Individualizing Thresholds of Cerebral Perfusion Pressure Using Estimated Limits of Autoregulation

Abstract

Objectives: In severe traumatic brain injury, cerebral perfusion pressure management based on cerebrovascular pressure reactivity index has the potential to provide a personalized treatment target to improve patient outcomes. So far, the methods have focused on identifying "one" autoregulation-guided cerebral perfusion pressure target-called "cerebral perfusion pressure optimal". We investigated whether a cerebral perfusion pressure autoregulation range-which uses a continuous estimation of the "lower" and "upper" cerebral perfusion pressure limits of cerebrovascular pressure autoregulation (assessed with pressure reactivity index)-has prognostic value.

Design: Single-center retrospective analysis of prospectively collected data.

Setting: The neurocritical care unit at a tertiary academic medical center.

Patients: Data from 729 severe traumatic brain injury patients admitted between 1996 and 2016 were used. Treatment was guided by controlling intracranial pressure and cerebral perfusion pressure according to a local protocol.

Interventions: None.

Methods and main results: Cerebral perfusion pressure-pressure reactivity index curves were fitted automatically using a previously published curve-fitting heuristic from the relationship between pressure reactivity index and cerebral perfusion pressure. The cerebral perfusion pressure values at which this "U-shaped curve" crossed the fixed threshold from intact to impaired pressure reactivity (pressure reactivity index = 0.3) were denoted automatically the "lower" and "upper" cerebral perfusion pressure limits of reactivity, respectively. The percentage of time with cerebral perfusion pressure below (%cerebral perfusion pressure < lower limit of reactivity), above (%cerebral perfusion pressure > upper limit of reactivity), or within these reactivity limits (%cerebral perfusion pressure within limits of reactivity) was calculated for each patient and compared across dichotomized Glasgow Outcome Scores. After adjusting for age, initial Glasgow Coma Scale, and mean intracranial pressure, percentage of time with cerebral perfusion pressure less than lower limit of reactivity was associated with unfavorable outcome (odds ratio %cerebral perfusion pressure < lower limit of reactivity, 1.04; 95% CI, 1.02-1.06; p < 0.001) and mortality (odds ratio, 1.06; 95% CI, 1.04-1.08; p < 0.001).

Conclusions: Individualized autoregulation-guided cerebral perfusion pressure management may be a plausible alternative to fixed cerebral perfusion pressure threshold management in severe traumatic brain injury patients. Prospective randomized research will help define which autoregulation-guided method is beneficial, safe, and most practical.

Figure 1. Schematic depicting the theoretical relationships between CPP and PRx including estimation of ‘optimal’ CPP (CPPopt), CPP LLR, and CPP ULR

The relationship between CPP and PRx can be approximated by fitting a U-shaped curve (2^nd order polynomial) whereby with both high or low values of CPP, the PRx is impaired (top right panel). With impaired PRx there is a positive correlation between changes in MAP and changes in ICP (normally calculated over a 5-minute window) However, for intermediate CPP values, PRx is (probably) efficient (bottom right panel), and the CPP at which PRx is most negative is termed the CPPopt (as indicated by the black dot). By applying a threshold for impaired cerebral PRx (here PRx= +0.30), the CPP at which PRx switches from being intact to impaired can be calculated to give an estimate of the lower and upper CPP limits of reactivity (LLR and ULR respectively). Summarising the relationships between CPP and the variables(LLR, ULR, CPPopt), we can appreciate how far a patients CPP is from their autoregulation guided target or range. CPP - cerebral perfusion pressure; PRx- Pressure reactivity index; MAP - Mean arterial pressure; ICP-intracranial pressure; CPPopt- CPP optimal; PRxopt - PRx optimal. ULR- Upper limit of reactivity; LLR- Lower limit of reactivity

Figure 2. Comparison of %time spent in different ‘zones’ of CPP as defined by fixed thresholds (left), optimal CPP based thresholds (middle), or flexible limits of reactivity (right)

A. Using the BTF recommended fixed CPP values of 60 and 70 mm Hg as lower and upper thresholds, most patients spent the majority of time with a CPP above the upper threshold. In those that died, the proportion of time with CPP > 70 mm Hg was lowest, and proportion of time with CPP < 60 mm Hg the highest. B. The CPPopt-based thresholds were estimated as follows: lower threshold was a CPP more than 10 mm Hg below CPPopt (ΔCPPopt < -10); while the upper threshold was a CPP more than 10 mm Hg above CPPopt (ΔCPPopt > +10). Those with severe disability or vegetative state spent the most amount of time above the upper CPPopt threshold, while those who died spent the most time with CPP below the lower CPPopt threshold. C. Referencing patients current CPP to their individually estimated lower and upper limits of reactivity (LLR and ULR respectively) reveals the most consistent pattern; those patients with increasing burden of disability spent more time with a CPP below their LLR and above their ULR. GR-good recovery; MD-moderate disability, SD-severe disability; VS- vegetative state; D-death; GOS- Glasgow outcome score; CPP- cerebral perfusion pressure; LLR-Lower limit of reactivity; ULR-upper limit of reactivity; WLR-within limits of reactivity; ΔCPPopt - CPP-CPPopt; BTF- Brain Trauma Foundation

Figure 3. Comparison of receiver operator characteristic (ROC) curves for predicting mortality (top) and unfavourable outcome (bottom).

Percentage of time with CPP below the lower limit of reactivity (%CPP