yes	No	No	No	Longitudinal Prospective	11
Ding C; Arthritis & Rheumatism; 2005; 1632033980	325			45.2(6.5)	190	Yes	No	Yes	Yes	No	No	No	No	No	Longitudinal Prospective	10
Bruyere O; Annals of the Rheumatic Diseases; 2006; 16396978190	62	62	0	64.9(10.3)	49	Yes	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Longitudinal Prospective	10
Katz JN; Osteoarthritis & Cartilage; 2006; 1641321069	83			61(11), (Range: 45–89)	50(60%)	No	No	Yes	Yes	Yes	Yes	Yes	Yes	No	Longitudinal Prospective	9
Raynauld JP; Arthritis Research & Therapy; 2006; 1650711972	110	110	0	62.4(7.5)	(64%)	Yes	No	Yes	Yes	No	No	Yes	Yes	No	Longitudinal Prospective	11
Hunter DJ; Arthritis & Rheumatism; 2006; 1650893081	257	257	0	66.6(9.2), (Range: 47–93)	(41.6%)	No	No	Yes	Yes	No	No	No	Yes	No	Longitudinal Prospective	10
Ding C; Archives of Internal Medicine; 2006; 1656760582	325			Decrease defects: 45.4(6.4); Stable defects: 44.2(7.1); Increase defects: 46.1(5.9)	(58.1%)	No	No	Yes	Yes	No	No	No	No	No	Longitudinal Prospective	14
Brandt KD; Rheumatology; 2006; 16606655191	30	20	10	62	29	No	No	No	No	Yes	No	No	No	No	Other	10
Hunter DJ; Arthritis & Rheumatism; 2006; 1664603783	217	217	0	66.4(9.4)	(44%)	No	No	Yes	Yes	No	No	Yes	No	No	Longitudinal Prospective	10
Wluka AE; Arthritis Research & Therapy; 2006; 16704746192	105	105	0	All eligible: 62.5 (10.7); MRI at FU: 63.8(10.6); Lost to FU: 61.6(11.3)	59(53%)	Yes	No	No	Yes	No	Yes	No	No	No	Longitudinal Prospective	17
Hunter DJ; Osteoarthritis & Cartilage; 2007; 16857393193	127	127		67(9.05)	(46.7%)	No	Yes	No	Yes	No	No	No	No	No	Cross-sectional	12
Bruyere O; Osteoarthritis & Cartilage; 2007; 1689046173	62	62	0	64.9(10.3)	46	No	No	Yes	Yes	Yes	Yes	Yes	Yes	Yes	Longitudinal Prospective	10
Amin S; Annals of the Rheumatic Diseases; 2007; 17158140194	196	196	0	68(9)	0	No	No	Yes	Yes	No	No	No	No	No	Longitudinal Prospective	13
Nevitt MC; Arthritis & Rheumatism; 2007; 1746912674	80	39	0	73.5(3.1)	(63.6%)	No	No	Yes	Yes	No	No	No	No	No	Longitudinal Prospective	10
Hill CL; Annals of the Rheumatic Diseases; 2007; 1749109624	270	270	0	66.7(9.2)	112	No	No	Yes	Yes	Yes	No	No	No	No	Longitudinal Prospective	9
Pelletier JP; Arthritis Research & Therapy; 2007; 1767289171	110	110	0	Q1 greatest loss global: 63.7(7.2); Q4 least loss gobal: 61.3(7.5); Q1 greatest loss_medial: 64.1 (7.4); Q1 least loss_medial: 61.6(7.8)	(68.3%)	No	No	Yes	Yes	No	No	Yes	Yes	No	Longitudinal Prospective	15
Davies-Tuck ML; Osteoarthritis & Cartilage; 2008; 17698376195	117	117	0	63.7(10.2)	68(58%)	Yes	No	Yes	Yes	No	Yes	No	No	No	Longitudinal Prospective	14
