Neuropathology and quantitative methods

Each hemi-SN block was randomly cut at different anatomical levels along the rostro-caudal anatomical axis of the structure. This anatomically random cutting minimized possible anatomical selection biases. We were unable to obtain the entire SN for each subject, but we were confident that at least three quarters of the entire structure was actually sampled for each case. This conclusion was based on our samples measurements (see below) in comparison to previous anatomical studies that assessed the entire length of SN.37,38

Tissue cutting and procedures for morphometric-quantitative analyses

The total length of each block of formalin-fixed SN received ranged from 8–10 mm. Tissues were processed at the Neuropathology Research labs at the Biomedical Research Institute of New Jersey (BRInj). Tissue blocks were processed with an automated tissue-processor (Tissue-Tek V.I.P. 1000 Vacuum Infiltration Processor, Ames Division, Miles Laboratories, Inc., Elkhart, IN, USA) using standard protocols. Tissue blocks were then embedded in paraffin, oriented with the rostral area facing the bottom of the paraffin mold, and serially cut along the rostro-caudal direction using a semi-automatic microtome (Leica RM2255, Leica Biosystems, Nussloch, Germany). Each block was cut for its entire length in series of 40 µm-thick consecutive sections, alternating to a series of 16 consecutive 10 µm-thick sections. These serial, consecutive, and alternating sectioning procedures guaranteed the constancy of the established sampling rate for this investigation (one every five 40 µm-thick sections) and simultaneously allowed the use of 10 µm-thick consecutive sections anatomically adjacent to the 40 µm-thick sections for further immunohistochemistry assessments and volumetric measurements.39

The mean length of the sampled SN did not differ between Dys and C brains (mean lengths of 9,392.31±1,518.99 µm and 10,015.38±1,573.2 µm, respectively). The total number of consecutive sections (including 40 µm- and 10 µm-thick sections) obtained for Dys and C were 792.9±129.3 and 845.4±132.3, respectively. No statistical differences in terms of total length of sampled tissue and number of sections obtained were present between the Dys and C groups. Means of 48.6±7.5 and 51.6±7.5 40 µm-thick consecutive sections were respectively obtained for Dys and C.

Each set of 40 µm-thick sections was stained with a 1.0% cresyl violet (CV) solution for neuronal counting.40 Each set of 40 µm-thick sections was separately and randomly recoded by an investigator blinded to the clinical and pathologic diagnoses (M.G.E.). Quantitative measurements using well-established stereological probes (the Optical Fractionator and Nucleator, see below) were performed by a second investigator (D.I.). Slides codes were opened, and statistical analyses were performed only after all measurements were completed.

A single 10 µm-thick tissue section was randomly chosen, stained with hematoxylin and eosin (H&E), and microscopically inspected at low (2.5×) and high (20×) magnifications to assess for possible micro-ischemic vascular pathologies, microhemorrhages, tissue rarefaction, and other possible histologic abnormalities not visible on gross examination.

Procedures for SN neuropathologic quantitative measurements

Each section was stereoscopically inspected (2×) across the entire sectional area to localize SN anatomical borders. The pars compacta and pars reticulata were identified and marked as based on previous detailed anatomical descriptions.41

For neuronal counting, the following histologic criteria were applied:

• 1)  Use of 40 µm-thick sections;
• 2)  Inclusion of any neuron, of any size, containing neuromelanin pigment;
• 3)  Exclusion of nonneuronal cells containing neuromelanin pigment, such as macrophages.

For neuronal volumetric measurements, the following histologic criteria were applied:

• 1)  Use of 10 µm-thick sections;
• 2)  Individuation of a well-defined nonneuromelanin-covered nucleolus for each randomly sampled pigmented neuron;
• 3)  Establishing the nucleolus as the only point of reference for all volumetric measurements in each randomly sampled pigmented neuron.

