Advertisement

Potential surrogate outcomes in individuals at high risk for incident knee osteoarthritis

Open AccessPublished:January 13, 2023DOI:https://doi.org/10.1016/j.joca.2023.01.003

      Summary

      Objective

      To study potential surrogate outcomes for osteoarthritis (OA) incidence by evaluating the association of short-term changes in clinical and imaging biomarkers with long-term clinical knee OA incidence.

      Design

      Middle-aged women with overweight/obesity, but free of knee symptoms were recruited through their general practitioners. At baseline, after 2.5 years, and after 6.5 years, questionnaires, physical examination, radiographs, and Magnetic resonance imaging (MRI) scans were obtained. The percentage of knees with a minimal clinically important difference for knee pain severity, Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) pain/stiffness/function, and joint space narrowing, and of those with progression/regression of medial knee alignment, chronic knee pain, radiographic osteophytes, and cartilage defects, bone marrow lesions, osteophytes, and effusion/synovitis on MRI were determined. For each of these potential surrogate outcomes with ≥10% improvement or progression in the population over 2.5 years, the association with incident clinical knee OA, defined using the combined ACR-criteria, after 6.5 years was determined.

      Results

      Most pre-defined potential surrogate outcomes showed ≥10% change in the population over 2.5 years, but only worsening of TF cartilage defects, worsening of TF osteophytes on MRI, and an increase in pain severity were significantly associated with greater clinical knee OA incidence after 6.5 years. These potential surrogate outcomes had high specificity and negative predictive value (89–91%) and low sensitivity and positive predictive value (20–28%)

      Conclusions

      Worsening of TF cartilage defects and TF osteophytes on MRI, and increased pain severity could be seen as surrogate outcomes for long-term OA incidence. However, higher positive predictive values seem warranted for the applicability of these factors in future preventive trials.

      Keywords

      Introduction

      The worldwide prevalence of osteoarthritis (OA) and the associated social and healthcare costs are very high
      • Hunter D.J.
      • Bierma-Zeinstra S.
      Osteoarthritis.
      • Katz J.N.
      • Arant K.R.
      • Loeser R.F.
      Diagnosis and treatment of hip and knee osteoarthritis: a review.
      • Leyland K.M.
      • Gates L.S.
      • Sanchez-Santos M.T.
      • Nevitt M.C.
      • Felson D.
      • Jones G.
      • et al.
      Committee PS
      Knee osteoarthritis and time-to all-cause mortality in six community-based cohorts: an international meta-analysis of individual participant-level data.
      . Without a drastic change in the way we handle OA, the number of OA patients and the burden of OA will become unmanageable in the near future. Experts in the field of OA have suggested that a shift in research focus towards primary prevention is needed
      • Roos E.M.
      • Arden N.K.
      Strategies for the prevention of knee osteoarthritis.
      ,
      • Whittaker J.L.
      • Runhaar J.
      • Bierma-Zeinstra M.A.
      • Roos E.M.
      A lifespan approach to osteoarthritis prevention.
      .
      A key challenge for studies in OA prevention is its slowly developing nature. Using traditional outcomes of OA incidence, e.g., Kellgren & Lawrence (KL) grade ≥2 for radiographic knee OA
      • Kellgren J.H.
      • Lawrence J.S.
      Radiological assessment of osteo-arthrosis.
      or the ACR-criteria for clinical knee OA
      • Altman R.
      • Asch E.
      • Bloch D.
      • Bole G.
      • Borenstein D.
      • Brandt K.
      • et al.
      Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association.
      , annual incidence rates within high-risk subjects, e.g., overweight/obese individuals or those sustained a serious knee injury, are only ±2–5%
      • de Vos B.C.
      • Landsmeer M.L.A.
      • van Middelkoop M.
      • Oei E.H.G.
      • Krul M.
      • Bierma-Zeinstra S.M.A.
      • et al.
      Long-term effects of a lifestyle intervention and oral glucosamine sulphate in primary care on incident knee OA in overweight women.
      • Lo G.H.
      • Strayhorn M.T.
      • Driban J.B.
      • Price L.L.
      • Eaton C.B.
      • McAlindon T.E.
      Subjective crepitus as a risk factor for incident symptomatic knee osteoarthritis: data from the osteoarthritis initiative.
      • Lohmander L.S.
      • Englund P.M.
      • Dahl L.L.
      • Roos E.M.
      The long-term consequence of anterior cruciate ligament and meniscus injuries: osteoarthritis.
      • Runhaar J.
      • van Middelkoop M.
      • Reijman M.
      • Willemsen S.
      • Oei E.H.
      • Vroegindeweij D.
      • et al.
      Prevention of knee osteoarthritis in overweight females: the first preventive randomized controlled trial in osteoarthritis.
      • Schiphof D.
      • Runhaar J.
      • Waarsing J.H.
      • van Spil W.E.
      • van Middelkoop M.
      • Bierma-Zeinstra S.M.A.
      The clinical and radiographic course of early knee and hip osteoarthritis over 10 years in CHECK (Cohort Hip and Cohort Knee).
      . Therefore, although advocated as a plausible strategy for OA prevention, clinical trials evaluating the incidence of knee OA in these high-risk groups require large samples and a very long follow-up
      • Whittaker J.L.
      • Runhaar J.
      • Bierma-Zeinstra M.A.
      • Roos E.M.
      A lifespan approach to osteoarthritis prevention.
      ,
      • Runhaar J.
      • Zhang Y.
      Can we prevent OA? Epidemiology and public health insights and implications.
      . Surrogate outcomes for the incidence of knee OA can boost the research on preventive treatments
      • Lohmander L.S.
      • Eyre D.R.
      From biomarker to surrogate outcome to osteoarthritis--what are the challenges?.
      .
      A surrogate outcome can lead to a “reduction in sample size and trial duration when a […] distant disease is replaced by a more frequent or proximal endpoint”
      • Prentice R.L.
      Surrogate endpoints in clinical trials: definition and operational criteria.
      . This could potentially “reduce costs and enhances feasibility” of clinical trials
      • Prentice R.L.
      Surrogate endpoints in clinical trials: definition and operational criteria.
      • Ciani O.
      • Buyse M.
      • Drummond M.
      • Rasi G.
      • Saad E.D.
      • Taylor R.S.
      Time to review the role of surrogate end points in health policy: state of the art and the way forward.
      • Vanderweele T.J.
      Surrogate measures and consistent surrogates.
      . To illustrate the potential, for the evaluation of treatment effects of certain medications in other chronic diseases, surrogate outcomes have successfully substituted traditional primary outcomes that have a direct and serious impact on patients (e.g., cardiovascular events or mortality)
      • Gandhi G.Y.
      • Murad M.H.
      • Fujiyoshi A.
      • Mullan R.J.
      • Flynn D.N.
      • Elamin M.B.
      • et al.
      Patient-important outcomes in registered diabetes trials.
      . For instance, in 436 registered RCTs on diabetes drugs, most trials (82%) used surrogate or laboratory markers as primary outcome rather than ‘patient-important’ outcomes, such as death or quality of life measures
      • Gandhi G.Y.
      • Murad M.H.
      • Fujiyoshi A.
      • Mullan R.J.
      • Flynn D.N.
      • Elamin M.B.
      • et al.
      Patient-important outcomes in registered diabetes trials.
      .
      Potential surrogate outcomes can be validated using a stepwise approach
      • Ciani O.
      • Buyse M.
      • Drummond M.
      • Rasi G.
      • Saad E.D.
      • Taylor R.S.
      Time to review the role of surrogate end points in health policy: state of the art and the way forward.
      , which is also advocated by the European Medicines Agency (EMA)
      • Agency EM
      ICH Topic E 9 - Statistical Principles for Clinical Trials.
      . Biological plausibility of the relation between the surrogate outcome and the true outcome is the lowest form of evidence (level 3). Evidence is considered level 2 if “consistent associations between the surrogate and final outcome” from observational studies exists. Results from multiple RCTs that show that the changes in the surrogate outcome are associated with commensurate changes in the true outcome are seen as level 1 evidence
      • Ciani O.
      • Buyse M.
      • Drummond M.
      • Rasi G.
      • Saad E.D.
      • Taylor R.S.
      Time to review the role of surrogate end points in health policy: state of the art and the way forward.
      .
      The objective of the current study was to study potential surrogate outcomes for OA incidence by evaluating the association of short-term changes (2.5 years) in several clinical and imaging biomarkers with long-term (6.5 years) clinical knee OA incidence, among high-risk middle-aged women with overweight and obesity, free of knee OA at baseline. Given their sex, age and body weight status, this population should be regarded as a high-risk population for future knee OA incidence and with that form an excellent target population for future preventive trials.

