gms | German Medical Science

Objective: Intracranial pressure (ICP) reflects volumetric loads in the cerebrospinal system in rest and under pathophysiological states. It is a typical experience in inspecting intracranial pressure (ICP) curves that the modulation degree in intracranial pulse-pressure (PP) is changing by ICP magnitude. Previous studies investigated coherence from sinus heart beats (RR) and ICP in short-term analyses. Here, we focus on a further modulation parameter, i.e., the time difference between PP subpeaks p1 and p2 (∆t12), revealed some insight in statistical interdependency of directional causality between cardiac and cerebrospinal processes.

Methods: In twelve unsupervised episodes of 50 min with distinct dynamics, ICP was recorded for diagnostic reasons in patients with idiopathic intracranial hypertension (IIH). An offline beat-to-beat analysis of PP waveforms was carried out extracting ∆t12 parameter (X) along with the corresponding mean ICP (Y). For all such data pairs, Granger causality (GC) and stationarity were evaluated in overlapping data blocks. In 9/12 episodes the working hypothesis was that bi-directional causalities (XY, YX) differ in epochs before and after an ICP transient or transition. Three episodes were rated controls because no change in ICP dynamics was noticeable.

Results: Granger causality was formally accepted for significance levels p<0.01 in overall 688/3114 resp. 678/3496 analyses before and after a change in ICP dynamics. Directed GC was found in 384 (XY) vs. 304 (YX) before the onset of change; after that, GC amounted to 414 (XY) vs. 264 (YX) counts. Contingency analyses were dominated by XY GC in 7/9 episodes plus 3/3 controls; two analyses showed an inverted result. In all but one analysis stratification revealed significant differences at p<0.01.

Conclusion: In all episodes, bi-directional GC was found independently from ICP levels and dynamics. Our results also show that parameter ∆t12 predominantly Granger-causes mean ICP before and after a transient change in ICP. In only one episode inverse GC (YX) became evident. Our findings comply in part with elaborate frequency domain conclusions from other groups found at baseline levels and during B-wave activity. Unlike these findings, however, in which the ICP influenced RR mainly while B-wave activity was present, the omnipresence of both XY and YX causalities allow the presumption that arbitrary wave activity seems to be a natural component of low- and high-level ICP at least in patients with IIH.