Raynauld JP; Annals of the Rheumatic Diseases; 2008; 1772833384	107	107	0	62.4(7.5)	(64%)	Yes	No	Yes	Yes	No	No	No	Yes	No	Longitudinal Retrospective	15
Felson DT; Arthritis & Rheumatism; 2007; 1776342770	330	110	220	Cases: 62.9(8.3); Controls: 61.2(8.4)	211(63.9%)	No	No	Yes	No	No	Yes	Yes	No	No	Case control	12
Kornaat PR; European Radiology; 2007; 17823802196	182	71		59(Range: 43–76)	157(80%)	No	No	Yes	No	No	No	Yes	No	No	Longitudinal Prospective	8
Hunter DJ; Arthritis Research & Therapy; 2007; 17958892197	160	80	80	67(9)	(46%)	No	No	Yes	Yes	No	No	No	No	No	Case control	11
Englund M; Arthritis & Rheumatism; 2007; 1805020140	310	102	208	Cases: 62.9(8.3)	211(68.1%)	No	No	Yes	No	No	No	No	Yes	No	Case control	15
Davies-Tuck ML; Osteoarthritis & Cartilage; 2008; 1809384785	74	0	74	Meniscal tear: 58.8(6); No meniscal tear: 55.5(4.3)	74(100%)	No	Yes	Yes	Yes	No	Yes	No	Yes	No	Longitudinal Prospective	13
Hernandez-Molina G; Arthritis & Rheumatism; 2008; 18163483198	258	258	0	66.6(9.2)	(42.6%)	No	No	Yes	Yes	No	No	Yes	No	Yes	Longitudinal Prospective	11
Teichtahl AJ; Osteoarthritis & Cartilage; 2009; 18194873199	99	99	0	63 (10)	(60%)	Yes	No	No	Yes	No	Yes	No	No	No	Longitudinal Prospective	14
Amin S; Osteoarthritis & Cartilage; 2008; 1820362986	265	265		67(9)	(43%)	No	No	Yes	Yes	No	No	No	Yes	Yes	Longitudinal Prospective	11
Teichtahl AJ; Obesity; 2008; 18239654200	297		297	58(5.5)	186	Yes	No	Yes	Yes	No	Yes	No	No	No	Longitudinal Prospective	14
Blumenkrantz G; Osteoarthritis & Cartilage; 2008; 18337129201	18	8	10	Cases: 55.7(7.3); Controls: 57.6(6.2)	18(100%)	No	Yes	Yes	Yes	No	No	No	No	No	Case control	12
Song IH; Annals of the Rheumatic Diseases; 2009; 18375537202	41	41		65(6.7)	26	No	No	Yes	No	No	No	No	No	Yes	Randomized controlled trial	14
Scher C; Skeletal Radiology; 2008; 1846386567	65	65	0	OA-only: 49.3 (Range: 28–75); OA&BME group: 53.5(35–82)		No	No	Yes	Yes	No	No	Yes	No	No	Longitudinal Retrospective	10
Sharma L; Arthritis & Rheumatism; 2008; 1851277787	153	153	0	66.4(11)		Yes	No	Yes	Yes	No	No	No	Yes	No	Longitudinal Prospective	11
Owman H; Arthritis & Rheumatism; 2008; 18512778203	15	9	7	50(Range: 35–70)		No	Yes	No	Yes	No	No	No	No	No	Longitudinal Prospective	10
Madan-Sharma R; Skeletal Radiology; 2008; 1856681375	186	74	112	60.2(Range: 43–76)	150	No	No	Yes	Yes	No	No	Yes	Yes	No	Longitudinal Prospective	11
Amin S; Journal of Rheumatology; 2008; 18597397168	192	192		69(9)		No	No	Yes	Yes	No	No	No	No	No	Cross-sectional	10
Pelletier JP; Osteoarthritis & Cartilage; 2008; 1867238625	27	1		64.1(9.6)	14	No	No	Yes	Yes	Yes	Yes	Yes	Yes	No	Other	9
Amin S; Arthritis & Rheumatism; 2009; 19116936204	265	265	0	67(9)		No	No	Yes	Yes	No	No	No	No	No	Longitudinal Prospective	16