      Methods

      Data of the PRevention of knee Osteoarthritis in Overweight Females (PROOF) study were used
      • Runhaar J.
      • van Middelkoop M.
      • Reijman M.
      • Willemsen S.
      • Oei E.H.
      • Vroegindeweij D.
      • et al.
      Prevention of knee osteoarthritis in overweight females: the first preventive randomized controlled trial in osteoarthritis.
      . In PROOF, 407 women (eligibility criteria: aged 50–60 years, body mass index (BMI) ≥27 kg/m2, and free of knee symptoms)
      • Altman R.
      • Asch E.
      • Bloch D.
      • Bole G.
      • Borenstein D.
      • Brandt K.
      • et al.
      Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association.
      were recruited (between 2006 and 2009) through 50 general practices in the greater Rotterdam region and followed for 6.5 years. Originally, PROOF was an intervention study evaluating the preventive effects of a lifestyle intervention and glucosamine sulfate on OA incidence after 2.5 and 6.5 years, but due to the absence of any intervention effects the data is considered as a cohort
      • de Vos B.C.
      • Landsmeer M.L.A.
      • van Middelkoop M.
      • Oei E.H.G.
      • Krul M.
      • Bierma-Zeinstra S.M.A.
      • et al.
      Long-term effects of a lifestyle intervention and oral glucosamine sulphate in primary care on incident knee OA in overweight women.
      ,
      • Runhaar J.
      • van Middelkoop M.
      • Reijman M.
      • Willemsen S.
      • Oei E.H.
      • Vroegindeweij D.
      • et al.
      Prevention of knee osteoarthritis in overweight females: the first preventive randomized controlled trial in osteoarthritis.
      . All participants gave written informed consent prior to any measurements and the procedures followed were in accordance with the ethical standards of the responsible committee on human experimentation (institutional and national) and with the Helsinki Declaration of 1975, as revised in 2000. The PROOF study was approved by the Medical Ethical Committee of Erasmus MC University Medical Center.
      At baseline, 2.5 years, and 6.5 years, questionnaires, physical examination, standardized semi-flexed posterior-anterior (PA) radiographs, and multi-sequential MRI scans were obtained. The questionnaires included questions on demographics (i.e., age, menopausal status, history of knee injury), knee symptoms (i.e., presence of chronic pain, defined as ‘pain on most days of the last month’, pain severity on a 0–10 numeric rating scale, WOMAC scores for subscales of pain, functional limitations, and stiffness on a 0–100 scale, and the presence of knee joint stiffness <30 min). From the radiographs, osteophytes on the medial and lateral femur and tibia were scores on a 0–3 scale, minimal joint space width was measured in mm, and the medial knee alignment was measured in degrees and stratified into varus, neutral and valgus aligned knees
      • Runhaar J.
      • van Middelkoop M.
      • Reijman M.
      • Willemsen S.
      • Oei E.H.
      • Vroegindeweij D.
      • et al.
      Prevention of knee osteoarthritis in overweight females: the first preventive randomized controlled trial in osteoarthritis.
      . MRIs at baseline and 2.5 years MRIs were evaluated for the presence and severity of osteophytes, cartilage defects, and bone marrow lesions (for the patellofemoral (PF) and tibiofemoral (TF) compartments separately), meniscal pathologies, meniscal extrusion, and effusion/synovitis according to the MRI Osteoarthritis Knee Score (MOAKS)
      • Hunter D.J.
      • Guermazi A.
      • Lo G.H.
      • Grainger A.J.
      • Conaghan P.G.
      • Boudreau R.M.
      • et al.
      Evolution of semi-quantitative whole joint assessment of knee OA: MOAKS (MRI Osteoarthritis Knee Score).
      .

