Cerebrospinal fluid lactate: measurement of an adult reference interval

Introduction

Differentiating bacterial meningitis from viral meningitis is diagnostically challenging. Cases of bacterial meningitis must be treated rapidly with antibiotic therapy, whilst the majority of patients with viral meningitis have a self-limiting course, which will resolve without specific treatment and without any serious sequelae. ¹

Conventional techniques of pathogen identification by culture are considered the gold standard and are used to confirm the diagnosis of bacterial meningitis. ² Unfortunately, these results are not immediately available to the clinician; cultures are left to grow for 48 h before being reported.

While lactate is not a specific marker for meningeal inflammation, it has been shown repeatedly to be useful in differentiating viral from bacterial meningitis.^3–5 Meta-analysis has shown that cerebrospinal fluid (CSF) lactate has a good level of overall accuracy (AUC 0.98) and performs better than CSF glucose, CSF protein and CSF leukocyte count when differentiating bacterial from viral meningitis. ⁶

However, information on lactate reference intervals in adults is largely historical, and is based on discontinued analytical methods and derived from small data-sets. In a survey of UK laboratories, only 30% of respondents measured CSF lactate and nearly half of these laboratories did not state a reference interval for CSF lactate. Quoted intervals varied from a lower limit of normal of 0.8 mmol/L to an upper limit of 3.1 mmol/L with four laboratories using a reference interval of 1.2–2.1 mmol/L. ⁷

Methods

This prospective observational study measured CSF lactate on samples received within the laboratory after obtaining consent. Any patient, aged 18 years and over, on whom a clinician had requested a CSF glucose measurement was approached for consent and CSF lactate was measured on the sample preserved in fluoride oxalate. Wherever possible, consent was obtained on the same day as sample receipt, and analysis took place on the same day. Where this was not possible, the CSF sample was frozen at −20℃ within 24 h until consent was obtained. No samples were stored frozen for more than one week prior to analysis. ⁸

Samples from patients who had no leukocytes, neutrophils or erythrocytes detected in the CSF using standard cell counting techniques, no culture growth after 48 h and no significant bilirubin (net bilirubin absorbance <0.007) were included in the statistical analysis for the reference interval.

Cell counting was performed manually using Fast Read 102 counting chambers under a light microscope. A differential cell count was performed on all samples with a raised white cell count. A cytocentrifugation deposit was stained using the MERCK Hemacolor kit and the relative numbers of polymorphs and lymphocytes defined. CSF cultures were performed by plating a centrifuged deposit onto blood and chocolate agar and incubating at 37℃ for 48 h in 5% CO₂.

The presence of bilirubin (absorption maximum 476 nm), was measured by spectrophotometric scanning using a Helios spectrophotometer in accordance with national guidelines. ⁹

Radiological data were checked on all patients suspected of subarachnoid haemorrhage (SAH) and patients included in the reference interval cohort if a computed tomography (CT) scan of the head was reported as negative with no indication of SAH or other intracranial abnormality.

Lactate was measured on the Beckman AU2700 analyser using the enzymatic method in which L-lactate is oxidized to pyruvate and hydrogen peroxidase by lactate oxidase. The CSF lactate method is linear from 0.22 to 12 mmol/L and had CVs of 2.9% and 2.4%, at mean concentrations of 1.4 mmol/L and 5.8 mmol/L, respectively, during the period of the study. The assay was performed according to the manufacturer’s instructions and the method achieved satisfactory performance on the external quality assurance (EQA) programme throughout the duration of the study.

A reference interval was derived from results obtained from 120 patient samples. Statistical analysis was performed using Analyse-it for Microsoft Excel, Leeds UK. The Kolmogorov–Smirnov D test was used to evaluate the distribution of results and non-parametric methods were used to compare medians and to define the 95% reference interval.

Results

A cohort composed of 82 females and 38 males was used to derive the reference interval for CSF lactate. Patient ages ranged from 18 to 90 years with a median of 42 years. There was no significant difference between median CSF lactate concentrations in males (1.45 mmol/L) and females (1.5 mmol/L), (P = 0.624, Mann–Whitney U test).

However, median CSF lactate concentrations were significantly higher in patients aged over 50 years (1.6 mmol/L) than they were in patients aged less than 50 years (1.4 mmol/L), (P = 0.001, Mann–Whitney U test).

