Article Text

Download PDFPDF

Extended report
Clinical significance of worsening versus stable preradiographic MRI lesions in a cohort study of persons at higher risk for knee osteoarthritis
  1. Leena Sharma1,
  2. Michael Nevitt2,
  3. Marc Hochberg3,
  4. Ali Guermazi4,
  5. Frank W Roemer4,5,
  6. Michel Crema4,
  7. Charles Eaton6,
  8. Rebecca Jackson7,
  9. Kent Kwoh8,
  10. Jane Cauley9,
  11. Orit Almagor1,
  12. Joan S Chmiel10
  1. 1Internal Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA
  2. 2Epidemiology and Biostatistics, University of California at San Francisco, San Francisco, California, USA
  3. 3School of Medicine, University of Maryland, Baltimore, Maryland, USA
  4. 4Department of Radiology, Boston University, Boston, Massachusetts, USA
  5. 5University of Erlangen-Nuremberg, Erlangen, Germany
  6. 6Brown University, Pawtucket, Rhode Island, USA
  7. 7The Ohio State University, Columbus, Ohio, USA
  8. 8Division of Rheumatology, University of Arizona Arthritis Center, University of Arizona, Tucson, Arizona, USA
  9. 9Department of Epidemiology, University of Pittsburgh, Pittsburgh, Pennsylvania, USA
  10. 10Preventive Medicine, Northwestern University Feinberg School of Medicine, Chicago, Illinois, USA
  1. Correspondence to Professor Leena Sharma, Division of Rheumatology, Feinberg School of Medicine, Northwestern University, 240 E. Huron, M300, Chicago, IL 60611, USA; L-Sharma{at}northwestern.edu

Abstract

Background Whether preradiographic lesions in knees at risk for osteoarthritis are incidental versus disease is unclear. We hypothesised, in persons without but at higher risk for knee osteoarthritis, that: 12–48 month MRI lesion status worsening is associated with 12–48 month incident radiographic osteoarthritis (objective component of clinical definition of knee osteoarthritis) and 48–84 month persistent symptoms.

Methods In 849 Osteoarthritis Initiative participants Kellgren/Lawrence (KL) 0 in both knees, we assessed cartilage damage, bone marrow lesions (BMLs), and menisci on 12 month (baseline) and 48 month MRIs. Multivariable logistic regression was used to evaluate associations between 12–48 month worsening versus stable status and outcome (12–48 month incident KL ≥1 and KL ≥2, and 48–84 month persistent symptoms defined as frequent symptoms or medication use most days of ≥1 month in past 12 month, at consecutive visits 48–84 months), adjusting for age, gender, body mass index (BMI), injury and surgery.

Results Mean age was 59.6 (8.8), BMI 26.7 (4.2) and 55.9% were women. 12–48 month status worsening of cartilage damage, meniscal tear, meniscal extrusion, and BMLs was associated with 12–48 month incident radiographic outcomes, and worsening of cartilage damage and BMLs with 48–84 month persistent symptoms. There was a dose-response association for magnitude of worsening of cartilage damage, meniscal tear, meniscal extrusion, and BMLs and radiographic outcomes, and cartilage damage and BMLs and persistent symptoms.

Conclusions In persons at higher risk, worsening MRI lesion status was associated with concurrent incident radiographic osteoarthritis and subsequent persistent symptoms. These findings suggest that such lesions represent early osteoarthritis, and add support for a paradigm shift towards investigation of intervention effectiveness at this stage.

  • Knee Osteoarthritis
  • Osteoarthritis
  • Epidemiology
  • Magnetic Resonance Imaging