The analysis included data from 61 manuscripts of which 1 pertains to the hip and the remainder to the knee. The mean Downs criteria score for these manuscripts was 11.5 (range 6–17). What follows below are important excerpts from this data pertaining to different aspects of predictive validity. The data is further summarized in Table IV to discretely identify the associations examined and those where a significant association was found.

Table IV.

Summary of Predictive Validity of MRI in OA

Outcome of interest	Number of studies examining this outcome	Number of studies finding significant associations (P < .05)
Joint replacement	3 studies	3 of 3 (100%)
Change in symptoms	6 studies	5 of 6 (83%)
Radiographic progression	8 studies	5 of 8 (63%)
MRI progression	19 studies	16 of 19 (84%)

Prediction of joint replacement

Three studies examined the predictive relation of MRI findings to joint replacement. Of these, 100% demonstrated a statistically significant association, defined as P < 0.05.

• One study investigated the relation of change in quantitative cartilage volume to risk of knee replacement. For every 1% increase in the rate of tibial cartilage loss there was a 20% increase risk of undergoing a knee replacement at four years (95% CI, 10%–30%). Those in the highest tertile of tibial cartilage loss had 7.1 (1.4–36.5) higher odds of undergoing a knee replacement than those in the lowest tertile. Change in bone area also predicted risk of TKR OR 12 (95% CI 1–14)⁶⁵.

• Higher total cartilage defect scores (8–15) were associated with a 6.0-fold increased risk of joint replacement over 4 yr compared with those with lower scores (2–7) (95% CI 1.6, 22.3), independently of potential confounders⁶⁶.

• A separate smaller study investigated the relation of bone marrow lesions (assessed semi-quantitatively) to need for TKR. Subjects who had a bone marrow lesion were 8.95 times as likely to progress rapidly to a TKA when compared to subjects with no BME (P = 0.016). There was no relation of TKR with meniscal tear or cartilage loss⁶⁷.

Prediction of change in symptoms

Six studies examined the predictive relation of MRI findings to change in symptoms. Of these, 83% demonstrated a statistically significant association, defined as P < 0.05.

• Weak associations between worsening of symptoms of OA and increased cartilage loss: pain [r(s) = 0.28, P = 0.002], stiffness [r(s) = 0.17, P = 0.07], and deterioration in function [r(s) = 0.21, P = 0.02]²⁰.

• Small study did not find a significant relation between changes in WOMAC scores with the amount of cartilage loss and the change in BME (P > 0.05)⁶⁸.

• Multivariate analyses of knee pain 1 year following arthroscopic partial meniscectomy demonstrated that medial tibial cartilage damage accounting for 13% of the variability in pain scores⁶⁹.

• The BOKS study examined the relationship between longitudinal fluctuations in synovitis with change in pain and cartilage in knee osteoarthritis. Change in summary synovitis score was correlated with the change in pain (r = 0.21, P = 0.0003). An increase of one unit in summary synovitis score resulted in a 3.15-mm increase in VAS pain score (0–100 scale). Effusion change was not associated with pain change. Of the three locations for synovitis, changes in the infrapatellar fat pad were most strongly related to pain change²⁴.

• A nested case-control study examined if enlarging BMLs are associated with new knee pain. Case knee was defined as absence of knee pain at baseline but presence of knee pain both times at follow-up. Controls were selected randomly from among knees with absence of pain at baseline. Among case knees, 54 of 110 (49.1%) showed an increase in BML score within a compartment, whereas only 59 of 220 control knees (26.8%) showed an increase (P < 0.001 by chi-square test). A BML score increase of at least 2 units was much more common in case knees than in control knees (27.5% vs 8.6%; adjusted odds ratio 3.2, 95% CI 1.5–6.8)⁷⁰.

