Introduction:
The diagnosis of meningitis due to Mycobacterium tuberculosis (TBM) in general can be extremely difficult in the absence of culture confirmation. A non-definitive test such as adenosine deaminase (ADA) could potentially assist in this regard, although its current value for the diagnosis of TBM remains unclear. The literature on the usefulness of ADA measurement in CSF to assist in the diagnosis of TBM shows inconsistencies especially from an analytical point of view regarding the actual ADA assay methodology.
Methods:
Clinical and laboratory data relating to cerebrospinal fluid (CSF) ADA requests during 2009 and 2010 at the Charlotte Maxeke Johannesburg Academic Hospital (CMJAH) were extracted from patient files and the National Health Laboratory Service (NHLS) laboratory information system. An optimal cut-off for CSF ADA for the diagnosis of TBM was calculated using Receiver Operating Characteristic (ROC) curve analysis. In addition, the performance of CSF ADA in different infective and non-infective categories was assessed. Specifically the performance of CSF ADA was compared between the ‘Confirmed TBM’ category and the categories for ‘Confirmed bacterial meningitis’, ‘Confirmed ventriculitis’, ‘Confirmed cryptococcal meningitis’, and ‘Confirmed viral meningitis / encephalitis’. An attempt was made to develop a prediction rule using the data collected, including the CSF ADA result, to improve the clinical diagnosis of TBM in the absence of culture confirmation. Results:
Total CSF ADA requests considered for these 2 years amounted to 3548. Of these 1490 accounted for patients who had both a CSF ADA and a culture for mycobacteria requested. The optimal CSF ADA cut-off to assist in the diagnosis of TBM was calculated at 2.0 U/l (AUC of 0.86; 95% CI of 0.82 – 0.89; p – value of < 0.0001). The sensitivity at this cut-off was 85.9% and the specificity was 77.7%. A considerable overlap was noted in the 95% distribution of CSF ADA values as well as the outliers for each of the categories considered. No statistically significant difference was noted between the ‘Confirmed TBM’ and the ‘Confirmed bacterial meningitis’ categories as well as between the ‘Confirmed TBM’ and the ‘Confirmed ventriculitis’ categories. Stepwise logistic regression analysis failed to produce a combination of factors with appropriate performance characteristics to be used as a prediction rule.
Discussion and Conclusion:
The CSF ADA cut-off determined in this study is unusually low. This cut-off as well as those reported in other studies, in addition to the actual CSF ADA result, are of dubious value in the diagnosis of TBM and may potentially mislead clinicians. Fundamental issues of specimen integrity, ADA assay standardisation and the overlap in the performance of the assay in different diagnostic categories (especially between ‘Confirmed TBM’ and ‘Confirmed bacterial meningitis’) affect interpretation of the CSF ADA result. An inadequate number of correlations between variables chosen possibly prevented generation of an acceptable prediction rule that included CSF ADA.
| Identifer | oai:union.ndltd.org:netd.ac.za/oai:union.ndltd.org:wits/oai:wiredspace.wits.ac.za:10539/12291 |
| Date | 24 January 2013 |
| Creators | Ekermans, Pieter Anton |
| Source Sets | South African National ETD Portal |
| Language | English |
| Detected Language | English |
| Type | Thesis |
| Format | application/pdf, application/pdf, application/pdf |
Page generated in 0.0018 seconds