Statistics from Altmetric.com

Introduction

Knowledge concerning early knee osteoarthritis (OA) is limited. By the time osteophytes are present radiographically (Kellgren/Lawrence (KL) grade ≥2, the definition of knee OA1–6), it is likely that the disease process has been underway for years. MRI lesions have been identified at KL 0 (normal) or 1 (possible/small osteophytes);7–21 however, few previous studies have evaluated their significance at these stages.17 ,18 Strongly associated with risk of incident KL ≥2,22 KL 1 is increasingly viewed as an early stage of OA. The significance of MRI lesions is likely to differ at each stage of OA; findings from studies including KL1 knees may or may not persist in knees that are normal by X-ray. We undertook MRI readings in a cohort of Osteoarthritis Initiative (OAI) participants who were KL 0 in both knees (since risk of knee OA is increased by contralateral knee OA) in an effort to better understand the significance of MRI lesions in a large segment of the high-risk population, those with normal knee X-rays. We initially found baseline cartilage damage, bone marrow lesions (BMLs), and meniscal tears were associated with symptoms over 4 years; baseline BMLs with incident patellofemoral (PF) cartilage damage; and a dose-response association for severity of cartilage damage and BMLs with symptoms, and BMLs with cartilage damage.21

OA appears to have stable/compensated (anabolic balancing catabolic activity) and progressive/decompensated phases. White et al23 illustrated that changing knee structure may be more consequential than stable structure: persons with incident (vs no incident) radiographic OA and worsening (vs stable) OA were at greater risk for incident severe function limitation. If MRI lesions represent early OA, an association, within the bilateral KL 0 cohort, between worsening (vs stable) lesion status and important clinical outcomes would be expected.

Previous studies evaluating the impact of changing lesions in knees KL <2 have, with rare exception,24 analysed the association between lesion change and other MRI change8 ,12 ,15 ,25 but not radiographic outcome, and none have been limited to KL 0 knees. Whether, in knees with normal X-rays, changing lesion status is associated with incident radiographic OA (the objective component of the clinical definition of knee OA) or persistent symptoms in a subsequent period is not known. Understanding whether preradiographic MRI lesions represent early disease will aid prevention and disease-modifying strategy design. If such lesions constitute early OA, they could become targets; such treatment, given before the downward spiral of tissue interactions of established knee OA, may be able to alter disease course.14 At present, there are no agents approved for the indication of OA disease modification.

We tested the hypotheses, in persons at higher risk of knee OA but KL 0 in both knees, that: 12–48-month MRI lesion status worsening is associated with greater risk of 12–48-month incident radiographic OA and 48–84-month persistent symptoms.

Methods

The OAI is a prospective, observational cohort study of persons with or at higher risk (on the basis of knee symptoms, overweight by gender/age-specific cut points, knee injury, knee surgery, family history of total knee replacement for OA in a biological parent or sibling, Heberden's nodes, repetitive knee bending, and/or age 70–79 years) for symptomatic, radiographic knee OA.26 Incidence subcohort eligibility required absent symptomatic, radiographic knee OA in either knee and present characteristics associated with increased risk of developing it27 ,28 Exclusion criteria were: inflammatory arthritis; severe bilateral joint space narrowing; total knee replacement and severe contralateral narrowing; bilateral total knee replacement or plan for it within 3 years; MRI or radiography contraindications; inability to provide blood sample; aides other than one straight cane for >50% of ambulation; severe comorbidity; double-blind trial participation.27

Current study inclusion additionally required KL 0 in both knees at the 12-month OAI visit, our ancillary study's baseline MRI assessment. The Institutional Review Board at each site approved the study.

Knee radiography

The posteroanterior fixed-flexion weight-bearing protocol27 ,29 with a SynaFlexer frame was used. Two experts, blinded to each other's reading, hypotheses and all other data, assessed KL grade.1 Weighted κ, between-reader agreement, was 0.79 for KL. Prespecified discrepancies (KL 1 vs 2) were adjudicated by consensus with a third reader.30

Knee MRI

Image acquisition used 3.0 T Siemens Trio scanners installed at each OAI site.31 Sequences included: coronal intermediate-weighted turbo spin echo, sagittal intermediate-weighted turbo spin echo with fat-suppression, and 3D Double Echo Steady State (DESS) water excitation (WE) sequence acquired in the sagittal plane with coronal and axial multiplanar reformat (MPR) reconstruction.31