The CSF lactate concentration in the reference interval cohort ranged from 0.9 mmol/L to 2.6 mmol/L (median 1.5; IQR 0.3) and was not normally distributed (Kolmogorov–Smirnov D method P = 0.0007). Non-parametric methods were used to define the 95% reference interval of 1.0 (90% CI 0.9–1.1)–2.2 mmol/L (90% CI 2.0–2.6).

Discussion

It is acknowledged that the increasing incidence of pneumococcal penicillin resistance worldwide is making the management of patients problematic, so it is pertinent that the correct diagnosis is made to avoid inappropriate treatment. ¹⁰ The reference interval for CSF lactate defined in this study will assist with the future interpretation of results.

Eighty-three samples (69% of the reference interval cohort) were from patients admitted to the Trust with a headache, in whom SAH was subsequently excluded. The CSF lactate reference interval of 1.0–2.2 mmol/L derived in this study is comparable to that observed in previous reports.^11–14 Previous small studies on adult populations gave comparable results using historic methods. Vamosi et al. ¹² and Herold et al. ¹³ proposed adult reference intervals of 0.68–2.10 mmol/L and 0.6–2.4, respectively, the latter on derived from a very small cohort (n = 36). The largest study in terms of numbers in the literature measured CSF lactate concentrations in 7614 samples in patients of all ages, and derived age-related reference intervals using an enzymatic method on the Mira Plus analyser, Roche Diagnostics, UK. Data from Leen et al. ¹⁴ show no difference in reference intervals between males and females, consistent with the findings in the present study. They also demonstrated that CSF lactate reference intervals are age specific; suggesting a reference interval of 1.2–2.2 mmol/L in young adults (18–30 years), 1.3–2.4 mmol/L in patients aged 30–50 years with a gradual increase in the 95th percentile to 2.7 mmol/L in patients >80 years of age. Our data show that patients <50 years have a significantly lower CSF lactate concentration than those >50 years (P = 0.0001) thus confirming an age-dependent increase in CSF lactate within our reference interval cohort.

Confidence in the intervals derived by Leen et al. ¹⁴ is limited by the fact they did not exclude patients with relevant pathology. Neither clinical information nor radiology results were obtained on their reference interval cohort. CSF lactate concentrations ranged from 0.16 mmol/L to as high as 12.5 mmol/L, with 149 samples (1.9% of the data-set) having CSF lactate values of >3 mmol/L. CSF lactate concentrations exceeding 3 mmol/L are considered to be pathological, based on previous studies and could have biased the data-set. ⁶ Exclusion of CSF samples with lactate values of >3 mmol/L did not lead to a change of >10% of the original value of the 95th percentile in adults; however, this could still be significant. The authors did not indicate if CSF values of >3 mmol/L were more common in the older patient cohorts.

More work is required to accurately determine the degree of age-dependent bias on larger data-sets in which relevant pathology has been excluded.

Alterations in CSF dynamics are thought to explain the age-dependent changes in CSF parameters. May et al. ¹⁵ demonstrated an age-related reduction in CSF turnover, which results in changes in the concentrations of CSF constituents. Age-dependent changes in the CSF proteome also occur, with an increase in markers of inflammation and wound repair. ¹⁶

The lower adult reference interval of 1.0–2.2 mmol/L, derived in our study may reflect the fact that relevant disease was excluded in our patient cohort, and our data were not positively biased by the presence of any pathology, which causes increased CSF lactate concentrations.

The upper limit of normal quoted for CSF lactate by UK laboratories, was as high as 3.1 mmol/L in some hospitals. ⁷ Our data indicate that upon excluding relevant pathology, an accurate adult CSF lactate reference interval is significantly lower than this.

Viallon et al. ¹⁷ state that a CSF lactate value greater than 3.8 mmol/L is highly discriminative in the differential diagnosis of bacterial and viral meningitis. In the meta-analysis reported by Huy et al., ⁶ cut-off values for bacterial meningitis varied from 2.1 to 4.4 mmol/L. It is important that local laboratories differentiate between the CSF lactate reference interval, and any cut-off is used for the diagnosis of bacterial meningitis.

Bacterial meningitis is an important cause of preventable morbidity and mortality in the UK. The measurement of CSF lactate can be used in conjunction with other findings and the clinical presentation to differentiate viral from bacterial meningitis. Reference interval data derived in this study will assist in the interpretation of these results in the future.