Sympathetic and parasympathetic subtypes of body-first Lewy body disease observed in postmortem tissue from prediagnostic individuals

Abstract

Recent studies suggest the existence of brain-first and body-first subtypes within the Lewy body disorder (LBD) spectrum, including Parkinson's disease. These studies primarily focused on α-synuclein propagation through the parasympathetic vagal and olfactory bulb routes, leaving the possibility of a sympathetic nervous system spreading route unexplored. In the present study, we analyzed two postmortem datasets, which included 173 and 129 cases positive for Lewy pathology. We observed a clear distinction between brain-first and body-first subtypes in early prediagnostic cases with mild Lewy pathology. Brain-first cases displayed minimal peripheral organ pathology in prediagnostic phases, contrasting with marked autonomic involvement in prediagnostic body-first cases. Utilizing the SuStaIn machine learning algorithm, we identified two distinct body-first subtypes, one with vagal predominance and another with sympathetic predominance, in equal proportions. Our study supports the existence of three prediagnostic LBD subtypes and highlights the sympathetic nervous system alongside the parasympathetic system in LBD onset and progression.

Fig. 1. Postmortem data and immunohistochemistry photomicrographs.

a, A total of 16 anatomical regions staged for Lewy pathology in 173 BBAR cases. b, A plot depicting the eight dissemination categories on the x axis and the percentage of diagnosed PD or DLB cases on the y axis. As diagnosed PD and DLB cases were primarily seen in the two final categories, the first six categories were defined as prediagnostic and the two most disseminated categories were postdiagnostic. c, Representative photomicrographs showing phosphorylated α-synuclein immunohistochemical staining in cortical, limbic, brainstem and peripheral tissues. Scale bars, OB 500 µm (insert 50 µm); AMY, DMV and SY 100 µm; all others 25 µm.

Fig. 2. Lewy pathology profiles in dissemination categories.

Lewy pathology severity scores of brain-first (red) and body-first (blue) subgroups plotted according to the eight dissemination categories. Blue boxes highlight autonomic structures (the SY and DMV) and red boxes the AMY and OB. Sample sizes of brain- and body-first groups for each dissemination category in each plot are listed in Table 1. Data are plotted as average and s.e.m. TrE, transentorhinal cortex; CIN, anterior cingulum, T, temporal cortex; F, frontal cortex; P, parietal cortex.

Fig. 3. Lewy pathology profiles in 16 anatomical regions.

Lewy pathology severity scores of brain-first (red) and body-first (blue) subgroups plotted for each of the 16 anatomical regions examined. a, The four regions used for stratifying the brain-first and body-first cases listed in the leftmost column. Those pathology profiles are necessarily biased by the brain- versus body-first stratification procedure. b, The other 12 regions not used for stratification that are independent. Body-first patients show peripheral organ pathology at earlier dissemination categories, whereas brain-first patients show earlier limbic or cortical pathology in comparable dissemination categories. Sample sizes of brain- and body-first groups for each dissemination category in each plot are listed in Table 1. Data are plotted as average and s.e.m.

Fig. 4. Peripheral organ pathology in brain- and body-first groups.

a, Frequency of cases showing Lewy pathology in at least one peripheral tissue in brain-first (red) and body-first (blue) groups. Sample sizes of brain- and body-first groups for each dissemination category are listed in Table 1. b, Summed Lewy pathology severity scores across the four peripheral organs (HE, ESO, ADR and skin) in brain- and body-first groups. Sample sizes of brain- and body-first groups for each dissemination category are listed in Table 1. Data are plotted as average and s.e.m. c, A comparison of the absolute number of cases with confirmed pathology in a single region, within the BBAR and Newcastle datasets. In the Newcastle dataset, no information on OB or AMY was consistently available.

Fig. 5. Three subtypes of LBD.

The SuStaIn algorithm inferred three different patterns of LBD progression. Maps of Lewy pathology deposition are shown for increasing SuStaIn stages with color codes representing pathology severity (cyan = mild, purple = moderate, blue = severe or very severe). Boxes show the positional variance diagrams, with each box representing the certainty that a region has reached a specific level of pathology at the listed SuStaIn stage, with darker colors representing higher certainty. a, A brain-first subtype with disease initiation in the AMY and OB followed by spreading to limbic and upper brainstem regions. Later, the pathology appears in the DMV, SY and cortex. Peripheral tissues start to become positive only when CNS pathology is widely disseminated. b, A parasympathetic body-first type, in which the DMV and LC are the first positive regions. The NBM, SN and SY follow next and then the AMY and OB. Peripheral tissue pathology follows soon after and shows robust pathology earlier compared with the brain-first subtype. c, A sympathetic body-first type presenting initial pathology in the SY and HE. The SY already shows severe pathology when DMV and ADR pathology appears in SuStaIn stages 5–6. Only later, the LC, AAMY and OB becomes positive followed by the SN and NBM. d,e,f, Pearson’s correlations between SuStaIn stage and total Lewy pathology burden. Diagnosed LBD cases (red dots) are almost exclusively seen above SuStaIn stage 29: brain-first (d), parasympathetic body-first (e), sympathetic body-first (f). g,h,i, Total peripheral pathology across SuStaIn stages: brain-first (g), parasympathetic body-first (e), ympathetic body-first (f). Lines show modeled trajectories calculated using LOESS regression.

