Statistics from Altmetric.com
We read the paper by Kahlon et al with great interest.1 Comparative studies about the use of different diagnostic techniques to predict the response to shunting in hydrocephalus are of great value as they are likely to form a landmark for future clinical practice. Therefore, it is of paramount importance that the procedures taken for comparison are methodologically sound.
Unfortunately, the interpretation of the lumbar infusion study given by the authors raises our concern. For unknown reasons, the authors have taken into account only the end equilibrium pressure obtained during a constant rate lumbar infusion and neglected the baseline CSF pressure. The authors presumed that this pressure was the same in everybody and equal to the value resulting from the mean taken from the whole cohort (11 mm Hg).
Interpretation of the infusion study can be based only partially on the resistance to CSF outflow (Rcsf), with other parameters describing CSF dynamics like the baseline pressure, elasticity, etc, taken into account.2–4 The resistance to CSF outflow is, undoubtedly the most important parameter, about which a number of independent studies have been conducted in the past,2,5 including a quite recent multicentre Dutch trial.6
The proper way to calculate Rcsf results from the well known Davson’s formula2:
ICP reached during infusion is equal to:
Subtracting the first equation from the second it can be derived:
From the last equation it seems obvious that the Rcsf must be calculated as the difference between end equilibrium CSF pressure and baseline CSF pressure, divided by the infusion rate (in this case 0.8 ml/min).
In cases when the pressure increases above 40 mm Hg during infusion, the Rcsf can be presumed to be greater than 40 minus baseline CSF pressure divided by the infusion rate.
There is still not complete agreement about which value Rcsf is normal or pathological (thresholds are ranging from 12 mm Hg/(ml/min)5 to 19 mm Hg/(ml/min),6 and are probably age dependent). Between 13 and 19 mm Hg/(ml/min) there is a “grey zone” in which patients may but they not necessarily should improve after shunting.
Is neglecting the baseline pressure likely to be a meaningful error?
Although the standard deviation of baseline CSF pressure mentioned by the authors is quite low (3 mm Hg with mean value of 11 mm Hg), this may theoretically mean that some of the patients (although less than 5%) classified as having normal pressure hydrocephalus may have baseline pressure above 17 mm Hg (that is, mean plus twice standard deviation). The “critical threshold” for CSF pressure reached during infusion study should be set for as 28, not 22 mm Hg.
We can recall our own material from Cambridge: in 133 adult patients presenting with clinical and radiological symptoms/signs of normal pressure hydrocephalus, constant rate computerised infusion studies were performed.3 We calculated properly Rcsf, (that is, taking into account the baseline pressure) and recorded the end equilibrium pressure obtained during the test. The end equilibrium pressure, compared with the threshold calculated as in Kahlon et al1 (that is, resulting from the averaged baseline pressure plus 14 mm Hg/(ml/min) multiplied by the infusion rate), was in disagreement with Rcsf below or above 14 mm Hg/(ml/min) in 23 patients. Therefore, by neglecting the baseline CSF pressure as done by Kahlon et al,1 we would misjudge 18% of our tests. It is not huge, but still a meaningful fraction.
The comment by Czosnyka and collaborators to Kahlon et al1 brings up a highly relevant question concerning the lumbar infusion test in patients with suspected normal pressure hydrocephalus. Namely, if calculations of the resistance to outflow of CSF (Rcsf) is a more adequate measure for predicting the outcome of a CSF shunting procedure than merely recording the steady state CSF pressure level reached during constant rate infusion as originally described by Katzman and Hussey.2 Czosnyka et al claim that the latter does not take the initial pressure level (that is, before infusion is started) into account. This is not totally correct. In fact in the equation for calculating Rcsf, the initial pressure level is deducted and only the effect of the fluid volume infused per unit of time is considered. This assumes that the patient’s own CSF production is similar before as well as during the infusion of artificial CSF, which may be true but is in fact not known. CSF production may well be influenced (downregulated) by longstanding hydrocephalus.3
If the measured initial resting pressure is in the high range, the difference to the infusion steady state plateau pressure level will tend to decrease and if low, the difference will tend to increase. Thus, a high initial resting pressure will tend to disqualify the patient from shunt surgery and vice versa if low. These considerations stimulated us to use the uncorrected infusion steady state pressure level as was originally described2 to predict the outcome of shunt surgery.
Several studies have found Rcsf to be a good predictor of outcome of shunt surgery, but almost a similar number of studies have shown a less favourable predictive value (for references, see Boone et al4). In two recent studies Rcsf and CSF outflow conductance were calculated in patients with suspected normal pressure hydrocephalus, undergoing shunt surgery based on pure clinical symptoms combined with ventricular widening.4,5 The results were partly divergent and while Malm et al5 concluded that outflow conductance (reciprocal to Rcsf) had no predictive value, Boone et al4 found that Rcsf could predict outcome of surgery with the best likelihood ratio at a cut off level of about 18 mm Hg/ml/min. In our study the consequence of using Rcsf calculations with cut off levels of 14 or 18 mm Hg/ml/min had been that 3% (1 of 32 patients) or 22% (7 of 32), respectively, of patients with verified improvement after shunt surgery should have been excluded from treatment.1
At present we cannot see any obvious reason for not using the steady state infusion plateau level as a simple measure of CSF absorption capacity in clinical practice. From a theoretical basis we can agree to the reasoning by Czosnyka et al, but if calculation of Rcsf in clinical practice is a better predictor than merely recording the steady state plateau pressure level remains to be proved and further data are warranted. We are currently scrutinising lumbar infusion test curves to further elucidate the role of other details than only the plateau pressure for selection of patients likely to be helped by shunt operations.
If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.