. Aberrations in Peripheral Inflammatory Cytokine Levels in Parkinson Disease: A Systematic Review and Meta-analysis. JAMA Neurol. 2016 Sep 26; PubMed.


Please login to recommend the paper.


  1. Qin et al. have undertaken a meta-analysis of peripheral blood cytokine levels in Parkinson’s disease compared to healthy controls, using data on 1,547 PD patients and 1,107 controls from 25 studies. They showed that peripheral levels of IL-6, TNF-α, Il1β, IL-2, IL-10 C-reactive protein, and RANTES were increased in PD patients. Levels of IFN-γ, IL-4, and IL-8 were not elevated.

    Peripheral cytokines would be attractive markers. They are relatively easy to measure since there are multiple assays available, and thus they have been studied (and still are being studied) for several neurological diseases. Qin et al. report small to moderate effect sizes for the majority of these cytokines, and larger effect sizes for RANTES and IL-2. However, these were reported in only three or five studies, with low patient numbers (n=171 and n=282, respectively). Patient numbers were small for many of the included cytokines, and heterogeneity was large. This means that in this meta-analysis, we cannot assume that the statistical properties of every single study are roughly the same as those of any other part. Thus, it remains to be seen if these reported effects are still large if more data become available. 

    For multiple sclerosis, one might assume signals with larger effect sizes, as this is primarily a neuroinflammatory illness; however, I cannot find a similar meta-analysis in the literature. The same is true for frontotemporal dementias. For FTD, the role of cytokines is only just emerging, and was so far mostly interrogated/anticipated for GRN mutation carriers, due to observations of inflammatory involvement in animal studies. For Alzheimer’s a recent meta-analysis exists (see Swardfager et al., 2010). According to this meta-analysis, similar cytokines changes appear in AD, i.e., IL-6, TNF-α, IL-1β, and a few others, such as TGF-β, IL-12 and IL-18. 

    It is always difficult to compare cytokine levels of different studies in any matrix, be it CSF or blood, as there is no reference material or reference value. The performance of current ELISAs in the lower concentration ranges, which is the range usually seen in neurological diseases and healthy controls, is usually far from optimal. This also applies between ELISA systems in the studies of Qin et al. and Swardfager et al., as the systems were obtained from different suppliers. Non-ELISA methods are equally, or maybe even more, non-comparable, as we show in a paper currently under revision (Malekzadeh et al.). Therefore the likelihood that heterogeneity is small in these non-ELISA methods appears of low relevance to me.

    Even so, the study of Qin et al. shows very consistently elevated levels of several cytokines. This raises the question of what the significance is of this increase, which the authors discuss thoughtfully. Even if some cytokines have some direct movement across the blood-brain barrier, their levels in peripheral blood are related to changes in many organs. Therefore, it is quite possible that the found effects are related to co-morbidity rather than to CNS changes.

    Co-morbidity and other factors that influence cytokine levels, such as age, smoking, physical exercise, and diurnal rhythm, were not taken into account in the studies meta-analyzed here. That is understandable, as when the 1,547 PD patients and 1,107 controls were equally distributed over the 25 included studies, an average 62 PD patients and 44 controls were included in each. Such low numbers do not allow for multiple corrections besides age and sex, and very often some of these confounders are not considered at patient inclusion for long-term biobanking, or during study design. But importantly, these confounders are very relevant to consider in peripheral blood studies for diseases of the central nervous system, where changes should be expected to be small. 

Make a Comment

To make a comment you must login or register.