• Increases in Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain and patient global scores over time are associated with change in cartilage volume of the medial tibial plateau and medial femoral condyle⁷¹.

• Weak association of cartilage volume loss with less knee pain. Medial cartilage volume loss and simultaneous pain change at 24 months (beta coefficient −0.45, P = 0.03) and SF-36 physical components (beta coefficient 0.22, P = 0.04)⁷².

Prediction of radiographic progression

Eight studies examined the predictive relation of MRI findings to radiographic progression. Of these, 63% demonstrated a statistically significant association, defined as P < 0.05.

• No significant association between reduction in JSW and cartilage volume (R < 0.13). Trend toward a significant association between change in medial tibiofemoral cartilage volume and joint replacement at 4 years (OR = 9.0, P = 0.07) but not change in medial tibiofemoral JSW (OR = 1.1, P = 0.92)⁵⁰.

• No correlation between the cartilage volume loss changes (either by using absolute or percentage values) and the JSW changes at 24 months (global cartilage volume, r = 0.11; medial compartment cartilage volume, r = 0.19)⁷².

• Medial femorotibial JSN after 1 year, assessed by radiography, was significantly correlated with a loss of medial tibial cartilage volume (r = 0.25, P = 0.046) and medial tibial cartilage thickness (r = 0.28, P = 0.025), over the same period⁷³.

• Higher baseline composite cartilage scores and increases in composite cartilage scores during follow-up were moderately correlated with greater joint space loss (r = 0.33, P = 0.0002 and r = 0.26, P = 0.01, respectively)⁷⁴.

• Loss in JSW correlated with the loss of cartilage volume on the central weight-bearing area of the condyles and the plateaus as well as on the medial compartment⁷¹.

• Study examined the relation of MRI features at baseline with radiographically determined JSN in the medial compartment of the knee after 2 years in a group of patients with symptomatic osteoarthritis. A significant association was observed for meniscal tears (RR 3.57; CI 1.08–10.0) and meniscal subluxation (RR 2.73; CI 1.20–5.41), between KL < 2 and meniscal subluxation (RR 11.3; CI 2.49–29.49) and KL ≥ 2 and meniscus tears (RR 8.91; CI 1.13–22.84) and radiographic JSN 2 years later⁷⁵.

Prediction of MRI progression

Nineteen studies examined the predictive relation of MRI findings to MRI progression. Of these, 84% demonstrated a statistically significant association, defined as P < 0.05.

• Patients who had sustained meniscal tears showed a higher average rate of progression of cartilage loss (22%) than that seen in those who had intact menisci (14.9%) (P ≤ 0.018). Anterior cruciate ligament (ACL) tears had a borderline significant influence P ≤ 0.06) on the progression of cartilage pathology. Lesions located in the central region of the medial compartment were more likely to progress to more advanced cartilage pathology (progression rate 28%; P ≤ 0.003) than lesions in the anterior (19%; P ≤ 0.564) and posterior (17%; P ≤ 0.957) regions or lesions located in the lateral compartment (average progression rate 15%; P ≤ 0.707). Lesions located in the anterior region of the lateral compartment showed less progression of cartilage degradation (6%; P ≤ 0.001). No specific grade of lesion identified at baseline had a predilection for more rapid cartilage loss (P ≤ 0.93)⁷⁶.

• There was a significant correlation between the degree of loss of tibial cartilage and the degree of loss of femoral cartilage, in both tibiofemoral joints (r = 0.81, P < 0.001 at the medial tibiofemoral joint; r = 0.71, P < 0.001 at the lateral tibiofemoral joint)⁷⁷.