We undertook 12-month and 48-month right (left, if right technically unacceptable) readings in persons determined by the coordinating centre to meet the KL criterion. Three experts who had not performed the X-ray readings (MC, AG and FWR) used modified MRI OA Knee Score (MOAKS),32 blinded to hypotheses, KL criterion and all other data. Evidence of their reliability using MOAKS to assess OAI images has been published, with very good to excellent inter-rater reliability for nearly all features.32 Paired images were read, with the chronology known.33 Cartilage morphology was scored separately for 4 PF and 10 tibiofemoral (TF) subregions: 0, normal; 1.0, 1–10% area damaged, no full thickness; 1.1, 1–10% area, 1–10% full thickness; 2.0, 10–75% area, no full thickness; 2.1, 10–75% area, 1–10% full thickness; 2.2, 10–75% area, 10–75% full thickness; 3.0, >75% area, no full thickness; 3.1, >75% area, 1–10% full thickness; 3.2, >75% area, 10–75% full thickness; 3.3, >75% area, >75% full thickness. BMLs were scored in the same subregions: 0, none; 1, <25% of subregions; 2, 25–50%; 3, >50%. For each meniscus, three subregions were scored: 0, normal; 1, signal abnormality; 2, horizontal tear; 3, vertical tear; 4, complex tear; 5, root tear; 6, partial maceration; 7, complete maceration; 8, progressing maceration. Extrusion was scored in each meniscus34: 0, none; 1, <50% extruded; 2, ≥50% extruded.

Covariate measurement

Covariates included age, gender, body mass index (BMI, body weight by standard balance beam scale, height by wall mounted stadiometer), knee injury (ever injured (study knee) badly enough to limit walking ≥2 days) and knee surgery (any previous, study knee).

Statistical analysis

All analyses used one knee per person. Multivariable logistic regression was used to evaluate associations between worsening lesion status (12–48-month incident lesion or worsening of established lesion) versus stable lesion status and each outcome (12–48-month incident KL ≥1 and incident KL ≥2, and 48–84-month persistent symptoms defined as frequent knee symptoms or medication use for them more than half the days in at least one of the past 12 months, at two consecutive annual visits between 48 months and 84 months), adjusting for age, gender, BMI, knee injury and knee surgery. Results are summarised using adjusted ORs and 95% CIs.

For cartilage damage, BMLs and meniscal extrusion, incident lesion was defined by 48-month occurrence of a score >0 in ≥one subregion among knees with all 12-month subregions 0 for that lesion, and worsening of established lesion was defined by 12–48-month worsening in ≥one subregion among knees with a 12-month score >0 for that lesion in ≥one subregion. For meniscal tear, incident lesion was defined by 48-month occurrence of a score ≥2 in ≥one subregion among knees with all 12-month subregions <2, and worsening of established lesion by 12–48 month worsening in ≥one subregion among knees with a 12-month score ≥2 in ≥one subregion.

We also defined an ordered categorical variable for number of lesion types with 12–48-month worsening status and used logistic regression to analyse associations of this variable (to assess trends across ordered categories) with each outcome. We explored a dose-response association between greater magnitude of worsening (using TF region with greatest worsening, for KL outcomes, and TF or PF region with greatest worsening, for persistent symptoms outcome) and each outcome. For ‘dose’ of worsening area of damage, we recoded cartilage damage severity: 0=normal; 1=1.0 or 1.1; 2=2.0, 2.1 or 2.2; and 3=3.0, 3.1, 3.2 or 3.3. For ‘dose’ of worsening full thickness involvement, we recoded cartilage damage severity: 0=normal; 1=1.0, 2.0 or 3.0; 2=1.1, 2.1 or 3.1; 3=2.2 or 3.2; and 4=3.3. For ‘dose’ of worsening meniscal lesions, we recoded meniscal lesions: 0=0 or 1; 1=2 or 3; 2=4 or 6; and 3=7 or 8. SAS software V.9.3 (SAS Institute, Cary, North Carolina, USA) was used.

