Calf muscle fat and frailty as serial culprits in worsening knee symptoms after 5 years in women: the 7-year ambers cohort study

We previously demonstrated that thigh intramuscular fat (IMF) predicts painful and progressive knee OA in the OAI. Another analysis of site-specific associations identified calf IMF as a key correlate of knee pain and function, albeit only within cross-sectional data. More studies are showing the importance of frailty in knee OA progression. While muscle is known to impact frailty, it is unclear whether IMF imparts long-term effects on knee OA through frailty or other pathways. 1) To evaluate how calf IMF predicts knee symptom progression beyond minimal detectability; and 2) to consider frailty and accelerated frailty as mediators along this pathway. The Appendicular Muscle and Bone Extension Research Study (AMBERS) comprises 312 postmenopausal women 60-85 years old recruited at baseline through primary care in Hamilton, Canada. Participants completed a 1T peripheral MRI (FSE, 10 slices, 0 gap, 0.312 × 0.312 × 2.0mm, TR/TE: 600/23ms, FA=40o) scan of the 66% calf at baseline. At the same site, a 1-slice peripheral QCT scan was prescribed (0.500 × 0.500 × 2.3mm, 38 kVp energy, 0.3 mA current). Calf IMF was segmented using our fully-automated iterative threshold-seeking algorithm. We explored a second round of IMF segmentation using muscle masks with first round of IMF removed (Fig 1). Muscle was separated from subcutaneous fat and bone using fixed thresholds on pQCT scans after filtration (filter: F03F05F05, inner/outer thresholds: 40 mg/cm3). Muscle density was computed as mass/volume. Participants were followed annually for 6 years, capturing activities of daily living, quality of life, and comorbidity data to calculate frailty using the cumulative deficits approach (CaMos Frailty Index (CFI)). At year 5 and 6, KOOS, Pain Detect and Gender Role Expectations in Pain (GREP) questionnaires were administered per knee. Use of knee-specific pain medications and corticosteroids was captured. ANALYSES: Group-based trajectory modeling classified CFI trajectory patterns from year 1 to 4, limited to third order polynomials (Fig 2). Change in KOOS from year 5 to 6 was dichotomized based on exceeding minimal detectable change (MDC) per subcategory. To address objective 1, a binary logistic regression analysis measured odds ratios (OR) and 95% confidence intervals. For objective 2, path analysis was applied, exploring CFI at years 1 through 4 or CFI trajectory class as mediators. Total, direct and indirect effects were evaluated. Analyses adjusted for age, BMI, diabetes, corticosteroids, knee pain medications, and presence of neuropathic pain. GREP was used as a weight to mitigate gender bias in self-reporting pain. Knee-level analyses coordinated side of calf scans with side of knee questionnaires. Among 298 women with complete baseline, year 5&6 data, 112(37.6%) reported knee pain below the significant threshold of 72 by year 6. Individuals with 1-year declines ≥MDCs ranged from 30.3% (symptoms) to 45.6% (quality of life). Mean CFI score increased from 0.173(0.118) at baseline to 0.214(0.131) at year 5. Each SD (4.45%) higher calf IMF was associated with a 1.58(1.06, 2.35)-fold increased odds for KOOS symptom progression ≥MDC, with effect sizes increasing when including round 2 IMF (OR: 1.70(1.13,2.56)). A lower muscle density (SD: -8.21 mg/cm3) showed a similar effect for KOOS QOL (OR: 2.16(1.50,3.12)). Individually, calf IMF predicted higher/ accelerated CFI (B=0.031, p=0.008, per +SD IMF%; and OR: 1.30(0.97,1.75) for accelerated versus stable CFI); and CFI significantly predicted 1-year KOOS subcategory changes ≥MDCs (ORs: 1.43-2.19 for Year 4 CFI; and 1.70-2.16 for accelerated versus stable CFI). However, the mediated effects of CFI were not significant in path analyses. Total effects of calf IMF on KOOS changes were explained largely by significant direct effects. Effect size and precision was improved after applying GREP as a weight. Although frailty was both a significant outcome of calf muscle adiposity, and predictor of detectable clinical change in KOOS symptoms, there may be more complex mediation pathways at play, obscuring a clear mediation effect for the investigated pathway. Future studies should investigate more proximal mediators before frailty development such as inflammation. Residual low signal fat left unsegmented from an initial review of MR images may contribute to improved distinction across individual knee symptoms despite some of it representing partial voluming. Arthritis Society Stars Career Award 21-035, CIHR Project Grant PJT166012,156274. None. CORRESPONDENCE ADDRESS: [email protected].