Fig. 6. SuStaIn analysis restricted to CNS regions.

a, Two subtypes suggesting when the SuStaIn analysis was restricted to ten CNS regions. The brain-first subtype (54%) was characterized by initiation in the OB and AMY, followed by propagation to the brainstem and limbic structures, and then the cortical regions. The body-first subtype (46%), characterized by initiation in the DMV followed by caudorostral spreading and involvement of the AMY and OB, occurred only after the LC and SN were involved. b, The complete SuStaIn analysis defined one brain-first and two distinct body-first subtypes, as shown here. The arrows in the anatomical figures show simplified theoretical spreading routes as indicated by the SuStaIn positional variance diagrams in the boxes. The diagrams suggest that in parasympathetic-predominant LBD, pathology propagates initially via the vagus (purple arrow), followed slightly later by pathology propagating through the sympathetic connections (blue arrows). In sympathetic-predominant LBD, this order of propagation is reversed. Illustration templates created using BioRender.com.

Fig. 7. Comparison to multimodal imaging data.

Brain- and body-first postmortem subtypes shown alongside previously published multimodal in vivo imaging data from healthy controls (HC), de novo body-first PD with premotor RBD and de novo brain-first PD without premotor RBD^,. a, De novo brain- and body-first PD patients showing similar magnitude of dopaminergic denervation at diagnosis (P = 0.44). SBR, specific binding ratios. b, De novo body-first PD showing reduced neuromelanin density in the LC compared with de novo brain-first PD (P = 0.032). The arrows show the LC. c, De novo body-first PD showing reduced binding of [¹¹C]donepezil in the colon compared with the de novo brain-first PD. This signifies parasympathetic denervation in body-first PD during the prodromal stage (P = 0.005). The arrow shows the colon. SUV, standard uptake values. d, De novo body-first PD showing severely reduced uptake (invisible heart) of [¹²³I]MIBG in the HE compared with de novo brain-first PD (P = 0.00019). An invisible heart signifies that cardiac denervation in many cases of prodromal body-first LBD starts ≥10 years before the dopamine system degenerates. The red arrow indicates the heart. The box plots depict the median, 25th and 75th percentiles and the whiskers indicate minimum and maximum values. Sample sizes of all in vivo imaging groups are shown in Extended Data Table 2. Group comparisons of imaging data were conducted using one-way ANOVA with Tukey’s multiple-comparison test. NM, neuromelanin; H/M, heart/mediastinum ratio.

Extended Data Fig. 1. Lewy pathology in substantia nigra of brain- and body-first subgroups.

Left. The cumulative area under the curve (AUC) of LBD severity scores is larger in the SN of brain-first (red) than in body-first (blue) LBD in comparable intervals of regions. Right. The AUC values in the table suggest that brain-first cases may be diagnosed with PD based on motor symptoms at earlier dissemination stages compared to body-first, since brain-first cases accumulate more Lewy pathology in the SN at earlier dissemination stages.

Extended Data Fig. 2. Determination of optimal number of subtypes.

a. After cross- validation of the complete dataset (score 0-3), the cross-validation information criterion (CVIC) indicates a preferable model fit for 2 N-splits, corresponding to 3 subtypes. b. High log- likelihood supports the conclusion that a model with 2 N-splits (3 subtypes) provides the best fit. c. Markov Chain Monte Carlo (MCMC) trace supports three different subtypes. The optimal number of subtypes was guided by 10-fold cross-validation. Model uncertainty was estimated through 100.000 Markov Chain Monte Carlo (MCMC) iterations.

Extended Data Fig. 3. Three subtypes of Lewy body disease.

The SuStaIn algorithm inferred three different patterns of Lewy body disease progression. Maps of Lewy pathology deposition are shown for increasing SuStaIn stages with color codes representing pathology severity (red = mild, purple = moderate, blue = severe, black = very severe). Boxes show the positional variance diagrams, with each box representing the certainty that a region has reached a specific level of pathology at the listed SuStaIn stage, with darker colors representing higher certainty. a. A brain-first subtype with disease initiation in the amygdala and OB, followed by spreading to limbic and upper brainstem regions. Later, the pathology appears in the DMV, SY and cortex. Peripheral tissues only start to become positive when CNS pathology is widely disseminated. b. A parasympathetic body-first type, in which the DMV and LC are the first positive regions. The NBM, SN, and SY follows next and then the amygdala and OB. Peripheral tissue pathology follows soon after and shows robustness much earlier compared to the brain-first subtype. c. A sympathetic body-first type presenting initial pathology in the SY and heart. The SY already shows severe pathology when DMV and adrenal gland pathology appears in SuStaIn stage 5-6. Only later, the LC, amygdala and OB becomes positive followed by the SN and NBM.

Extended Data Fig. 4. Determination of optimal number of subtypes in restricted dataset analysis.

Markov Chain Monte Carlo (MCMC) trace shows that only two sufficiently different subtypes were seen, when restricting the dataset to 10 CNS regions. The optimal number of subtypes guided by 10-fold cross-validation. Model uncertainty was estimated through 100.000 Markov Chain Monte Carlo (MCMC) iterations.