• A highly significant difference in global cartilage volume loss was observed between severe medial meniscal tear and absence of tear [mean (SD), −10.1 (2.1)% v −5.1 (2.4)%, P = 0.002]. An even greater difference was found between the medial meniscal changes and medial compartment cartilage volume loss [−14.3 (3.0)% in the presence of severe tear v −6.3 (2.7)% in the absence of tear; P < 0.0001]. Similarly, a major difference was found between the presence of a medial meniscal extrusion and loss of medial compartment cartilage volume [−15.4 (4.1)% in the presence of extrusion v −4.5 (1.7)% with no extrusion; P < 0.001]⁷⁸.

• Annual patellar cartilage loss was highest in those with defects compared with no defects (5.5% vs 3.2%, P = 0.01). Tibial cartilage loss was not associated with defects in the medial (4.6% vs 5.8%, P = 0.42) or lateral (4.7% vs 6.5%, P = 0.21) tibial cartilages⁶⁶.

• Baseline cartilage defect score was negatively associated with the progression of cartilage defects in each compartment (all P < 0.001)⁷⁹.

• Baseline cartilage defect scores at the medial tibia, lateral tibia, and patella had a dose-response association with the annual rate of change in knee cartilage volume at the corresponding site (beta = −1.3% to −1.2% per grade; P < 0.05 for all comparisons). In addition, an increase in knee cartilage defect score (change of more than or equal to 1) was associated with higher rates of knee cartilage volume loss at all sites (beta = −1.9% to −1.7% per year; P < 0.01 for all comparisons). Furthermore, a decrease in the knee cartilage defect score (change of less than or equal to −1) was associated with an increase in knee cartilage volume at all sites (beta = 1.0%–2.7% per year; P < 0.05 for all comparisons)⁸⁰.

• Predictors of fast progression included the presence of severe meniscal extrusion (P = 0.001), severe medial tear (P = 0.005), medial and/or lateral bone edema (P = 0.03), high body mass index (P < 0.05, fast vs slow), weight (P < 0.05, fast vs slow) and age (P < 0.05 fast vs slow)⁷².

• In the medial tibiofemoral joint, each measure of meniscal malposition was associated with an increased risk of cartilage loss. There was also a strong association between meniscal damage and cartilage loss⁸¹.

• A worsening in cartilage defect score was significantly associated with tibiofemoral osteophytes (OR, 6.22 and 6.04 per grade), tibial bone area (OR, 1.24 and 2.07 per square centimeter), and cartilage volume (OR, 2.91 and 1.71 per ml in the medial tibiofemoral and patellar compartments)⁸².

• Knee compartments with a higher baseline BML score had greater cartilage loss. An increase in BMLs was strongly associated with further worsening of the cartilage score⁸³.

• Despite cartilage loss occurring in over 50% of knees, synovitis was not associated with cartilage loss in either tibiofemoral or patellofemoral compartment²⁴.

• Significant correlations were seen between the loss of cartilage volume and edema size change in the medial condyle (−0.40, P = 0.0001) and the medial tibial plateau (−0.23, P = 0.03), and the changes in cyst size in the medial condyle (−0.29, P = 0.01). A multivariate analysis showed that the edema size change was strongly and independently associated with medial cartilage volume loss (−0.31, P = 0.0004)⁸⁴.

• Medial meniscal tear was associated with 103 mm(2) greater tibial plateau bone area within the medial (95% CI 6.2, 200.3; P = 0.04) and a lateral meniscal tear with a 120 mm(2) greater area within the lateral compartment (95% CI 45.5, 195.2; P = 0.002)⁸⁵.

• Adjusting for age, body mass index, gender and baseline cartilage scores, complete ACL tear increased the risk for cartilage loss at the medial tibiofemoral compartment (OR: 1.8, 95% CI: 1.1, 3.2). However, following adjustment for the presence of medial meniscal tears, no increased risk for cartilage loss was further seen (OR: 1.1, 95% CI: 0.6, 1.8)⁸⁶.