      Potential surrogate outcomes

      For the present study, validated patient-reported clinical outcomes (e.g., WOMAC and visual analogue scale (VAS) pain scores), structural outcomes on radiography [e.g., joint space narrowing (JSN) and osteophytes], and structural items on MRI (e.g., cartilage defects and meniscal pathologies) were selected for evaluation. Published measures of the Minimal Clinically Important Difference (MCID) were used to determine the number of knees/subjects with changes over 2.5 years, based on knee pain severity (≥1 point change
      • Salaffi F.
      • Stancati A.
      • Silvestri C.A.
      • Ciapetti A.
      • Grassi W.
      Minimal clinically important changes in chronic musculoskeletal pain intensity measured on a numerical rating scale.
      ), WOMAC pain (≥8.3 points change
      • Angst F.
      • Aeschlimann A.
      • Michel B.A.
      • Stucki G.
      Minimal clinically important rehabilitation effects in patients with osteoarthritis of the lower extremities.
      ), WOMAC stiffness (≥8.0 points change
      • Angst F.
      • Aeschlimann A.
      • Michel B.A.
      • Stucki G.
      Minimal clinically important rehabilitation effects in patients with osteoarthritis of the lower extremities.
      ), WOMAC function (≥10.1 points change
      • Angst F.
      • Aeschlimann A.
      • Michel B.A.
      • Stucki G.
      Minimal clinically important rehabilitation effects in patients with osteoarthritis of the lower extremities.
      ), and medial and lateral JSN (≥30% change
      • Ornetti P.
      • Brandt K.
      • Hellio-Le Graverand M.P.
      • Hochberg M.
      • Hunter D.J.
      • Kloppenburg M.
      • et al.
      OARSI-OMERACT definition of relevant radiological progression in hip/knee osteoarthritis.
      ). For changes in medial knee alignment, no MCID was available. Hence, we used one unit (1°) to define its change; ≥1° more varus in varus/neutral aligned knees or more valgus in valgus/neutral aligned knees was defined as ‘worsening’, while ≥1° less varus in varus aligned knees or less valgus in valgus aligned knees was defined as ‘improvement’. Incident chronic knee pain was defined as no chronic knee pain at baseline and presence of chronic knee pain at 2.5 years, improvement was defined as chronic knee pain at baseline and no chronic knee pain at 2.5 years, and when chronic knee pain status was identical for baseline and at 2.5 years, knees were labeled as ‘no change’. For osteophytes scored on radiographs, progression ≥1 grade in ≥1 anatomical location was used to determine the 2.5 years progression (excluding progression from 0 to 1 to match the incidence or progression of definite osteophytes). For MOAKS features, published definitions by Runhaar et al.
      • Runhaar J.
      • Schiphof D.
      • van Meer B.
      • Reijman M.
      • Bierma-Zeinstra S.M.
      • Oei E.H.
      How to define subregional osteoarthritis progression using semi-quantitative MRI osteoarthritis knee score (MOAKS).
      were used to define progression per compartment (TF/PF), given the known clinical importance of the PF compartment in OA incidence
      • Macri E.M.
      • van Middelkoop M.
      • Damen J.
      • Bos P.K.
      • Bierma-Zeinstra S.M.
      Higher risk of knee arthroplasty during ten-year follow-up if baseline radiographic osteoarthritis involves the patellofemoral joint: a CHECK Cohort Study.
      . Additionally, any change ≥1 grade in either ‘effusion/synovitis’ or ‘Hoffa synovitis’ was defined as a change in effusion/synovitis
      • Banuls-Mirete M.
      • Lombardi A.F.
      • Posis A.I.B.
      • Shadyab A.H.
      • Chang E.Y.
      • Lane N.E.
      • et al.
      Effusion-synovitis worsening mediates the association between body mass index and Kellgren-Lawrence progression in obese individuals: data from the Osteoarthritis Initiative.
      .

      Long-term OA incidence

      Knee OA incidence after 6.5 years was defined by using the combined ACR-criteria
      • Altman R.
      • Asch E.
      • Bloch D.
      • Bole G.
      • Borenstein D.
      • Brandt K.
      • et al.
      Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association.
      ; presence of chronic knee pain, ≥1 definite osteophyte on radiography (grade ≥2), and one or more of the following: age >50 years, stiffness <30 min, or crepitus.

      Statistical analysis

      All knees free of clinical knee OA according to the combined ACR-criteria at baseline, with all potential surrogate outcomes at 2.5 years and the long-term OA incidence measures available were selected for the analyses. Baseline characteristics were determined and compared between included and excluded knees.
      First, for each of the potential surrogate outcomes, the proportion of knees/subjects showing improvement, no change, and progression over 2.5 years was determined, using MCID measures described above. Those potential surrogate outcomes with ≥10% improvement or with ≥10% progression in the study sample over 2.5 years were selected to assess the association of the change in the potential surrogate outcome (improvement or progression) with clinical knee OA incidence after 6.5 years. Three-level surrogate outcomes were restructured into two-level outcomes if one level of the outcome (improvement or progression) did not reach the 10% threshold, by merging the specific level with the ‘no change’ level. For potential surrogate outcomes measured on a knee-level (i.e., knee pain severity, medial and lateral JSN, medial knee alignment, osteophytes on radiography, and all MOAKS features), the association of the change in the potential surrogate outcome with clinical OA incidence in a knee (yes/no) was determined using Generalized Estimating Equations to account for repeated measures within persons. For potential surrogate outcomes measured on a subject-level (i.e., all WOMAC scores), a simple logistic regression was used where clinical OA incidence was recoded into yes/no within a person (i.e., incident clinical OA in ≥1 knee vs no incident clinical OA). As we aimed to identify surrogate outcomes that predict long-term OA incidence, rather than those causally linked to long-term OA incidence, confounding is of little importance and presented odds ratios are unadjusted
      • Vanderweele T.J.
      Surrogate measures and consistent surrogates.
      .
      Next, for all potential surrogate outcomes showing a significant association with incident clinical knee OA, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were calculated. All analyses were performed using IBM SPSS version 25, and a P-value <0.05 for statistical significance.
      As sensitivity analyses, potential imaging surrogate outcome markers in the tibiofemoral compartment were analyzed separately for medial and lateral involvement.