Results

Figure 1 delineates sample derivation. Among 1114 eligible, 77 missed the 48-month evaluation (withdrew (28), difficulty contacting/scheduling (32), died (14), health problems (2), caregiving responsibilities (1)), 176 attended without MRI, and 12 had inadequate images. We assessed right knee MRIs (left, if right inadequate) at 12 months and 48 months in the remaining 849 (mean age 59.6 years (SD 8.8), BMI 26.7 kg/m2 (4.2), 475 (55.9%) women; 88.7% Caucasian, 9.0% African-American, 1.1% Asian, 1.2% other non-white; 8 (0.9%) Hispanic, including 2 Caucasian, 1 African-American, 5 other non-white). Persons KL 0 in both knees and not in the analysis sample were similar (age 58.7 years (9.3), BMI 27.0 (4.8), 63.0% women). An additional 57 were not included in radiographic analyses due to missing radiographic or covariable data; their characteristics were similar (mean age 58.3 years (7.8), BMI 26.9 (3.6), 63.2% women, 11.8% non-white). Of the 792 knees from 792 persons in the radiographic analysis sample, 50 developed incident KL ≥1 and 25 incident KL ≥2 by 48 months. Of the 841 knees from 841 persons, 200 had persistent symptoms in the 48–84-month period.

Figure 1

Derivation of analysis samples. The figure shows the derivation of analysis samples. KL, Kellgren and Lawrence.

Tables 1 and 2 show outcome frequency by 12–48-month lesion status groups: none→none, none→incident lesion, lesion→no worsening and lesion→worse. As shown in table 1, status worsening (vs no worsening) in TF cartilage damage, PF cartilage damage, meniscal tear, meniscal extrusion and TF BMLs was associated with incident KL ≥1 and incident KL ≥2. In post hoc analyses further adjusting for TF cartilage damage worsening, PF cartilage damage worsening was associated with incident KL ≥1 (adjusted OR 2.28, 95% CI 1.18 to 4.39) but not incident KL ≥2 (adjusted OR 2.16, 95% CI 0.88 to 5.32); in these models, TF cartilage damage worsening was associated with both outcomes (adjusted OR 2.49, 95% CI 1.38 to 4.51, and adjusted OR 8.44, 95% CI 3.10 to 23.03, respectively). As shown in table 2, status worsening in TF cartilage damage, PF cartilage damage, TF BMLs and PF BMLs was associated with 48–84-month persistent symptoms (findings persisting after further adjustment for 12–48-month symptoms, not shown). In sensitivity analyses further adjusting for comorbidity (ie, having two or more comorbid conditions) as well as side of leg dominance, these covariables were not consistently associated with the outcomes and the OR estimates for the associations between lesion worsening and outcomes were minimally affected.

Table 1

12-month to 48-month MRI lesion status worsening and 12–48-month incident KL ≥1 and incident KL ≥2 radiographic outcomes

Table 2

12-month to 48-month MRI lesion status worsening and 48–84-month persistent knee symptoms

Cartilage damage status worsened in TF and PF compartments in 49 knees; combined worsening (vs all other possibilities) was associated with persistent symptoms (adjusted OR 2.18, 95% CI 1.19 to 3.99). BML status worsened in both compartments in 21 knees; combined worsening was associated with persistent symptoms (adjusted OR 3.65, 95% CI 1.50 to 8.86). As shown in table 3, there was a significant linear increase (p for trend) in the odds of each outcome (on the logit scale) as number of lesion types with worsening increased.

Table 3

Number of lesion types with 12–48-month status worsening and 12–48-month incident KL ≥1, 12–48-month incident KL ≥2, and 48–84-month persistent knee symptoms

As shown in table 4, greater 12–48-month worsening of TF cartilage damage (surface area and extent of full thickness involvement), meniscal tear, meniscal extrusion and TF BML scores was associated with 12–48-month incident radiographic KL ≥1 and KL ≥2; greater 12–48-month worsening of TF or PF cartilage damage (surface area and full thickness involvement) and TF or PF BMLs was associated with 48–84-month persistent symptoms.

