Longitudinal neuromelanin changes in prodromal and early Parkinson's disease in humans and rat model

Abstract

Neuromelanin-sensitive MRI has been proposed as a biomarker of Parkinson's disease pathology. However, the biological and physical origins of this contrast are debated. A recent rodent model of controlled neuromelanin accumulation in the substantia nigra has been developed and recapitulates several features of Parkinson's disease. In this work, we first combined neuromelanin-sensitive-MRI and histology to study neuromelanin accumulation and neurodegeneration in a humanized rat model of Parkinson's disease. Neuromelanin-sensitive-MRI signal changes were biphasic with an initial increase due to the accumulation of neuromelanin in dopaminergic neurons, followed signal decrease due to neurodegeneration. In healthy subjects and patients with isolated rapid eye movement sleep behaviour disorder, neuromelanin-sensitive-MRI signal increased initially and then decreased similarly as in rodents after reaching a similar maximum signal intensity in both groups. In early Parkinson's disease and converted isolated rapid eye movement sleep behaviour disorder patients, neuromelanin-sensitive-MRI signal drop was greater than in healthy individuals. Results in animals and humans show that neuromelanin-sensitive-MRI is a marker of the intracellular neuromelanin accumulation and then of neuronal degeneration and originates mainly from T₁ reduction effect of neuromelanin.

Keywords: Parkinson’s disease.

Graphical Abstract

Figure 1

Study design of the longitudinal imaging experiment on AAV–hTyr rats cohort. Forty rats were scanned with multiparametric MRI for NM imaging 1 month before, and 1, 2, 4, and 8 months after injection of viral vector expressing human tyrosinase. Ten animals were euthanized after each timepoint post injection for histological analysis.

Figure 2

NM accumulation and neurodegeneration following AAV–hTyr injection in rat SN. (A) OD of intracellular NM in ipsilateral SN of rats euthanized at different mpi (1 mpi: 3 rats, 2 mpi: 5 rats, 4 mpi: 6 rats, 8 mpi: 6 rats). One-way ANOVA F_Time = 632.8, Tukey’s post hoc test between timepoints *P < 0.05, ***P < 0.001. (B) Number of extracellular NM aggregates in ipsilateral SN of rats euthanized at different mpi (1 mpi: 3 rats, 2 mpi: 6 rats, 4 mpi: 3 rats, 8 mpi: 6 rats). One-way ANOVA F_Time = 72.13, Tukey’s post hoc test between timepoints *P < 0.05, ***P < 0.001. (C) Ratio of the number of tyrosine hydroxylase positive (TH⁺) cells in the ipsilateral (TH + ipsi) compared with the contralateral SN (TH + contra) in rats euthanized at different mpi (1 mpi: 3 rats, 2 mpi: 6 rats, 4 mpi: 3 rats, 8 mpi: 6 rats). Two-way ANOVA F_Time = 3.8, Tukey’s post hoc test between timepoints *P < 0.05, ***P < 0.001; F_side = 31.4, Tukey’s post hoc test between ipsi- and contralateral sides ^###P < 0.001 with Tukey’s post hoc test. Histological images of the SN ipsilateral to AAV–hTyr injection with TH staining showing. (D) intracellular NM (arrows) and (E) extracellular NM (arrows). (F) Histological images of the SN contralateral and ipsilateral to AAV–hTyr injection with TH staining showing reduced TH + neurons in the injected ipsilateral side and NM accumulation.

Figure 3

Progression of NM contrast and correlation with NM accumulation and neurodegeneration. (A) Representative NM-MRI image of an AAV–hTyr-injected rat at 1 mpi. Melanized SN is detected as an ipsilateral hyperintense area (arrow). (B) CNR between the ipsilateral and contralateral SN as function of time from AAV–hTyr injection. Grey dots represent single CNR values from individual rats. Black dots represent average value across rats for each timepoint. Continuous line and shading represent estimated marginal means and confidence intervals. LMM, χ² = 99.6, post hoc pairwise comparisons versus baseline using the FDR correction *P < 0.05 and ***P < 0.001, S_ipsi = signal in the ipsilateral SN, S_contra = signal in the contralateral SN. (C) Correlation matrix of CNR parameter with histological quantification between 1 and 8 mpi Spearman’s rho correlation coefficient. Significant correlations are highlighted. *P < 0.05, ***P < 0.001, Spearman’s correlation test.

Figure 4

Progression of MT and longitudinal relaxation parameters in time. (A–C) Representative brain images of (A) R₁, (B) MTR and (C) MPF. (D–F) Variation over time from AAV–hTyr injection of (D) R₁, (E) MTR, and (F) MPF parameters in the ipsilateral SN compared with the contralateral SN. Grey dots represent single values from individual rats. Black dots represent average value across rats for each time point. Continuous line and shading represent estimated marginal means with 95% confidence intervals. LMM, (D) χ² = 44.6, (E) χ² = 5.1, (F) χ² = 8.4; post hoc pairwise comparisons versus baseline using the FDR correction ***P < 0.001.

Figure 5

Progression of susceptibility and transverse relaxation parameters in time. (A–C) Representative brain images of (A) QSM, (B) R₂* and (C) R₂. (D–F) Correlation of magnetic susceptibility parameters with CNR in AAV–hTyr rats. (D) Differences in QSM between ipsilateral and contralateral SN over time from injection. Variations in (E) R₂* and (F) R₂ parameters in the ipsilateral compared with the contralateral SN over time from injection. Grey dots represent single values from individual rats. Black dots represent average value across rats for each timepoint. Continuous line and shading represent estimated marginal means with 95% confidence intervals. LMM, (D) χ² = 32.2, (B) χ² = 5.5, (C) χ² = 2.5; post hoc pairwise comparisons versus baseline using the FDR correction *P < 0.05, ***P < 0.001.

Figure 6

NM-MRI NSI in the SN as a function of age in years in HV (black), unconverted iRBD patients (cyan), converted iRBD patients (orange) and Parkinson’s disease patients (red). NSI values were obtained by dividing each value by the mean NSI value (110.9) of the control subjects at the baseline visit (V1). Circles, triangles and squares represent values measured at baseline (V1), first follow-up (V2) and last follow-up (V4) visits respectively. Vertical dotted lines represent the mean ages of onset in Parkinson’s disease (red) and converted iRBD (orange). For the sake of comparison between groups and within the Parkinson’s disease group, all Parkinson’s disease patients’ ages were aligned at baseline with the mean age of onset to study the effect of disease duration as the actual age was meaningless. Solid curves were added to indicate the NSI evolution of each group using the ‘geom_smooth’ function of the ggplot2 R package with either a linear fit for Parkinson’s disease and converted iRBD patients or a quadratic fit for HV and unconverted iRBD patients. Dashed lines represent extrapolated prolongation of linear trajectory outside of observed age intervals. NSI loss in percentage (%) as indicated by the regression lines at the mean ages of onset was indicated for Parkinson’s disease and iRBD converted patients. Estimated marginal means of linear trends: converted iRBD compared with unconverted iRBD r = −0.537, SE = 0.056, P = 0.10; Parkinson’s disease compared with HV: r = −0.326, SE = 0.326, P < 0.0001; Parkinson’s disease compared with converted iRBD: r = −0.211, SE = 0.331, P = 0.53. NSI, normalized signal intensity; yo, years old; HV, healthy volunteers; iRBD, idiopathic REM sleep behaviour disorder; PD, Parkinson’s disease.