      Results

      In total, 236 individuals and 464 knees were free of clinical knee OA at baseline, had required follow-up measures available, and thus were selected for the analyses. Baseline characteristics for the analytic and the excluded populations were presented in Table I. After 2.5 years, 7.2% of knees developed clinical knee OA, after 6.5 years this was 11.3%.
      Table IBaseline characteristics for the analytic cohort and excluded individuals
      Analytic cohort (Mean ± sd
      unless indicated otherwise.
      )
      Excluded (Mean ± sd
      unless indicated otherwise.
      )
      N (individuals)236171
      Age55.7 ± 3.1 years55.7 ± 3.3 years
      BMI31.8 ± 3.7 kg/m233.1 ± 4.8∗∗
      WOMAC Pain (0–100)5.5 ± 9.18.3 ± 13.6∗∗
      WOMAC Stiffness (0–100)10.4 ± 15.020.3 ± 1.5∗∗
      WOMAC Function (0–100)4.8 ± 7.58.7 ± 14.1∗∗
      N (knees)464346
      Radiographic OA severity
      • -
        KL grade 0
      53%48%
      • -
        KL grade 1
      44%42%
      • -
        KL grade 2
      4%11%∗∗
      Joint space width, medial4.7 ± 0.7 mm4.7 ± 0.9 mm
      Joint space width, lateral6.1 ± 1.0 mm6.1 ± 1.0 mm
      Medial knee alignment angle182 ± 2.2°182 ± 2.6°
      Osteophytes on radiography8%4%∗∗
      Chronic pain12%2%∗∗
      Pain severity (0–10)0.4 ± 1.30.6 ± 1.5∗∗
      BMI, KL, sd: standard deviation.
      ∗∗ significant difference (P < 0.05) between groups.
      unless indicated otherwise.
      Table II presents the 2.5-years changes in the potential surrogate outcomes. Medial knee alignment angle, WOMAC pain, and WOMAC stiffness showed worsening and improvement in ≥10% of subjects/knees over 2.5 years. PF and TF BMLs, PF and TF cartilage defects, TF osteophytes, meniscus pathologies, meniscus extrusion, pain severity, and WOMAC function only showed worsening in ≥10% of knees over 2.5 years, while JSN was the only factor showing improvement in ≥10% of knees over 2.5 years.
      Table IIShort-term (2.5 years) changes in potential surrogate outcomes
      Potential surrogate outcomeWorsening (%; mean ± sd
      For continuous measures.
      )
      Improvement (%; mean ± sd
      For continuous measures.
      )
      No change (%; mean ± sd
      For continuous measures.
      )
      Radiography markers
      JSN (change ≥30%)3%13%85%
      Medial knee alignment (change ≥1°
      Worsening: ≥1° more varus in varus/neutral aligned knees or more valgus in valgus/neutral aligned knees, improvement: ≥1° less varus in varus aligned knees or less valgus in valgus aligned knees.
      )
      37%42%21%
      Osteophytes (change ≥1 grade)3%0%97%
      MRI markers
      BML
      As defined by Runhaar et al. (OA&C 2014).
       PF24%9%63%
       TF13%9%73%
      Cartilage defects
      As defined by Runhaar et al. (OA&C 2014).
       PF23%2%70%
       TF12%1%83%
      Osteophytes
      As defined by Runhaar et al. (OA&C 2014).
       PF7%1%88%
       TF12%0%83%
      Meniscal pathologies
      As defined by Runhaar et al. (OA&C 2014).
      25%2%68%
      Meniscal extrusion
      As defined by Runhaar et al. (OA&C 2014).
      18%2%74%
      Effusion/synovitis (change ≥1 grade)9%3%84%
      Clinical markers
      Chronic pain (incident at 2.5 years)9%1%90%
      Pain severity (change ≥1 point)11%;

      4.6 ± 2.2
      7%;

      −3.2 ± 1.5
      79%;

      0.0 ± 0.0
      WOAMC pain
      Determined on a subject-level. Bold figures represent change in ≥10% of the population.
      (change ≥8.3)
      20%;

      28.2 ± 16.0
      11%;

      −15.8 ± 10.7
      69%;

      −0.5 ± 2.8
      WOMAC stiffness
      Determined on a subject-level. Bold figures represent change in ≥10% of the population.
      (change ≥8.0)
      22%;

      31.1 ± 18.9
      19%;

      −19.2 ± 7.9
      59%;

      0.0 ± 0.0
      WOMAC function
      Determined on a subject-level. Bold figures represent change in ≥10% of the population.
      (change ≥10.1)
      14%;

      29.2 ± 16.7
      4%;

      −17.8 ± 6.3
      81%;