Table 4

Greater magnitude of 12–48-month worsening of each lesion type: association with 12–48-month incident KL ≥1, 12–48-month incident KL ≥2 and 48–84-month persistent knee symptoms

In analyses using an alternative definition of worsening (subregion maximum score worse at 48 months than 12 months among knees with that lesion at 12 months), findings were similar. Twelve-month to 48-month radiographic medial and lateral TF joint space grade progression occurred in 10 knees and 5 knees, respectively, precluding examination of TF compartment-specific radiographic outcome.

Discussion

In persons at higher risk of knee OA but KL 0 in both knees, 12–48-month status worsening in TF cartilage damage, meniscal tear, meniscal extrusion and TF BMLs by MRI was associated with 12–48-month incident radiographic KL ≥1 and incident KL ≥2. Twelve-month to 48-month status worsening in TF cartilage damage, PF cartilage damage, TF BMLs and PF BMLs was associated with 48–84-month persistent symptoms. More lesion types with worsening were associated with greater odds of each outcome. There was a dose-response association for magnitude of worsening of TF cartilage damage, meniscal tear, meniscal extrusion and TF BML score and each radiographic outcome, and for magnitude of worsening of TF or PF cartilage damage and TF or PF BMLs and the persistent symptom outcome.

Conceptually, the study of early OA is compelling. As stated by Ding, preradiographic MRI lesions “allow us to explore the very subtle, potentially reversible, early aberrations in joint morphology, prior to the onset of radiographic disease… we can identify reversible structural changes and risk factors for these, to enable us to turn the individual off the pathway to OA and avert progressive change”.14 A recent symposium dealt with strategies to transform the approach to OA from palliation to prevention.35 The need to better understand the ‘tilting point’ at which initial anabolic events shift to net catabolic activity in early OA was described.36 McGonagle et al37 proposed knee OA classification based on site(s) of earliest joint involvement, noting that treatment for early disease could theoretically target specific tissues to delay whole-organ involvement. Luyten et al38 suggested patient categories that fit and would aid work within regenerative medicine and tissue engineering: MRI tissue lesions in otherwise healthy joints; tissue lesions with early radiographic signs; established OA by American College of Rheumatology criteria. As stimulating as these ideas are, whether preradiographic lesions are important or incidental is unclear. Advancing this knowledge clarifies whether they can serve as targets or as parameters to define knees requiring early treatment.

MRI lesions were not infrequent in knees without radiographic OA in the Framingham study: TF cartilage damage in 69%, TF BMLs in 52%, meniscal tears in 25%.11 ,20 Rates in persons at higher risk to develop knee OA are not substantially higher: in the Multicenter Osteoarthritis Study, TF or PF cartilage damage in 67–81%, BMLs in 55–74%;13 and, in our recent report regarding the current sample, cartilage damage in 76%, BMLs in 61%, meniscal tears in 21%, meniscal extrusion in 14%.21 Different findings may reflect different MRI methods and inclusion of persons with contralateral knee OA.

To our knowledge, no prior study has examined the impact of worsening MRI lesions in persons with normal knee X-rays. In knees without OA by American College of Rheumatology clinical criteria or knee pain, incident BML development (but not change in prevalent BMLs) was associated with pain development.12 In other cohorts largely without radiographic OA (KL <2), cartilage defect change was associated with cartilage volume change.8 ,25 In persons without radiographic OA (KL <2), BML size change was associated with concurrent pain and function change.15 In a case-control study involving OAI participants with KL 0 or 1 in at least one knee, Roemer et al24 recently reported that the presence of specific MRI features 2 years prior to incident radiographic OA was associated with risk of incident OA, and incident lesions were associated with greater risk (than fluctuating or prevalent lesions).

We previously reported that, within the bilateral KL 0 sample at higher risk, lesions at 12 months were associated with 48-month incident MRI cartilage damage and persistent symptoms by 60 months.21 The current study extends this work by showing that, within this sample, worsening (vs stable) lesion status is associated with incident radiographic OA over the same time interval and persistent symptoms in the subsequent 3 years. The pattern of findings was similar for KL ≥1 and KL ≥2 outcomes (although we had lower power for the latter), in keeping with a report that KL 1 is associated with future radiographic knee OA (KL ≥2).22 As anticipated, TF lesion status change was associated with KL outcomes. The association between PF cartilage damage status worsening and KL outcomes was at least partially explained by TF status worsening. Notably, PF lesion status change was important for the persistent symptom outcome.