Measurements were performed using the Stereo-Investigator system, Version 10.0 (MBF Bioscience, Williston, VT, USA), equipped with a digital camera (AxioCam MRm, and MRc, Zeiss, Oberkochen, Germany) and multiple objective head (2.5–100× oil-immersion). The sampling grid area was 500×500 µm, the counting frame was 40×40 µm, and the disector height was 25 µm with guard zones of ±2 µm. The total number of SN pigmented neurons was estimated using the Optical Fractionator probe.42,43

Series of 10 µm-thick consecutive sections adjacent to each previously analyzed 40 µm-thick section were stained with CV to measure the cellular, nuclear, and nucleolar volumes of pigmented neurons. Volumetric measurements were performed using the Nucleator probe.38 Pigmented neuron volumes were measured by placing six rays that automatically centered on the nucleolus and randomly intersecting the cell membrane (Figure 1A). Pigmented neurons were measured if their nucleolus was inside the counting frame intersecting or touching the green inclusion line; they were excluded if their nucleolus intersected or touched the red exclusion line (Figure 1A, upper right quadrant). This second-run of volumetric measurements on separate sets of 10 µm-thick consecutive sections was chosen to increase the number of available pigmented neurons showing a well-defined nucleolus, that is, the established point of spatial reference for volumetric measurements in this study. In fact, a consistent number of pigmented neurons that could be clearly counted on 40 µm-thick sections could not be volumetrically measured due to the presence of neuromelanin covering, partially or totally, the nucleolus (Figure 1B).

Immunohistochemistry

Each Dys and C case underwent an extensive immunohistochemical assessment. For each case, we examined 16 10-µm thick sections adjacent to the median 40 µm-thick set of sections. The 10 µm-thick tissue sections were deparaffinized, hydrated, and treated to block endogenous peroxidase activity (3% hydrogen peroxide in water). They were then rinsed in buffer (Tris with 0.05% Triton-X), microwaved with antigen retrieval solution (sodium citrate, pH 6.0) for 10 minutes, and cooled. Protein block (1.5% horse serum) was applied for 30 minutes. The following antibodies were used: mouse anti-β-amyloid, 17–24 (dilution 1:500, 4G8; SIG-39220, Covance, Princeton, NJ, USA) with overnight incubation at 4°C (sections were pre-treated with 90% formic acid, 5 minutes); mouse anti-PHF-tau (dilution 1:500, MN1020; Thermo Fisher Scientific, Waltham, MA, USA); mouse anti-α-synuclein (dilution 1:500, Ab27766; Abcam, Cambridge, UK) rabbit anti-ubiq (dilution 1:250, ab7780, Abcam); rabbit anti phospho-TDP-43 (dilution 1:2,000, TIP-PTD-P02; Cosmo Bio Co. LTD, Carlsbad, CA, USA). With the exception of the anti-β-amyloid antibody, sections were incubated overnight at 4°C, rinsed with buffer, incubated with biotinylated horse anti-mouse or anti-rabbit secondary antibodies (30 minutes), Vector Kit reagents (Vector Labs, Inc., Burlingame, CA, USA), then with a 3, 3'-diaminobenzidine substrate system (D3939; Sigma, St. Louis, MO, USA) for 5 minutes. Slides were counterstained with hematoxylin, dehydrated in xylene, and coverslipped.

Statistical analyses

Statistical analyses for quantification and volumes were performed to compare Dys and C, adulthood-onset Dys and C, childhood-onset Dys and C, childhood-onset Dys and adulthood-onset Dys, and generalized and segmental/focal Dys. All comparisons were computed before and after excluding co-occurring pathology (LB, LN, tau-NFT, tau-th, ubiq, and pTDP43 inclusions, β-amyloid deposits). The childhood-onset Dys group included subjects with early adulthood-onset diagnoses and the only subject with posttraumatic Dys (Dys#10, onset at age 12).

Neuronal counting results revealed a Gaussian distribution, and a t-test was initially performed to compare Dys and C and Dys and C without copathologies. Furthermore, we performed analyses of variance (ANOVAs) with Tukey's posthoc tests across all comparisons. The Tukey's test compares every mean with every other mean and takes into account multiple comparisons and adjusted p-values for each comparison. Statistical significance was established at p<0.01 (adjusted for all five types of comparisons).

The preliminary analyses for the obtained neuronal volumetric measurements did not show a Gaussian curve, so nonparametric Mann-Whitney tests were performed across all comparisons for those measurements. Data refer to the values estimated for the bilateral SN. All computations were performed using GraphPad Prism software, version 5 (GraphPad Software, Inc., La Jolla, CA, USA).

Immunohistochemistry findings

Overall, 5/13 Dys and 5/13 C were positive for tau lesions. In the majority of cases, tau lesions were sparse or rare (tau-NFT or tau-th), and only two Dys cases (cases #7 and #8) had moderate levels of tau lesions, mostly tau-th.