      −0.3 ± 3.8
      For continuous measures.
      Worsening: ≥1° more varus in varus/neutral aligned knees or more valgus in valgus/neutral aligned knees, improvement: ≥1° less varus in varus aligned knees or less valgus in valgus aligned knees.
      As defined by Runhaar et al. (OA&C 2014).
      § Determined on a subject-level. Bold figures represent change in ≥10% of the population.
      Table III presents the long-term (6.5 years) incidence of clinical knee OA for subjects/knees with short-term worsening or improvement in the selected potential surrogate outcomes, and the corresponding odds ratios. Short-term worsening of TF cartilage defects and of TF osteophytes, and an increase in pain severity were significantly associated with greater clinical knee OA incidence after 6.5 years.
      Table IIIAssociations of potential surrogate outcomes with ≥10% short-term (2.5 years) change in the population to incident long-term (6.5 years) knee OA incidence
      Potential surrogate outcomeIncident OA for worseningIncident OA for improvementIncident OA for no changeOdds ratio for worsening
      Odds ratios for incident OA using a three-level determinant (worsening, improvement, and no change), with ‘no change’ as reference.
      Odds ratio for improvement
      Odds ratios for incident OA using a three-level determinant (worsening, improvement, and no change), with ‘no change’ as reference.
      Radiography markers
      JSN medial and/or lateral (change ≥30%)1/12 (8%)5/60 (8%)47/392 (12%)0.8 (95% CI 0.3–2.0)
      Medial knee alignment (change ≥1°)
      Worsening: ≥1° more varus in varus/neutral aligned knees or more valgus in valgus/neutral aligned knees, improvement: ≥1° less varus in varus aligned knees or less valgus in valgus aligned knees.
      20/171 (12%)19/194 (10%)14/99 (14%)0.7 (95% CI 0.4–1.3)0.6 (95% CI 0.3–1.2)
      MRI markers
      BML, PF
      As defined by Runhaar et al. (OA&C 2014). Bold figures represent significant odds ratios. ∼ Any worsening or improvement in the PF, TF medial, and/or TF lateral compartment.
      12/109 (11%)7/42 (17%)31/290 (11%)1.0 (95% CI 0.5–2.1)
      BML, TF
      As defined by Runhaar et al. (OA&C 2014). Bold figures represent significant odds ratios. ∼ Any worsening or improvement in the PF, TF medial, and/or TF lateral compartment.
      13/61 (20%)6/43 (14%)32/345 (9%)2.0 (95% CI 0.9–4.2)
      Cartilage defects, PF
      As defined by Runhaar et al. (OA&C 2014). Bold figures represent significant odds ratios. ∼ Any worsening or improvement in the PF, TF medial, and/or TF lateral compartment.
      12/105 (11%)2/9 (22%)35/325 (11%)1.2 (95% CI 0.6–2.3)
      Cartilage defects, TF
      As defined by Runhaar et al. (OA&C 2014). Bold figures represent significant odds ratios. ∼ Any worsening or improvement in the PF, TF medial, and/or TF lateral compartment.
      11/54 (20%)3/5 (60%)36/390 (9%)2.4 (95% CI 1.2–4.7)
      Osteophytes, TF
      As defined by Runhaar et al. (OA&C 2014). Bold figures represent significant odds ratios. ∼ Any worsening or improvement in the PF, TF medial, and/or TF lateral compartment.
      13/55 (24%)0/2 (0%)37/393 (9%)2.2 (95% CI 1.1–4.4)
      Meniscal pathologies
      As defined by Runhaar et al. (OA&C 2014). Bold figures represent significant odds ratios. ∼ Any worsening or improvement in the PF, TF medial, and/or TF lateral compartment.
      14/117 (12%)0/8 (0%)36/323 (11%)1.1 (95% CI 0.6–2.1)
      Meniscal extrusion
      As defined by Runhaar et al. (OA&C 2014). Bold figures represent significant odds ratios. ∼ Any worsening or improvement in the PF, TF medial, and/or TF lateral compartment.
      10/83 (12%)1/11 (9%)39/345 (11%)1.1 (95% CI 0.6–2.2)
      Clinical markers
      Pain severity (change ≥1 point)14/50 (28%)5/34 (15%)32/366 (9%)3.1 (95% CI 1.6–6.0)
      WOAMC pain
      Determined on a subject-level. no odds ratio provided since short-term change in the population was <10% in this direction (see Table II).
      (change ≥8.3)
      11/45 (24%)6/25 (24%)22/158 (14%)2.0 (95% CI 0.9–4.5)2.0 (95% CI 0.7–5.4)
      WOMAC stiffness
      Determined on a subject-level. no odds ratio provided since short-term change in the population was <10% in this direction (see Table II).
      (change ≥8.0)
      11/51 (22%)8/43 (19%)20/136 (15%)1.6 (95% CI 0.7–3.6)1.3 (95% CI 0.5–3.3)
      WOMAC function
      Determined on a subject-level. no odds ratio provided since short-term change in the population was <10% in this direction (see Table II).
      (change ≥10.1)
      6/32 (19%)2/9 (22%)29/186 (16%)1.3 (95% CI 0.5–3.3)
      Odds ratios for incident OA using a three-level determinant (worsening, improvement, and no change), with ‘no change’ as reference.
      Worsening: ≥1° more varus in varus/neutral aligned knees or more valgus in valgus/neutral aligned knees, improvement: ≥1° less varus in varus aligned knees or less valgus in valgus aligned knees.
      As defined by Runhaar et al. (OA&C 2014). Bold figures represent significant odds ratios. ∼ Any worsening or improvement in the PF, TF medial, and/or TF lateral compartment.
      § Determined on a subject-level.no odds ratio provided since short-term change in the population was <10% in this direction (see Table II).
      In Table IV, sensitivity, specificity, PPV, and NPV are presented for the worsening of TF cartilage defects, TF osteophytes, and increased pain severity against long-term clinical knee OA incidence. Overall, results showed the potential surrogate outcomes had high specificity and NPV (89–91%) and low sensitivity and PPV (20–28%) against knee OA incidence.
      Table IVDiagnostic accuracy measures for potential surrogate outcomes with significant associations to knee OA incidence
      Potential surrogate outcomeChange in surrogate outcomeSensitivity
      Calculated for binary change score of the potential surrogate outcome. PPV: positive predictive value. NPV: negative predictive value.
      (95% CI)
      Specificity
      Calculated for binary change score of the potential surrogate outcome. PPV: positive predictive value. NPV: negative predictive value.
      (95% CI)
      PPV
      Calculated for binary change score of the potential surrogate outcome. PPV: positive predictive value. NPV: negative predictive value.
      (95% CI)
      NPV
      Calculated for binary change score of the potential surrogate outcome. PPV: positive predictive value. NPV: negative predictive value.
      (95% CI)
      MRI markers

      Cartilage defects, TF
      As defined by Runhaar et al. (OA&C 2014).
      Worsening
      Associated to significant higher incidence of long-term knee OA incidence.
      22% (11–33)89% (86–92)20% (10–31)90% (87–93)
      Osteophytes, TF
      As defined by Runhaar et al. (OA&C 2014).
      Worsening
      Associated to significant higher incidence of long-term knee OA incidence.
      26% (14–38)89% (86–92)24% (12–35)91% (88–93)
      Clinical markers

      Pain severity (change ≥1 point)
      Worsening
      Associated to significant higher incidence of long-term knee OA incidence.
      27% (15–40)91% (88–94)28% (16–40)91% (88–94)
      As defined by Runhaar et al. (OA&C 2014).
      Calculated for binary change score of the potential surrogate outcome. PPV: positive predictive value. NPV: negative predictive value.
      Associated to significant higher incidence of long-term knee OA incidence.
      Sensitivity analyses indicated predominant involvement of the medial TF compartment. Only short-term worsening of medial TF osteophytes on MRI showed a significant association to long-term OA incidence, with comparable accuracy to the overall TF compartment (see Appendix tables).