• Medial meniscal damage predicted medial tibial cartilage volume loss and tibial and femoral denuded bone increase, while varus malalignment predicted medial tibial cartilage volume and thickness loss and tibial and femoral denuded bone increase. Lateral meniscal damage predicted every lateral outcome⁸⁷.

• A positive correlation was found between the global severity of synovitis at baseline and the loss of cartilage volume at 60 days (P < 0.03)²⁵.

Discussion

The performance of MRI as an outcome measure in OA has been extensively studied providing strong support for both its concurrent and predictive validity.

As outlined in this review numerous studies have examined the relation of MRI to related constructs such as symptom measures, plain radiography, histology and arthroscopy. These studies demonstrate the following:

1. Inconsistent relation of structural features to symptoms with 13 of 21 studies finding a significant relation. Generally strong relation of large bone marrow lesions, moderate relation of synovitis and effusion and weak relation of cartilage volume/thickness to presence of pain. No relation of meniscal tears to presence of pain.

2. In general there was an inconsistent relation of cartilage volume and thickness and compositional measures to presence of radiographic OA. Higher frequency of meniscal tears, synovitis, increased bone area, increased bone attrition/curvature in persons with radiographic OA. Radiographic change insensitive to early changes found on MRI. 39 of 43 studies found significant associations between MRI and radiographic features.

3. There was a strong relation of meniscal subluxation and increased subchondral bone area to reduced radiographic joint space. Inconsistent (but generally moderate) relation of reduced cartilage volume and thickness to reduced radiographic joint space. Nine of nine studies found significant associations between MRI and radiographic joint space.

4. In general there was a strong correlation of cartilage volume measures to histologic findings. Three of five studies found significant relation of MRI to histology/pathology.

5. Moderate to strong relation of arthroscopic findings to cartilage and meniscal findings on MRI with five of seven studies finding a significant association

6. Strong relation of CT arthrography to MRI cartilage volume with all four studies examining this relation finding a significant association.

An important obstacle to biomarker validation and qualification is the adequate delineation of a gold standard. Unlike other diseases where surrogate endpoints exist, OA does not have a clear gold standard clinical endpoint and further is a remarkably heterogeneous disease. Therefore, the ‘clinical endpoint’ is more difficult to establish. A number of experts in the field have advocated that joint replacement be the clinical outcome of interest but due to constraints over comorbidities, insurance status and a number of other factors that influence determining if a person receives a joint replacement, alternate suggestions have been recommended including the use of virtual TKR (vTKR)⁸⁸. This is a composite endpoint that includes domains of pain, physical function and joint structure on X-rays⁸⁹. At this point it remains to be validated and as a consequence the constituent literature in this review does not include this endpoint to establish the predictive validity of MRI.

This work may be susceptible to publication bias as there was no effort made to search either clinical trial registries or meeting abstracts for potential unpublished studies that might tend to invalidate the MRI biomarkers examined.

The literature on the predictive validity of MRI in OA demonstrated the following:

1. Quantitative cartilage volume change and presence of cartilage defects or bone marrow lesions are potential predictors of TKR. Three of three studies found a significant relation.

2. Inconsistent but generally weak relation of cartilage loss to symptom change. Moderate relation of BML change to incident symptoms and pain change. Weak relation of change in synovitis to change in pain. Five of six studies found significant association between MRI and change in symptoms.

3. At best a weak relation between change in cartilage thickness and change in joint space. Five of eight studies found a significant relation.

4. Presence of meniscal damage, cartilage defects and BMLs predicts MRI progression. 16 of 19 studies found a significant relation.

Some MRI biomarkers correlate with some other biomarkers. Moreover in a limited number of studies some MRI biomarkers correlate with clinical endpoints and/or predict clinical outcomes Future research should be directed toward improving the predictive validity of current structural measures as they relate to important clinical outcomes so their role as surrogate outcomes can be substantiated. In addition, studies to improve the precision of assessment of structural features more closely related to symptom change such as BMLs and synovitis are warranted.