Characterization and diagnosis spectrum of patients with cerebrospinal fluid pleocytosis

Cerebrospinal fluid (CSF) analysis is an important tool to diagnose diseases of the central nervous system (CNS) [1]. Under normal conditions, the blood–brain barrier (BBB) limits immune cell trafficking into the CNS parenchyma to activated T cells, to ensure CNS immune surveillance. During neuroinflammation, immune cells can enter the CSF-filled subarachnoid space at the level of post-capillary venules and maybe via the choroid plexus and across the arachnoid barrier [2]. Elevated leukocyte count within the CSF is an indicator of neuroinflammation that can be caused by a variety of diseases [3], such as CNS infections, autoimmune diseases, trauma, degenerative processes and cancer. Yet, in up to 50% of patients, no etiology can be found [4]. In patients with CSF pleocytosis, the number of cells and the cell type determined by cytological examination already give an indication about the underlying cause. While a predominance of neutrophil granulocytes suggests bacterial meningitis, viral infections usually display a CSF enriched with lymphocytes and monocytes. In addition, the presence of erythrophages or siderophages indicates subarachnoid hemorrhage, whereas prevalence of tumor cells within the CSF is evidence of meningeal carcinomatosis [5].

Finding the correct diagnosis for a patient with elevated white cell count in the CSF can be a challenging task, but is vital for implementing the correct treatment. Acute bacterial meningitis is one of the most dangerous causes of acute neuroinflammation with a high mortality rate and high risk for long-term neurological sequelae [6]. Early diagnosis and prompt start of antibiotic treatment are indispensable for improving clinical outcomes [7].

Our study had two main objectives: (1) to characterize the various causes of CSF pleocytosis through an extensive analysis of a large patient population across all clinical departments, (2) to develop a scoring system capable of distinguishing bacterial meningitis from other diseases presenting with CSF pleocytosis.

A retrospective cohort study was performed including all in-patients with an elevated leukocyte count in their lumbar CSF—defined as > 5 cells/µl—from January 1st, 2015, to December 31st, 2017, in a large university hospital in Germany.

CNS infections accounted for one-third of all cases with CSF pleocytosis, while autoimmune and neoplastic diseases were found in 21% and 16% of patients, respectively. Thirty percentages were due to “other diseases.” This underscores the importance of CNS infections as the cause of CSF pleocytosis on the one hand; on the other hand, it also demonstrates the necessity to think of alternative etiologies in the majority of patients who are found to have CSF pleocytosis.

A smaller study which examined 244 patients with elevated CSF cell count > 5 cells/µl found infectious causes in 18%—comparably to the percentage of patient with CNS infections with determined pathogen of 19.7% in our study. The number of patients with neoplastic (11%) as well as inflammatory (5%) and autoimmune diseases (4%) was lower than in our study [4]. In contrast, 53% of the cases were initially reported as unknown. Of those, 79 patients were treated for a presumptive diagnosis with the majority being suspected CNS infections (51/244, 20.9%)—correlating with our 14.1% of patients with CNS infections without determined pathogen. Another 51 patients (20.9%) of the category “unknown” remained undiagnosed. In our study, only 49 cases (4.3%) were reported as unknown reflecting the differences in diagnostic rigor of CSF pleocytosis. Interestingly, in another Danish study on the etiologies of patients with CSF pleocytosis, a much higher percentage of infections (61.4%) was reported followed by miscellaneous causes (12.7%), vascular (9.7), neurodegenerative (7%), neoplastic (5%) and inflammatory causes (4.2%) [8]. This discrepancy is mostly due to a different cell count cutoff: The Danish study included patients with > 10 cells/µl CSF, while this study included those with > 5 cells/µl, resulting in a higher percentage of patients with CNS infections in the Danish cohort. When using the same cutoff of 10 cells/µl CSF in our patient collective, CNS infections (n = 199, 24.4% [with detected pathogen] and n = 136, 16.7% [without pathogen]) comprised 41.1% of all cases, while autoimmune (n = 153), neoplastic (n = 135) and other CNS diseases (n = 192) contributed 17% and 24%. The Danish study analyzed patient data through the biochemical database from 5 hospitals in the capital region of Denmark from 2003 to 2010. Further differences in frequency of the various CNS diseases might be due to differences in patient cohorts or hospital care levels.

Regarding meningitis, CSF white cell count is a well-known marker that can help differentiate between bacterial and viral origin: One study reported that 99% of patients with bacterial meningitis show CSF cell count over 100 cells/µl, while 87% exceed 1000 cells/µl [9]. Yet, in another cohort study, only 66% of patients with bacterial meningitis revealed a white cell count over 1000/µl [10], thus rendering CSF cell count solely not a reliable predictive factor. A typical constellation with granulocytic pleocytosis > 1000 cells/µl, CSF protein > 1000 mg/l and lactate > 3.5 mmol/l is present in approximately 80% of cases [1]. Other parameters such as CSF lactate, serum procalcitonin and CSF glucose [11] have been suggested previously as predictive of bacterial meningitis with mixed results of sensitivity and specificity [12‐15]. Even though predictive factors indicating bacterial meningitis have been identified, highly reliable markers are still lacking.