      Discussion

      The current study showed that among overweight and obese, middle-aged women at high-risk for knee OA incidence, worsening of TF cartilage defects, worsening of TF osteophytes on MRI, and increased pain severity meet the requirements for surrogate outcomes, as their short-term change was significantly associated to long-term clinical knee OA incidence. However, based on the strength of the associations and the corresponding sensitivity, specificity, PPV, and NPV values, the relevance of these factors as surrogate outcomes for future intervention studies can be questioned.
      The presented significant ORs indicated ‘small’ to ‘moderate’ associations
      • Chen H.
      • Cohen P.
      • Chen S.
      How big is a big odds ratio? Interpreting the magnitudes of odds ratios in epidemiological studies.
      between the potential surrogate outcomes and the incidence of clinical knee OA after 6.5 years. To put it differently one could look at the predictive ability; for the entire population, the incidence of clinical knee OA after 6.5 years was 11% (pre-test probability), but for knees with worsening of TF cartilage defects, worsening of TF osteophytes on MRI, and increased pain severity, the incidence of clinical knee OA after 6.5 years (post-test probability) was 20%, 24% and 28%, respectively. These figures highlight two important observations: (1) among middle-aged women with overweight/obesity, most knees fulfilling these surrogate outcomes will not develop knee OA, and (2) the overall incidence of clinical knee OA after 6.5 years is fairly low, even in this high-risk population.
      One could argue that a good surrogate outcome should have a high PPV for the incidence of the long-term condition under investigation
      • Vanderweele T.J.
      Surrogate measures and consistent surrogates.
      . In the current population, middle-aged overweight and obese women at high risk for knee OA incidence, this would equal an incidence of the surrogate outcome of ≤11% after 2.5 years. Such incidence would still require a large sample size to demonstrate significant and clinically relevant reductions in an intervention study, limiting the feasibility of such a preventive study. To overcome this serious challenge in OA prevention research, establishing valid outcomes for early-stage knee OA, which are anticipated to have higher short-term incidence numbers among high-risk populations, is urgently warranted
      • Whittaker J.L.
      • Runhaar J.
      • Bierma-Zeinstra M.A.
      • Roos E.M.
      A lifespan approach to osteoarthritis prevention.
      ,
      • Runhaar J.
      • Zhang Y.
      Can we prevent OA? Epidemiology and public health insights and implications.
      ,
      • Mahmoudian A.
      • Lohmander L.S.
      • Mobasheri A.
      • Englund M.
      • Luyten F.P.
      Early-stage symptomatic osteoarthritis of the knee - time for action.
      ,
      • Emery C.A.
      • Whittaker J.L.
      • Mahmoudian A.
      • Lohmander L.S.
      • Roos E.M.
      • Bennell K.L.
      • et al.
      Establishing outcome measures in early knee osteoarthritis.
      .
      For the feasibility of a prevention study that aims to reduce the incidence of knee OA, a surrogate outcome with a relatively high short-term incidence would be preferred. In the population described here, any such surrogate outcome will have a short-term incidence larger than the long-term incidence of clinical knee OA, and hence would show high NPV and sensitivity, but low PPV and specificity. To establish significant and clinically relevant reductions in such a surrogate outcome will require a smaller sample size than for a surrogate outcome with high PPV. However, in the hypothetical situation of an effective intervention, the significant reduction in the surrogate outcome will then only translate in a small reduction in the long-term incidence of clinical knee OA, due to the low PPV. Of note, what reductions in long-term OA incidence or what short-term changes in a (potential) surrogate outcome qualify as ‘clinically relevant’ in a preventive setting, is still to be agreed upon by the OA community.
      Previously, using the same data set, we showed that short-term changes in cartilage thickness and short-term changes in cartilage surface integrity were not associated with long-term clinical knee OA incidence
      • Runhaar J.
      • Dam E.B.
      • Oei E.H.G.
      • Bierma-Zeinstra S.M.A.
      Medial cartilage surface integrity as a surrogate measure for incident radiographic knee osteoarthritis following weight changes.
      . Therefore, these markers did not meet the criteria for a surrogate outcome for long-term clinical knee OA. Interestingly, weight loss was associated to the short-term change in cartilage surface integrity and its change was significantly associated to long-term radiographic knee OA incidence
      • Runhaar J.
      • Dam E.B.
      • Oei E.H.G.
      • Bierma-Zeinstra S.M.A.
      Medial cartilage surface integrity as a surrogate measure for incident radiographic knee osteoarthritis following weight changes.
      . However, given the clinical importance of OA illness over OA disease
      • Whittaker J.L.
      • Runhaar J.
      • Bierma-Zeinstra M.A.
      • Roos E.M.
      A lifespan approach to osteoarthritis prevention.
      , the current study focused on incident clinical knee OA. Potentially, incident OA disease (i.e., radiographic knee OA incidence) could be a surrogate outcome itself for long-term clinical knee OA incidence, but this has not been thoroughly assessed thus far.
      The current study has its limitations. Over the follow-up period (6.5 years), drop-out was substantial. Therefore, only 236 of the original 407 participants (58%) were included in the current analyses. Despite significant baseline differences between those included and excluded in the current analyses that potentially limit the generalizability, differences were on average small and of questionable clinical relevance. Moreover, despite the absence of a significant intervention effect in the original PROOF trial, there is a potential for a limited generalizability towards observational data from an identical cohort, due to (small) intervention effects. However, in previous secondary data analyses which treated PROOF data as a cohort, additional adjustments for intervention effects never affected the main results (see e.g.,
      • Landsmeer M.L.A.
      • Runhaar J.
      • van Middelkoop M.
      • Oei E.H.G.
      • Schiphof D.
      • Bindels P.J.E.
      • et al.
      Predicting knee pain and knee osteoarthritis among overweight women.
      ,
      • van der Voet J.A.
      • Runhaar J.
      • van der Plas P.
      • Vroegindeweij D.
      • Oei E.H.
      • Bierma-Zeinstra S.M.A.
      Baseline meniscal extrusion associated with incident knee osteoarthritis after 30 months in overweight and obese women.
      ,
      • van der Voet J.A.
      • Wesselius D.
      • Zhang F.
      • Vroegindeweij D.
      • Oei E.H.
      • Bierma-Zeinstra S.M.A.
      • et al.
      Factors associated with longitudinal change of meniscal extrusion in overweight women without clinical signs of osteoarthritis.
      ). The current study evaluated the potential of several individual outcome measures to serve as a surrogate outcome for long-term OA incidence. In light of the multifactorial nature of OA incidence
      • Hunter D.J.
      • Bierma-Zeinstra S.
      Osteoarthritis.
      ,
      • Katz J.N.
      • Arant K.R.
      • Loeser R.F.
      Diagnosis and treatment of hip and knee osteoarthritis: a review.
      , a single surrogate outcome measures could very well not be suitable to capture all aspects of the disease incidence
      • Yetley E.A.
      • DeMets D.L.
      • Harlan Jr., W.R.
      Surrogate disease markers as substitutes for chronic disease outcomes in studies of diet and chronic disease relations.
      , and rather a combination of outcomes needs to be studied. Moreover, the presented results cannot be used to inform the incidence of new treatment targets, as the analyses focused on surrogate outcomes that numerically captured the incidence of long-term OA, not those that actually mediated the incidence of long-term OA. For that, other analyses and adjustment for important confounders would have been appropriate
      • Vanderweele T.J.
      Surrogate measures and consistent surrogates.
      . Finally, the chosen cut-off of ≥10% progression after 2.5 years is arbitrary. However, lower scores will inevitably hamper the feasibility of such an outcome measure in any future RCTs.
      In conclusion, the current study showed that worsening of TF cartilage defects, worsening of TF osteophytes on MRI, and increased pain severity after 2.5 years were associated to long-term knee OA incidence, among high-risk middle-aged, overweight/obese women. With that, these factors could be seen as surrogate outcomes for long-term OA incidence. However, based on the current results, the applicability of these factors in future RCTs is questionable. Therefore, efforts to validate other potential surrogate outcomes are required.