This study has limitations. The OAI focuses on persons at higher risk for knee OA, a group of public health importance that will grow with expansion of the aging and overweight populations. However, our findings may or may not apply in a population not at higher risk for knee OA. Our MRI readings did not include effusion/synovitis, another potentially important factor in early OA.24 Improvement in lesion status is possible, particularly with BMLs. In our analyses, BML improvement was grouped with ‘no worse’. It would have been interesting to further subdivide knees to examine whether BML improvement was associated with a reduced risk of outcomes, but we did not have sufficient power for this. MOAKS cartilage assessment ranges correspond to focal, extensive and diffuse involvement; the middle group has wide bounds (10–75%). However, MOAKS is an established, and among the most widely used, reading systems for knee OA, and affords the important opportunity to assess surface area of involvement and extent of full thickness involvement. It was not among our objectives to evaluate the clinical utility of MRI in this population; the role of MRI in the care of patients at this stage is, at present, uncertain.

That, within this sample worsening lesion status was associated with the concomitant new development of radiographic knee OA as well as persistent symptoms in the subsequent 3 years, along with robust dose-response findings, adds to the evidence that these lesions represent early OA and are clinically significant. Given the absence of disease-modifying therapy for OA, widespread clinical application of MRI is difficult to justify. Nevertheless, the findings have important implications. Prevention or delay of worsening of early-stage lesions should be considered as a target for emerging pharmacological and non-pharmacological treatments in an effort to prevent or delay full-blown disease. Candidate interventions should be studied at this stage, when they are more likely to be effective, potentially even in a shorter duration or intermittently applied. The findings support the recent call to no longer consider OA as a disorder of hyaline cartilage,39 and additionally support a paradigm shift towards investigation of tissue-targeted intervention effects at this early stage.

In conclusion, in persons at higher risk but with normal knee X-rays at baseline, worsening MRI lesion status over a 3-year interval was associated with concurrent development of radiographic OA and persistent symptoms over a subsequent 3-year follow-up. These findings support the concept of compensated/stable and decompensated/progressive phases at very early stages of OA and suggest that such lesions represent early disease and illness.

Acknowledgments

The authors acknowledge the dedication and commitment of the OAI study participants.

References

Footnotes

  • Handling editor Tore K Kvien

  • Correction notice This article has been corrected since it was published Online First. Changes have been made to the author affiliations and the shading in tables 1 and 2 has been corrected.

  • Contributors All authors contributed to the planning, conduct, and reporting of the work described and fulfil the ICMJE recommendations for authorship credit: Substantial contributions to the conception or design of the work, or the acquisition, analysis or interpretation of data. Drafting the work or revising it critically for important intellectual content. Final approval of the version published. Agreement to be accountable for all aspects of the work in ensuring that questions related to the accuracy or integrity of any part of the work are appropriately investigated and resolved.

  • Funding NIH/NIAMS R01 AR52918, R01 AR065473, P60 AR064464, and the Osteoarthritis Initiative. The OAI is a public-private partnership comprised of five contracts (N01-AR-2-2258; N01-AR-2-2259; N01-AR-2-2260; N01-AR-2-2261; N01-AR-2-2262) funded by the NIH and conducted by the OAI Study Investigators. Private funding partners include Merck Research Laboratories, Novartis Pharmaceuticals Corporation, GlaxoSmithKline, and Pfizer, Private sector funding for the OAI is managed by the Foundation for the NIH. This manuscript was prepared using an OAI public use data set (in addition to data obtained within NIH/NIAMS funded ancillary grants) and does not necessarily reflect the opinions or views of the OAI investigators, the NIH, or the private funding partners. The study sponsor played no role in study design or the collection, analysis, and interpretation of data and the writing of the article or the decision to submit it for publication; researchers were independent from funders and sponsors; researchers had access to all the data.

  • Competing interests None declared.

  • Provenance and peer review Not commissioned; externally peer reviewed.

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.