To our knowledge this is the first study using an unbiased cytokine microarray analysis approach to identify potential serum biomarkers for the discrimination of neurodegenerative parkinsonian disorders. In our study the primary screen using cytokine microarray analysis on pooled samples yielded 12 cytokines differentially regulated in PSP/CBS, MSA, PD and controls (GRO, ICAM-1, IL-2 R-alpha, IL-6 R, leptin, MCP-4, NAP-2, PDGF-BB, prolactin, RANTES, TIMP-2 and TRAIL R3). Functional annotation clustering revealed that these cytokines are associated with immune responses, chemotaxis and cell migration, whereas no association with neuronal or glial function was found. These results suggest that it is rather unlikely that the identified cytokines are markers reflecting specific pathophysiological processes for neurodegenerative parkinsonism.
By using a second independent, analytic method (ELISA or bead-based immunoassays), we have tried to confirm the results derived from the microarray analysis in the same samples for seven of these 12 cytokines (ICAM-1, IL-2 RA, leptin, MCP-4, PDGF-BB, prolactin and RANTES). However, only two (PDGF-BB and prolactin) of the seven cytokines were significantly different amongst patient groups (PSP/CBS, MSA and PD) and controls using both methods. In a second step we were able to confirm these results in a different cohort containing larger number of patients and controls. The striking difference between initial analysis and initial replication by a different method likely reflects differences in the sensitivity and specificity of the used antibodies and appears to be consistent with the frequent irreproducibility of many serum biomarker studies of neurodegenerative diseases. For instance the results of a recent study describing a cytokine-array based investigation of protein-panels enabling to distinguish patients with AD from HC  could not be reproduced in two subsequent studies [9, 10]. Beside this methodological concern, our results suggest that the serum prolactin levels are influenced by dopaminergic antiparkinsonian treatment, but not the patient group; dopaminergic antiparkinsonian treatment remained the only significant variable on prolactin levels in a multivariate analysis and patients with no such treatment had similar serum prolactin levels to controls. This is in line with studies indicating a crucial role for dopamine as an inhibitor of prolactin production as well as with studies suggesting that untreated PD patients have normal prolactin release, whereas pharmacologic stimulation of dopamine D2-receptors with dopaminergic antiparkinsonian treatment leads to decreased serum prolactin levels [16–18], corroborating the reliability of our cytokine-array screening approach.
Serum PDGF-BB levels were significantly increased in the patient groups compared to the controls with the highest levels found in PSP/CBS. PDGF-BB, a member of the platelet-derived growth factor family, is a homodimer encoded by the PDGFB gene . PDGF was originally discovered in serum and identified as a major mitogenic factor for connective tissue cells as well as some epithelial and endothelial cells. In addition, PDGF is chemotactic for fibroblasts, smooth muscle cells, neutrophils and mononuclear cells. However, PDGF also appears to be ubiquitous in neurons throughout the CNS, where it is suggested to play an important role in neural development, function and neuron survival as well as in mediation of glial cell proliferation and differentiation . Experimental studies from the 1990s demonstrated that PDGF-BB acts as a trophic factor for rat and human mesencephalic dopaminergic neurons promoting gene expression, survival and neurite outgrowth in culture [20, 21]. In the 6-OHDA rat model, PDGF could counteract the 6-OHDA-induced degeneration of mesencephalic DA neurons when administered prior to the insult . In the same in vivo model, PDGF-BB as well as BDNF administration post insult was capable of increasing the numbers of newly formed cells in the striatum and substantia nigra . To the best of our knowledge there are no studies reporting on PDGF concentrations in brain tissue or in the CSF from parkinsonian patients. In peripheral blood, levels of PDGF-BB have been analyzed in AD, but the results of studies were controversial [7, 9, 10, 24]. Interestingly, the most recent of these studies also included 11 demented PD patients and, in line with our results, they found an increase in PDGF-BB levels in their plasma .
In our cohort of patients with neurodegenerative parkinsonian syndromes, there was no association of PDGF-BB levels with the disease duration or the Hoehn and Yahr score. Also in the subgroups of early untreated patients PDGF was elevated to the same extent as in the whole groups. Thus, it is tempting to speculate that the increased serum PDGF-BB levels might reflect early compensatory mechanisms as a response to neurodegeneration. This would appear consistent with increasing evidence that immunological and inflammatory processes including microglial over-activation as well as increased synthesis and release of cytokines could be a key player in PD pathogenesis . PDGF-BB elevations could therefore represent an important factor in central and peripheral communication between neurons, glial cells and peripheral immune cells. Besides the expression in neurons and Schwann cells, PDGF-BB is also synthesized by vascular endothelial cells, macrophages, fibroblasts and megakaryocytes . Since PDGF-BB has several important functions in the peripheral circulation such as mitogenic and chemotactic effects on mesenchymal stem cells [26–29], it is more likely that the increased serum PDGF-BB levels observed in our study might reflect a response to pathological changes in the periphery. This could explain why PDGF-BB was also detectable in our control sera to a marked extent, which accounts for the suboptimal differentiation of neurodegenerative Parkinsonian syndromes from controls. Our study has some limitations: it was performed in patients with a clinical diagnosis of neurodegenerative parkinsonian syndromes without pathological confirmation. Hence, misdiagnosis in some patients, especially in the early disease stages, cannot be excluded. Also, some of the included PSP patients suffered from the most reliably identifiable classic picture of PSP (i.e. Richardson’s syndrome), whereas the true diagnostic dilemma lies with atypical presentations like PSP-parkinsonism . Given the pathological heterogeneity of a ‘corticobasal syndrome’ , most commonly including CBS and other neurodegenerative causes such as PSP with both diseases sharing the same tauopathy and due to the limited number of CBS patients (n = 8) included into the present study, these two groups were gathered together. However, in all validation experiments CBS and PSP patients were analyzed separately and we found no differences between these groups. Furthermore, we only analyzed seven out of twelve deregulated cytokines in the initial cytokine array, depending on the availability of commercial available test kits. Therefore, the deregulation of five proteins significantly altered in the initial screening was not further validated (GRO, IL-6 R = IL6R, NAP-2, TIMP-2, TRAIL R3). The cross sectional design of our study did not allow for a direct correlation of PDGF-BB levels and disease progression. Therefore a longitudinal study is now needed to address this important question. Finally, our control group included not only healthy controls, but also patients with internal diseases. We think that this is not a limitation but rather a strength since these controls could avoid confounding effects of internal diseases related to aging in patients with neurodegenerative diseases. However, a post-hoc analysis revealed no differences between INC and HC for the cytokines analyzed in this study.