Statistics from Altmetric.com
Leptomeningeal carcinomatosis, characterised by infiltration of neoplastic cells in the central nervous system (CNS), is rapidly progressive and fatal. It may develop from a neoplasm arising from leptomeningeal cells, such as leptomeningeal-melanocytosis, or by infiltration from a neoplasm from outside the CNS.1 This may be from direct local extension, haematogenous spread or migration along perineural or perivascular spaces.2 The prognosis is extremely poor, with a median overall survival from diagnosis typically less than 3 months.3
Cytological examination of the cerebrospinal fluid (CSF) is the gold standard diagnostic test but its sensitivity varies between 54 and 97%.4,–,10 With such a poor prognosis, prompt diagnosis is important so that appropriate palliative care can be started with minimal delay.
We present the factors that maximise CSF cytology sensitivity and provide a strategy to improve practice.
When to perform CSF cytology
Leptomeningeal carcinomatosis presents with a wide spectrum of neurological symptoms, ranging from confusion through to ataxia, seizures and eventually coma.1 The clinical features include signs of raised intracranial pressure, focal weakness, and isolated or progressive cranial nerve palsies, reflecting basal meningeal infiltration.1 ,2
Up to 19% of patients with known malignancy presenting with neurological symptoms or signs have leptomeningeal carcinomatosis.11 The incidence is 4–15% in solid tumours, 5–15% in leukaemia and lymphoma and 1–2% in primary brain tumours.11
CSF cytology should be considered in any patient with known malignancy or history of malignancy who presents with new neurological signs or symptoms, unless there are clinical contraindications to performing a lumbar puncture.
How to get the most out of CSF cytology
Improving clinical practice
In January 2011, we developed a local CSF cytology guideline, building on the existing evidence base, and through collaboration between the head of the diagnostic laboratory and the clinical director for neurology. We emphasised obtaining at least 10 mls of CSF with immediate fixation in formalin (figure 1).4,–,10 We disseminated the guideline to the clinical staff through educational sessions, clinical case discussion meetings and grand rounds. We also informed all junior doctors at induction. We made the guideline available on the ward in a CSF pack, which also included a dedicated CSF sample bottle containing formalin fixative.
We audited all CSF samples submitted to the laboratory for cytological analysis in the 6 months preceding the guideline's introduction (July–December 2010) and the 6 months following (January–June 2011), to assess their impact on clinical practice. We defined ‘suspected malignancy’ as being when this was documented on the request form. We defined the CSF cytological analysis as ‘positive’ when it showed either diagnostic or suggestive changes of malignancy.
Following the guideline's introduction, all the variables that we assessed improved (table 1). We showed statistically significant increases in the overall median CSF sample volume collected, the proportion of patients who had at least 10 mls of CSF collected, and the proportion of CSF samples fixed in formalin within 30 minutes. There was also a trend towards more patients with suspected malignancy having cells identified that were diagnostic or suggestive of malignancy.
Sample volume is important in determining the diagnostic sensitivity of CSF cytological analysis. The false negative rate may be as high as 32% with small sample volumes (3.5 ml), but only 3% with appropriate sample volumes (>10 ml).7
The median [range] sample volume of CSF was greater in samples that were positive compared to those that were negative (10 ml [2.5-20 ml] and 5 ml [0.5-20 ml] respectively, p=0.0007). When there was at least 6 mls, the odds ratio (OR) of identifying cells diagnostic or suggestive of malignancy was 2.88 (confidence interval [CI] 1.08–7.69, p = 0.01). If there was at least 10 mls, the OR was 3.87 (CI 1.44–10.42, p = 0.0098). Similarly, 21% were positive in the group where samples had at least 10 mls, compared with only 5% in the group where sample volume was between 5.0–9.9 mls and and 6% if below 5 mls (figure 2).
Minimising delays in fixation and analysis
The sensitivity of CSF cytology analysis also falls when there is a delay between sampling and fixation. One study found a reduction in the false negative rate from 55% to 35% when samples were fixed and analysed after 48 hours in a refrigerator compared to immediate fixation and analysis.7
In our study, fixing in formalin at the time of sampling increased the likelihood of identifying a malignancy, although this was not statistically significant (9 [15%] versus 5 [6%] respectively, p = 0.089). We need larger multicentre cluster randomised studies to assess the health economic impact of this intervention.
Leptomeningeal carcinomatosis is an important complication of cancer and has a very poor prognosis. It is essential to have both a high index of clinical suspicion and optimum CSF cytology sampling technique to facilitate timely diagnosis and appropriate palliation.
The sensitivity of CSF cytology is influenced by sample volume and time to fixation. Best practice is to take at least 10 ml for analysis and to add the samples to formalin fixative immediately on the ward, before sending to a laboratory for analysis.
Furthermore, simple local guidelines outlining best practice, educational sessions for junior doctors performing lumbar punctures and providing sampling bottles with fixative on the ward can all improve practice.
Suggestions for further reading
Iqbal Z, Michael BD, Pomeroy I, Ali R, Wilson M, Wieshmann U. Multiple cranial neuropathies: One diagnostic difficulty. Practical Neurology 2012 (in press).
Chamberlain M. Neoplastic Meningitis. The Oncologist 2008;13:967–977.
PM-M and BDM are joint first authors. The authors would like to acknowledge the support of Professor Carolyn Young, Dr Peter Enevoldson, Dr Nicholas Silver, Dr Geoff Keir and the Buxton Laboratory team.
Conflicts of interest BDM is an NIHR Doctoral Research Fellow, the authors have no other conflicts of interest to declare.
Provenance and Peer Review Not commissioned; externally peer reviewed. This paper was reviewed by Brendan MacLean, Truro, UK.
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.