No C cases had α-synuclein-positive lesions (LBs or LNs), whereas three Dys cases (#8, #10, and #11) did. These three cases were classified as possible Lewy Body disease (possLBD) of the brainstem.40 None of the SN samples were positive for 1–42 β-amyloid (diffuse amyloid or neuritic plaques), ubiq, or pTDP43-positive cytoplasmic intraneuronal inclusions. Ubiq-positive intranuclear inclusions (Marinesco bodies [MB])44 were observed in some cases, and more often in Dys than C. As a reminder, MB are eosinophilic intranuclear neuronal inclusions observed in human brains that have an uncertain origin and pathologic meaning. This intriguing finding has been reported previously; 30 however, larger studies are necessary to confirm the higher frequency of MB in Dys in general or in a specific Dys subtype. Table 4 summarizes the immunohistochemistry findings in both groups.

Quantitative findings: neuronal counting

A total number of 496,601±130,010 and 648,926±162,475 pigmented neurons were estimated in the SNs of the Dys and C groups, respectively. The estimated nigral neuronal population in C did not differ proportionally from previous stereologic studies that assessed the entire SN length in control subjects.45,46

Histograms depicting the estimated nigral pigmented neuronal counts and volumes are shown in Figure 2. The estimated nigral pigmented neuronal population was calculated using neuronal numbers weighted for each section thickness per each subject. A significant reduction of pigmented neurons in SN of Dys versus C was found (p = 0.01). A separate computation after exclusion of all positive LB/LN and/or tau-NFT/th Dys (#6, #7, #8, #10, and #11) and C (#15, #19, #20, #21, and #24) cases was also performed. This second narrower computation confirmed a significant reduction of pigmented neurons in the SN of Dys (n = 8) versus C (n = 7) (p = 0.01). No significant differences in terms of neuronal counts were observed between childhood-onset Dys and adulthood-onset Dys cases regardless of whether cases with co-occurring nigral pathologies were excluded. Table 5 summarizes all comparisons for neuronal counting between Dys and C with their numerical values and statistical significance. See also Figure 4.

The mean for each type of volumetric measurement (cell body, nucleus, and nucleolus) performed on nigral pigmented neurons of Dys and C are shown in Table 5. A mean of 179.7±56.3 nigral pigmented neurons were volumetrically measured (range: 97–297) across all cases. No difference was measured between the mean numbers of pigmented neurons volumetrically measured in Dys (157.5±40.0) and C (201.9±56.3). Importantly, no significant differences in mean cellular, nuclear, or nucleolar volumes were found across all types of comparisons between the Dys and C groups (Table 5). Table 6 shows the coefficient error (CE) values for each performed measurement (counting and volumes) for each subject in the study.

Limitations and strengths

Although the quantitative methods used in this study were very similar to those used in unbiased stereology investigations, and we tried to minimize confounding factors, it is important to consider our results in the context of some important limitations:

• 1)  The obtained quantitative estimations were not based on the entire length of the SN. Although we used well-established and reliable stereology tools (such as the Optical Fractionator and Nucleator probe) and precise histological, neuropathologic, and clinical approaches (i.e., serial sectioning, anatomical randomization, exclusion of co-occurring pathologies, same mean age at death, etc.), this investigation cannot be defined as an unbiased stereology study mainly due to the impossibility of obtaining the entire structure of interest (SN). Our preliminary novel findings need to be confirmed by unbiased stereological investigations that include the entire anatomical extension of SN.
• 2)  We assessed a low number of cases for each specific type of Dys. It is highly probable that different types of Dys have different causes and unique pathologic features. For example, it is highly probable that adulthood- and childhood-onset forms of Dys represent two different pathogenetic pathways. However, it cannot be totally excluded that they share some common clinical and pathologic features. Then, we cannot be certain that future larger neuropathologic quantitative stereologic studies focusing on a single type of Dys could confirm our findings only on some specific type or group of Dys.

This investigation, however, also has several strengths:

• 1)  To the best of our knowledge, this is the first clinicopathologic correlative study performed on a relatively large number of autopsy-Dys brains (n = 13) and an equivalent amount of autopsy-confirmed age-matched control brains applying rigorous neuropathologic quantitative methods to describe possible specific characteristics (e.g., a reduced number of pigmented neurons in the SN).
• 2)  Extensive histopathologic analyses were performed on each single Dys and C case, including possible co-occurring pathologies affecting the SN. In our opinion, this type of immunohistochemistry assessment is essential to decrease the possibility that a lower number of nigral pigmented neurons could be due to the simultaneous presence of nigral degenerative pathologies. These could indeed represent important confounders that are not directly or necessarily related to the pathogenesis of idiopathic forms of Dys. However, our findings do not exclude the possibility that nigral or extranigral neurodegenerative phenomena (e.g., during aging), could contribute to the progression or initial disease manifestations.
• 3)  There was a remarkable effort in reclassifying most of the non-DYT1 Dys autopsy cases available at UMBTB following the most recent classification system changes.4 The reclassification was based on the careful re-examination of all available medical records. This effort was an important attempt to describe new neuropathologic findings using precious brain tissues previously donated and already available to the scientific community in consideration of newer clinical views and modern neuropathologic quantitative and immunohistochemistry techniques.

