Original articles

Temporal evolution of CSF following initiation of treatment for Tuberculous Meningitis

Vinod Patel, Izak Burger, Cathy Connolly



When a decision is made to empirically treat for TBM clinicians will often do follow up lumbar punctures (LP) to document changes occurring in the CSF. Normalization of CSF is then used as validation for the decision to treat and serves as indirect confirmation of diagnosis. However the changes occurring in CSF following the initiation of anti-Tuberculous treatment are not well described. This makes the interpretation of a follow up LP difficult. We undertook a retrospective study to determine the temporal evolution of CSF in patients with TBM on anti-TB treatment in order to provide a more rational basis for the interpretation of repeat LPs
We identified all patients who had a lumbar puncture done at King George V Hospital in Durban KZN in the period from 1994-2003. During this time it was usual practice to perform repeat lumbar punctures on patients with a diagnosis of TBM. We selected those in whom a diagnosis of TBM was established. Demographic, clinical, laboratory and radiological data for all the selected patients were recorded. Appropriate statistical processing was done.
Ninety nine patients met the inclusion criteria. A total number of 327 LPs were done. The average number of LPs per patient was 3 (range 3-9). The time intervals between follow up LPs varied from patient to patient. On initial CSF examination, 67.7% of patients had a lymphocytic predominance and 32,3% a polymorphonuclear cells (PMN) predominance. There was a significant decrease in PMN count over time, p < 0.001. The geometrical mean of the PMN count is estimated to be 21.3 at day 0; 7.6 at day 25; 3.1 at day 50 and 0.64 at day 100. By day 150 the PMN count has levelled off at 0.51.The lymphocytes decreased significantly over time, p < 0.001. The geometrical mean of the lymphocyte count is estimated to be 56.0 at day 0; 40.7 at day 25; 29.6 at day 50 and 15.4 at day 100. The lymphocyte count continues to decline and by day 150 reaches 7.8. The geometrical mean of the protein is estimated to be 2.6 at day 0; 1.9 at day 25 (a drop of 15%); 1.3 at day 50 and 15.4 at day 100. By day 150 it is 0.85. The glucose shows a significant increase over time, p < 0.001. The geometrical mean of the sugar is estimated to be 1.8 at day 0; 2.2 at day 25 (an increase of 22%); 2.6 at day 50 and reaches a maximum of 2.8 at day 100. It then decreases slightly to stabilize at 2.5 by day 150.
Our results demonstrate clearly the tendency for CSF to normalize over time. The change is very slow and there is no correlation between rate of normalization of CSF and clinical outcome, for these two reasons the routine practise of documenting normalization of CSF in patients with TBM is of little value. However in the selected patient follow up LP may provide very useful information. The rapid change of CSF glucose and PMN allows us to make reasonable clinical decisions. When a repeat LP does not show definite improvement in these two parameters it should be considered highly atypical for TBM and an alternate diagnosis must be considered. Rapid, reliable and accurate diagnosis of TBM remains a major challenge and new diagnostic methods are needed.

Authors' affiliations

Vinod Patel, UKZN

Izak Burger, UKZN

Cathy Connolly, Department of Biostatistics, Medical Research Council, Durban South Africa

Full Text

PDF (2915KB)


Cerebrospinal fluid, Tuberculosis; Meningitis

Cite this article

South African Medical Journal 2008;98(8):610.

Article History

Date submitted: 2007-07-18
Date published: 2008-08-04

Article Views

Abstract views: 3834
Full text views: 2267

Comments on this article

*Read our policy for posting comments here