      Author contributions

      All authors made substantial contributions to analysis and interpretation of data. JR, MvM, and SBZ contributed to the conception and design of the study and acquisition of the data. JR drafted the manuscript, after which all authors revised the manuscript critically for important intellectual content and gave final approval of the version submitted.

      Conflict of interest

      All authors declare no conflicts of interest.

      Role of the funding source

      The PROOF study was funded by The Netherlands Organisation for Health Research and Development and a FP7 Program from the European Union.

      Acknowledgments

      N/A.

      Data statement

      Data are available at reasonable request.

      Appendix A. Supplementary data

      The following is the Supplementary data to this article.

      References

        • Hunter D.J.
        • Bierma-Zeinstra S.
        Osteoarthritis.
        Lancet. 2019 Apr 27; 393: 1745-1759
        • Katz J.N.
        • Arant K.R.
        • Loeser R.F.
        Diagnosis and treatment of hip and knee osteoarthritis: a review.
        JAMA. 2021 Feb 9; 325: 568-578
        • Leyland K.M.
        • Gates L.S.
        • Sanchez-Santos M.T.
        • Nevitt M.C.
        • Felson D.
        • Jones G.
        • et al.
        • Committee PS
        Knee osteoarthritis and time-to all-cause mortality in six community-based cohorts: an international meta-analysis of individual participant-level data.
        Aging Clin Exp Res. 2021 Mar; 33: 529-545
        • Roos E.M.
        • Arden N.K.
        Strategies for the prevention of knee osteoarthritis.
        Nat Rev Rheumatol. 2016 Feb; 12: 92-101
        • Whittaker J.L.
        • Runhaar J.
        • Bierma-Zeinstra M.A.
        • Roos E.M.
        A lifespan approach to osteoarthritis prevention.
        Osteoarthritis Cartilage. 2021 Dec; 29: 1638-1653https://doi.org/10.1016/j.joca.2021.06.015
        • Kellgren J.H.
        • Lawrence J.S.
        Radiological assessment of osteo-arthrosis.
        Ann Rheum Dis. 1957 Dec; 16: 494-502
        • Altman R.
        • Asch E.
        • Bloch D.
        • Bole G.
        • Borenstein D.
        • Brandt K.
        • et al.
        Development of criteria for the classification and reporting of osteoarthritis. Classification of osteoarthritis of the knee. Diagnostic and Therapeutic Criteria Committee of the American Rheumatism Association.
        Arthritis Rheum. 1986 Aug; 29: 1039-1049
        • de Vos B.C.
        • Landsmeer M.L.A.
        • van Middelkoop M.
        • Oei E.H.G.
        • Krul M.
        • Bierma-Zeinstra S.M.A.
        • et al.
        Long-term effects of a lifestyle intervention and oral glucosamine sulphate in primary care on incident knee OA in overweight women.
        Rheumatology (Oxford). 2017 Aug 1; 56: 1326-1334
        • Lo G.H.
        • Strayhorn M.T.
        • Driban J.B.
        • Price L.L.
        • Eaton C.B.
        • McAlindon T.E.
        Subjective crepitus as a risk factor for incident symptomatic knee osteoarthritis: data from the osteoarthritis initiative.
        Arthritis Care Res (Hoboken). 2018 Jan; 70: 53-60
        • Lohmander L.S.
        • Englund P.M.
        • Dahl L.L.
        • Roos E.M.
        The long-term consequence of anterior cruciate ligament and meniscus injuries: osteoarthritis.
        Am J Sports Med. 2007 Oct; 35: 1756-1769
        • Runhaar J.
        • van Middelkoop M.
        • Reijman M.
        • Willemsen S.
        • Oei E.H.
        • Vroegindeweij D.
        • et al.
        Prevention of knee osteoarthritis in overweight females: the first preventive randomized controlled trial in osteoarthritis.
        Am J Med. 2015 Aug; 128: 888-895 e4
        • Schiphof D.
        • Runhaar J.
        • Waarsing J.H.
        • van Spil W.E.
        • van Middelkoop M.
        • Bierma-Zeinstra S.M.A.
        The clinical and radiographic course of early knee and hip osteoarthritis over 10 years in CHECK (Cohort Hip and Cohort Knee).
        Osteoarthritis Cartilage. 2019 Oct; 27: 1491-1500
        • Runhaar J.
        • Zhang Y.
        Can we prevent OA? Epidemiology and public health insights and implications.
        Rheumatology (Oxford). 2018 May 1; 57: iv3-iv9
        • Lohmander L.S.
        • Eyre D.R.
        From biomarker to surrogate outcome to osteoarthritis--what are the challenges?.
        J Rheumatol. 2005 Jun; 32: 1142-1143
        • Prentice R.L.
        Surrogate endpoints in clinical trials: definition and operational criteria.
        Stat Med. 1989 Apr; 8: 431-440
        • Ciani O.
        • Buyse M.
        • Drummond M.
        • Rasi G.
        • Saad E.D.
        • Taylor R.S.
        Time to review the role of surrogate end points in health policy: state of the art and the way forward.
        Value Health. 2017 Mar; 20: 487-495
        • Vanderweele T.J.