Our neuropathologic morphometric-quantitative findings, which are similar to the results of other recent pathologic studies,31,32 seem to confirm the direct involvement of the SN in the pathomechanisms of Dys. Importantly, the involvement of the SN in Dys with regard to a reduced number of pigmented neurons (mainly dopaminergic cells) is not necessarily or directly related to dopaminergic dysfunction per se (i.e., dopaminergic function loss), or to a specific clinical phenomenology of Dys. In fact, in a hypothetical and more global context of a neurodevelopmental disorder, it is possible that other nondopaminergic circuits and neurotrasmettitorial abnormalities (i.e., gamma-aminobutyric acid, acetylcholine)50,51 could play important or even major roles. Moreover, results from two clinical and functional neuroimaging series52,53 support a more global view on Dys phenomenology where the motor and nonmotor abnormalities observed in Dys patients are associated with a broad spectrum of abnormalities in multiple cerebral regions, some of which are nondopaminergic. Furthermore, at a higher speculative level, a neurodevelopmental hypothesis for idiopathic Dys (or some specific form of it) that would not primarily involve dopaminergic dysfunction (i.e., dopaminergic function loss) could help to explain the constant and frustrating inefficacy of dopaminergic therapies in most Dys patients, with the remarkable exception of DRD.

Larger studies are needed to confirm our novel findings on idiopathic Dys. However, these preliminary results seem to suggest the presence of a specific neuropathologic substrate in idiopathic Dys in the form of a reduced number of nigral pigmented neurons that is not due to a neurodegenerative process.

Some very important questions remain:

• 1)  In Dys brains, is the nigral pigmented neuronal reduction a loss of neurons (neurodegenerative hypothesis), or does it represent, as we hypothesize, a possible neurodevelopmental defect (neurodevelopmental hypothesis)?54 In support of the latter hypothesis is the observation that volumetric measurements of nigral pigmented neurons in Dys cases were not significantly different from those in C brains as would be expected if atrophic-neurodegenerative processes were actually involved. These morphometric findings are especially relevant considering that the reduction of nigral pigmented neurons in Dys versus C remained significant even after immunohistochemical exclusion of some copathologies frequently associated with SN neuronal degeneration.

In addition, the absence of a difference in the number of nigral pigmented neurons between the only available posttraumatic dystonia case (#10) and C brains seems to support a neurodevelopmental hypothesis for idiopathic Dys. In the future, larger studies of posttraumatic and nonposttraumatic/idiopathic Dys could provide pathologic findings that support or exclude a neurodevelopmental defect of the SN in idiopathic Dys.

• 2)  Does the nigral pigmented neuronal reduction primarily or exclusively affect dopaminergic neurons? Does it specifically impact the pars compacta or pars reticulata of the SN? It is important to recall that the SN (and basal ganglia [BG] in general) contains a considerable level of somatotopic neuronal organization and neurotrasmettitorial complexity. It is not possible to exclude then that fewer nigral pigmented neurons could mainly affect specific subregions of the SN, thus influencing specific cellular and pathogenetic mechanisms and causing various clinical phenotypes and motor and nonmotor manifestations of Dys.
• 3)  Does the nigral neuronal reduction (or loss) represent a cause or an effect of other pathologic (or neurodevelopmental) phenomena?

Our preliminary quantitative findings, if confirmed, would offer novel insight into specific neuronal alterations in idiopathic Dys. However, these new findings need to take into account very complex pathophysiologic mechanisms and possible pathologic differences across the various forms of Dys.55–58 Nonetheless, partial overlapping or common mechanisms among groups or subgroups of Dys with similar clinical phenotypes remain possible.59

Future major efforts should include unbiased stereological analyses in larger groups of different genetic and nongenetic forms of Dys to confirm our observation of fewer pigmented neurons in the SN, and if so, clarify whether this decrease is “congenital” or neurodegenerative in nature. Another important issue is whether this cellular abnormality is unique to the SN or specific forms of Dys.