Machine learning-augmented and microspectroscopy-informed multiparametric MRI for the non-invasive prediction of articular cartilage composition

BACKGROUND

Articular cartilage degeneration is the hallmark change of osteoarthritis, a severely disabling disease with high prevalence and considerable socioeconomic and individual burden. Early, potentially reversible cartilage degeneration is characterized by distinct changes in cartilage composition and ultrastructure, while the tissue's morphology remains largely unaltered. Hence, early degenerative changes may not be diagnosed by clinical standard diagnostic tools.

METHODS

Against this background, this study introduces a novel method to determine the tissue composition non-invasively. Our method involves quantitative MRI parameters (i.e., T₁, T_1ρ, T₂ and [Formula: see text] maps), compositional reference measurements (i.e., microspectroscopically determined local proteoglycan [PG] and collagen [CO] contents) and machine learning techniques (i.e., artificial neural networks [ANNs] and multivariate linear models [MLMs]) on 17 histologically grossly intact human cartilage samples.

RESULTS

Accuracy and precision were higher in ANN-based predictions than in MLM-based predictions and moderate-to-strong correlations were found between measured and predicted compositional parameters.

CONCLUSION

Once trained for the clinical setting, advanced machine learning techniques, in particular ANNs, may be used to non-invasively determine compositional features of cartilage based on quantitative MRI parameters with potential implications for the diagnosis of (early) degeneration and for the monitoring of therapeutic outcomes.

Human articular cartilage mechanosensitivity is related to histological degeneration - a functional MRI study

OBJECTIVE

To investigate changes in response to sequential pressure-controlled loading and unloading in human articular cartilage of variable histological degeneration using serial T1ρ mapping.

METHOD

We obtained 42 cartilage samples of variable degeneration from the medial femoral condyles of 42 patients undergoing total knee replacement. Samples were placed in a standardized artificial knee joint within an MRI-compatible whole knee-joint compressive loading device and imaged before (δ₀), during (δ_ld1, δ_ld2, δ_ld3, δ_ld4, δ_ld5) and after (δ_rl1, δ_rl2, δ_rl3, δ_rl4, δ_rl5) pressure-controlled loading to 0.663 ± 0.021 kN (94% body weight) using serial T1ρ mapping (spin-lock multigradient echo sequence; 3.0T MRI system [Achieva, Philips]). Reference assessment included histology (Mankin scoring) and conventional biomechanics (Tangent stiffness). We dichotomized sample into intact (n = 21) and degenerative (n = 21) based on histology and analyzed data using Mann Whitney, Kruskal Wallis, one-way ANOVA tests and Spearman's correlation, respectively.

RESULTS

At δ₀, we found no significant differences between intact and degenerative samples, while the response-to-loading patterns were distinctly different. In intact samples, T1ρ increases were consistent and non-significant, while in degenerative samples, T1ρ increases were significantly higher (P = 0.004, δ0 vs δld1, δ0 vs δld3), yet undulating and variable. With unloading, T1ρ increases subsided, yet were persistently elevated beyond δ0.

CONCLUSION

Cartilage mechanosensitivity is related to histological degeneration and assessable by serial T1ρ mapping. Unloaded, T1ρ characteristics are not significantly different in intact vs degenerative cartilage, while load bearing is organized in intact cartilage and disorganized in degenerative cartilage.

Short-Term Oral Sorafenib for Therapy of Intratumoral Shunts of Hepatocellular Carcinoma to Enable Intraarterial Treatment

INTRODUCTION

Significant intratumoral shunts between tumor-supplying arteries and portal or liver veins are a contraindication for transarterial therapy of HCC because interventional treatment of these shunts is frequently insufficient. Sorafenib has anti-angiogenic effects and is indicated for palliative treatment of patients with HCC. Here, we report our experience with the use of sorafenib for the closure of intratumoral shunts in patients scheduled for transarterial therapy of HCC.

MATERIALS AND METHODS

Three patients with HCC, aged 65, 82 and 79 years, exhibited a significant intratumoral shunting from tumor artery to portal (n = 1) or liver veins (n = 2). In all cases, intratumoral shunting had already been suspected based on pre-interventional CT angiography, and DSA confirmed the shunt. Oral sorafenib (800 mg/day) was administered for at least four weeks, only and specifically to occlude the shunt. Hereafter, patients were re-evaluated by CT and DSA.

RESULTS

All patients tolerated the full prescribed dose for at least 4 weeks. In one case, therapy was prolonged with an adapted dose (400 mg/day) due to sorafenib-related hand-foot syndrome. After sorafenib treatment, CT and DSA confirmed a complete closure of intratumoral shunts for all patients. No tumor progression was observed. All three patients hereafter underwent successful transarterial treatment by TACE (n = 2) or TARE (n = 1) without complications. Progression-free survival according to mRECIST was 501, 397 and 599 days, respectively.

CONCLUSION

Even short-term oral sorafenib seems to effectively close intratumoral shunts in patients with HCC and thus might enable transarterial treatment of these patients.

Non-invasive T1ρ mapping of the human cartilage response to loading and unloading

OBJECTIVE

To define the physiological response to sequential loading and unloading in histologically intact human articular cartilage using serial T1ρ mapping, as T1ρ is considered to indicate the tissue's macromolecular content.

METHOD

18 macroscopically intact cartilage-bone samples were obtained from the central lateral femoral condyles of 18 patients undergoing total knee replacement. Serial T1ρ mapping was performed on a clinical 3.0-T MRI system using a modified prostate coil. Spin-lock multiple gradient-echo sequences prior to, during and after standardized indentation loading (displacement controlled, strain 20%) were used to obtain seven serial T1ρ maps: unloaded (δ₀), quasi-statically loaded (indentation₁-indentation₃) and under subsequent relaxation (relaxation₁-relaxation₃). After manual segmentation, zonal and regional regions-of-interest were defined. ROI-specific relative changes were calculated and statistically assessed using paired t-tests. Histological (Mankin classification) and biomechanical (unconfined compression) evaluations served as references.

RESULTS

All samples were histologically and biomechanically grossly intact (Mankin sum: 1.8 ± 1.2; Young's Modulus: 0.7 ± 0.4 MPa). Upon loading, T1ρ consistently increased throughout the entire sample thickness, primarily subpistonally (indentation₁ [M ± SD]: 9.5 ± 7.8% [sub-pistonal area, SPA] vs 4.2 ± 5.8% [peri-pistonal area, PPA]; P < 0.001). T1ρ further increased with ongoing loading (indentation₃: 14.1 ± 8.1 [SPA] vs 7.7 ± 5.9% [PPA]; P < 0.001). Even upon unloading (i.e., relaxation), T1ρ persistently increased in time.

CONCLUSION

Serial T1ρ-mapping reveals distinct and complex zonal and regional changes in articular cartilage as a function of loading and unloading. Thereby, longitudinal adaptive processes in hyaline cartilage become evident, which may be used for the tissue's non-invasive functional characterization by T1ρ.