        Surrogate measures and consistent surrogates.
        Biometrics. 2013 Sep; 69: 561-569
        • Gandhi G.Y.
        • Murad M.H.
        • Fujiyoshi A.
        • Mullan R.J.
        • Flynn D.N.
        • Elamin M.B.
        • et al.
        Patient-important outcomes in registered diabetes trials.
        JAMA. 2008 Jun 4; 299: 2543-2549
        • Agency EM
        ICH Topic E 9 - Statistical Principles for Clinical Trials.
        • Hunter D.J.
        • Guermazi A.
        • Lo G.H.
        • Grainger A.J.
        • Conaghan P.G.
        • Boudreau R.M.
        • et al.
        Evolution of semi-quantitative whole joint assessment of knee OA: MOAKS (MRI Osteoarthritis Knee Score).
        Osteoarthritis Cartilage. 2011 Aug; 19: 990-1002
        • Salaffi F.
        • Stancati A.
        • Silvestri C.A.
        • Ciapetti A.
        • Grassi W.
        Minimal clinically important changes in chronic musculoskeletal pain intensity measured on a numerical rating scale.
        Eur J Pain. 2004 Aug; 8: 283-291
        • Angst F.
        • Aeschlimann A.
        • Michel B.A.
        • Stucki G.
        Minimal clinically important rehabilitation effects in patients with osteoarthritis of the lower extremities.
        J Rheumatol. 2002 Jan; 29: 131-138
        • Ornetti P.
        • Brandt K.
        • Hellio-Le Graverand M.P.
        • Hochberg M.
        • Hunter D.J.
        • Kloppenburg M.
        • et al.
        OARSI-OMERACT definition of relevant radiological progression in hip/knee osteoarthritis.
        Osteoarthritis Cartilage. 2009 Jul; 17: 856-863
        • Runhaar J.
        • Schiphof D.
        • van Meer B.
        • Reijman M.
        • Bierma-Zeinstra S.M.
        • Oei E.H.
        How to define subregional osteoarthritis progression using semi-quantitative MRI osteoarthritis knee score (MOAKS).
        Osteoarthritis Cartilage. 2014 Oct; 22: 1533-1536
        • Macri E.M.
        • van Middelkoop M.
        • Damen J.
        • Bos P.K.
        • Bierma-Zeinstra S.M.
        Higher risk of knee arthroplasty during ten-year follow-up if baseline radiographic osteoarthritis involves the patellofemoral joint: a CHECK Cohort Study.
        BMC Musculoskelet Disord. 2022 Jun 22; 23: 600
        • Banuls-Mirete M.
        • Lombardi A.F.
        • Posis A.I.B.
        • Shadyab A.H.
        • Chang E.Y.
        • Lane N.E.
        • et al.
        Effusion-synovitis worsening mediates the association between body mass index and Kellgren-Lawrence progression in obese individuals: data from the Osteoarthritis Initiative.
        Osteoarthritis Cartilage. 2022 Sep; 30: 1278-1286
        • Chen H.
        • Cohen P.
        • Chen S.
        How big is a big odds ratio? Interpreting the magnitudes of odds ratios in epidemiological studies.
        Commun Stat Simul Comput. 2010; 39: 860-864
        • Mahmoudian A.
        • Lohmander L.S.
        • Mobasheri A.
        • Englund M.
        • Luyten F.P.
        Early-stage symptomatic osteoarthritis of the knee - time for action.
        Nat Rev Rheumatol. 2021 Oct; 17: 621-632
        • Emery C.A.
        • Whittaker J.L.
        • Mahmoudian A.
        • Lohmander L.S.
        • Roos E.M.
        • Bennell K.L.
        • et al.
        Establishing outcome measures in early knee osteoarthritis.
        Nat Rev Rheumatol. 2019 Jul; 15: 438-448
        • Runhaar J.
        • Dam E.B.
        • Oei E.H.G.
        • Bierma-Zeinstra S.M.A.
        Medial cartilage surface integrity as a surrogate measure for incident radiographic knee osteoarthritis following weight changes.
        Cartilage. 2019 Dec 12; 1947603519892305
        • Landsmeer M.L.A.
        • Runhaar J.
        • van Middelkoop M.
        • Oei E.H.G.
        • Schiphof D.
        • Bindels P.J.E.
        • et al.
        Predicting knee pain and knee osteoarthritis among overweight women.
        J Am Board Fam Med. 2019 Jul-Aug; 32: 575-584
        • van der Voet J.A.
        • Runhaar J.
        • van der Plas P.
        • Vroegindeweij D.
        • Oei E.H.
        • Bierma-Zeinstra S.M.A.
        Baseline meniscal extrusion associated with incident knee osteoarthritis after 30 months in overweight and obese women.
        Osteoarthritis Cartilage. 2017 Aug; 25: 1299-1303
        • van der Voet J.A.
        • Wesselius D.
        • Zhang F.
        • Vroegindeweij D.
        • Oei E.H.
        • Bierma-Zeinstra S.M.A.
        • et al.
        Factors associated with longitudinal change of meniscal extrusion in overweight women without clinical signs of osteoarthritis.
        Rheumatology (Oxford). 2021 Nov 3; 60: 5175-5184https://doi.org/10.1093/rheumatology/keab228
        • Yetley E.A.
        • DeMets D.L.
        • Harlan Jr., W.R.
        Surrogate disease markers as substitutes for chronic disease outcomes in studies of diet and chronic disease relations.
        Am J Clin Nutr. 2017 Nov; 106: 1175-1189