The values of blood and CSF glucose to calculate glucose index were determined simultaneously (within 3 h) in close to 80% of our patient cohort. Interestingly, in patients that were treated primarily in the department of neurology, the CSF/serum glucose index was calculated in over 90% of their patients—as it is standard there—while in patients treated in non-neurological departments, a CSF/serum glucose index was available in only ~ 50% of cases. These real-world data represent an important finding showing there is still room for improvement in CSF examination, especially in non-neurological departments. A replacement of the standard glucose index by the determination of CSF lactate could be a solution. The incomplete determination of glucose index in patients with CSF pleocytosis might have contributed to the absence of significance for differentiating between bacterial meningitis and other causes of elevated CSF count. Another reason might be the presence of e.g., patients with meningeal carcinomatosis and thus reduced glucose index within the control group. In line, when we tried to differentiate bacterial meningitis from viral infections, the glucose index was highly significant (0.19 vs. 0.56, p < 0.001), correlating with previous studies that reported significantly lower glucose index in patients with bacterial meningitis than in those with viral meningitis [14, 15]. Several recent studies proposed CSF lactate as most accurate parameter to discriminate bacterial from viral meningitis [12, 16, 17], attributing it higher diagnostic accuracy than CSF glucose, glucose quotient, CSF protein and CSF leukocyte count [18]. As of late, CSF lactate is increasingly used to discriminate CSF infections from non-infectious causes of CSF abnormalities in post-neurosurgical patients [19, 20] although it is not sufficient as a single marker [21‐23]. Yet, as CSF lactate parameters were not available for all patients in our study due to the need of an extra sample for its determination, it was not included in the CHANCE score. Moreover, as elevated CSF lactate occurs in various neurological diseases including stroke, seizures and mitochondrial pathologies [7, 24, 25], differentiating bacterial meningitis from other causes of CSF pleocytosis might be less accurate than in comparison with viral meningitis.

Regarding the etiology of CSF pleocytosis, the last group of CNS diseases “other” than infectious, autoimmune or neoplastic causes, represents a group of great clinical interest. There were 30% miscellaneous CNS diseases such as cerebral hemorrhage, ischemic stroke or epilepsy in our study. In some of these cases, for example patients with primary headaches, psychiatric disorders, ischemic strokes or seizures, the documented diagnoses do not explain the CSF pleocytosis. This discrepancy might be due to the retrospective design of this study where re-testing was not possible. In some cases, there could have been a coincidence of the documented patient’s diagnosis and a neuroinflammatory reaction with undetected etiology. Initially, there were 30 patients with neurodegenerative disorders and CSF pleocytosis. When analyzed in detail, we found the following main explanations for CSF pleocytosis: (1) contamination of CSF with peripheral blood or bone marrow and placement of a Tuohy drain before CSF analysis in cases of normal pressure hydrocephalus. Also, there are several documented cases in the literature of patients with seizures or ischemic strokes who concomitantly show elevated CSF cell count [26‐29]. Other studies have reported patients with CSF pressure disorders, post-intracranial surgery or even depression to display CSF pleocytosis [30‐34]. Therefore, in most cases of our miscellaneous group, elevated CSF cell count was a secondary finding without relevance for the patients’ primary disease. The group of 47 patients with “other neoplastic diseases” and CSF pleocytosis are not unexpected as our clinic is a large center for hematooncology and neurosurgery where many patients with brain tumors are treated who can present with elevated cell count in up to 30% of cases [35].

Assessing a combination of clinical features and laboratory parameters, we tried to develop a score to differentiate bacterial meningitis from other disease entities with CSF pleocytosis. The score was developed from the clinical data of our cohort and revealed that the presence of altered mental status, nuchal rigidity, CSF cell count > 100 cells/µl, CSF protein > 100 mg/dl, elevated white blood cell count and CRP > 5 mg/dl indicate bacterial meningitis in patients with CSF pleocytosis. For the score to reach the cutoff, 4 or more of those criteria must be fulfilled. Therefore, when a patient with elevated CSF cell count shows 4 or more of those criteria, bacterial meningitis is very likely. Still, sensitivity of 92% and specificity of 91% are not sufficient alone for clinical decision making and to rule out bacterial meningitis based on the score alone. As such, it is still important to start empiric antibiotic therapy as early as possible in cases with suspected bacterial meningitis, even if single characteristic symptoms are missing. This can especially be the case in immunosuppressed patients who are less likely to present with characteristic meningitis symptoms [36]. However, the CHANCE score can, among others, still be a helpful additional tool in clinical decision making,

Other scores for adults included CSF pleocytosis, CSF gram stain and abnormal immune status, examination or laboratory findings [37], CSF cell count, CSF protein, CSF lactate, CSF/serum glucose index and peripheral neutrophil count [38, 39]. Overall, several factors of other scores differentiating bacterial meningitis are also included in our CHANCE score.

In contrast to existing scores, the CHANCE score was developed from a large patient cohort for adults to discriminate bacterial meningitis not only against viral meningitis, but also against all other causes of CSF pleocytosis. The containing clinical and laboratory criteria can be gained very quickly, even at night, and give an indication about probability of bacterial meningitis without further microbiological testing.

A limitation of this study lies in its retrospective character which did not allow for retesting of the documented diagnoses. In several miscellaneous cases, CSF pleocytosis might be a secondary diagnostic finding not connected with the patients’ primary disease. Moreover, for the development of the CHANCE score, only factors that were examined in most patients could be evaluated, leading to exclusion of CSF lactate, for example.

In conclusion, this study illustrates the most common causes for CSF pleocytosis including CNS infections, neoplastic or autoimmune CNS diseases. We identified the following criteria as relevant for differentiating cases of bacterial meningitis from other causes of CSF pleocytosis: CSF cell count > 100 cells/µl, CRP > 5 mg/dl, altered mental status, nuchal rigidity, CSF protein > 100 mg/dl and elevated WBC count.