Concentration gradients of monoamines, their precursors and metabolites in serial lumbar cerebrospinal fluid of neurologically healthy patients determined with a novel LC–MS/MS technique

This study is a sub-study of the Anaesthetic Biobank of Cerebrospinal Fluid (ABC) of the University Medical Center Groningen (UMCG), The Netherlands. The purpose of the ABC is to collect and store CSF of relatively (neurologically) healthy adult patients for future neuroscientific research. The ABC was approved by the Medical Ethical Committee of the UMCG (registration number 2016-174). Details of the main study design have been described in detail [6].

All patients 18 years of age and older scheduled for elective surgery under spinal anaesthesia were approached for participation in the ABC study, except for patients scheduled for caesarean section. After informed consent was obtained, all patients were also screened for eligibility for this sub-study. Patients with relevant neurological, psychiatric, metabolic (e.g. diabetes mellitus) or inflammatory disorders, or relevant medication use (e.g. antidepressants, antipsychotics, immunosuppressive drugs, dopamine antagonists) were excluded. Additional exclusion criteria were: Body Mass Index (BMI) ≥ 30 kg/m², pregnancy, excessive smoking (more than five cigarettes daily), alcohol abuse (men: more than two units per day; women: more than one unit per day), and recreational drug use. Patients were excluded if the procedure was performed in lateral position, blood was visible in the CSF, less than 10 mL CSF was collected or CSF aspiration time exceeded 4 min. Those patients excluded for this study still participated in the ABC study.

Spinal anaesthesia was conducted using a standard protocol [6]. Prior to local anesthetic injection, 10 mL CSF was aspirated into five 2 mL syringes. The first 1 mL of each consecutive 2 mL fraction was immediately used for routine analyses. The remaining 1 mL of each fraction was transferred into a polypropylene tube wrapped in aluminium foil and directly put on ice for transport to the laboratory where it was centrifuged (1000 ×g, 10 min, 4 ℃) and stored in two aliquots of 500 µL at − 80ºC. For all five fractions, glutathione was added to one of the two aliquots for stabilization.

Monoamines, precursors and metabolites assayed included L-DOPA, 3-OMD, dopamine, DOPAC, 3-MT, HVA, noradrenaline, normetanephrine, MOPEG, adrenaline, metanephrine, 5-HIAA and serotonin. Concentrations of all analytes were analysed using online solid-phase extraction in combination with LC–MS/MS as previously described [35, 36]. In short: 50 µL plasma or CSF was pipetted in a 96-well plate mixed with 50 μL of internal standard working solution, 250 μL of 0.5 mol/L dipotassium phosphate, and 4 mmol/L K2EDTA, pH 8.5. Subsequently, 50 μL of 25% (v/v) propionic anhydride in acetonitrile was added, and the plate was vortex mixed for 15 min. Water was added to all wells to a total volume of 1.0 mL. The plate was vortex mixed and centrifuged for 15 min at 1500 g. Then, 50 μL of each calibrator and sample was injected onto the online solid-phase extraction (SPE) LC–MS/MS system. Intraassay imprecision was < 7.1% at three different concentrations for L-DOPA, 3-OMD, catecholamines and metanephrines (n = 20). Limit of quantification (LOQ) was 1, 1, 0.01, 0.01, 0.03, 0.01, 0.05, 0.04 nmol/L for L-DOPA, 3-OMD, DA, noradrenaline, adrenaline, 3-MT, normetanephrine, metanephrine, respectively. For the indoles, intraassay imprecision was < 6.1% at three different concentrations (n = 20). LOQ was 0.01 nmol/L for serotonin, and 15 nmol/L for 5-HIAA. DOPAC, HVA and MOPEG were analysed essentially as described by Ford et al. [37]. The same online SPE system in combination with LC–MS/MS was used as described above. Intraassay imprecision was < 4% for at three different concentrations (n = 10) for DOPAC, HVA, and MOPEG. LOQ was 0.06, 0.05, and 0.13 nmol/L for DOPAC, HVA, and MOPEG, respectively. All samples were analysed in the same analytical run to minimize contribution of analytical variation, with only one freeze thaw cycle. As the serotonin concentration in CSF is much lower than in platelet-rich plasma, the analysis for serotonin had to be repeated with optimal mass spectrometer settings. For the serotonin analysis, most of the samples underwent two freeze–thaw cycles and for a small part of the samples, three freeze–thaw cycles were necessary due to an error in the sample preparation step.

Routine CSF and plasma analyses included glucose, leucocyte count, albumin and total protein concentrations. CSF erythrocyte counts were determined, to assess for blood contamination (due to the spinal puncture) [22].

Data concerning patient demographics, medical history, medication use and the American Society of Anaesthesiologists (ASA) score [38] were obtained from patient records. If logistically possible, the Montreal Cognitive Assessment (MoCA) was performed preoperatively. Details of the lumbar puncture and CSF collection were recorded.

Based on the sample size of 20 and five fractions (repeated measures) we calculated the statistical power to detect a medium within-subject effect for fraction (i.e. gradient; η² = 0.06; [39]) in a RM-ANOVA, using G*Power [40]. With an alpha of 0.05, and expected (test–retest) correlation among repeated measurements at 0.7, the expected power was 0.96. In addition, the minimum effect that could be tested with a power of 0.8 was η² = 0.04 (4% of total variance accounted for by fraction).

Analyses were performed using SPSS statistics software (version 26; IBM Corp., Armonk, New York, Unites Stated of America). Continuous data are presented as mean ± standard deviation or as median [inter quartile range (IQR)] when appropriate, discrete data by category frequencies (%). For every normally distributed laboratory measurement, a possible linear RCG was investigated by running a repeated measures ANOVA (laboratory measurement as dependent variable and fraction as independent within-subject factor). Analyses were checked for assumption violations. To test the within-subject effect, the Greenhouse–Geisser correction was used in case the assumption of sphericity was violated (according to Mauchly’s test). To assess the within-subject differences between the first and each of the subsequent fractions, post hoc pairwise comparisons were performed. P-values were adjusted for multiple testing using the Bonferroni correction. For non-normally distributed outcomes, a Friedman test with post hoc Wilcoxon signed-rank tests was used to investigate the presence of within-subject differences between fractions. Pearson/Spearman correlation coefficients were calculated to investigate the relationships between blood and CSF fractions. Qalb was calculated by dividing the concentration of albumin in CSF by the concentration of albumin in blood, times one thousand.

Metanephrine and adrenaline concentrations were quantifiable in plasma, but were below the quantification limits in CSF (0.04 nmol/L and 0.03 nmol/L respectively).

Concentrations of dopamine, DOPAC, 3-MT, HVA, noradrenaline, normetanephrine and 5-HIAA increased significantly across CSF fractions (all p < 0.001) (Table 2 and Fig. 1). A substantially higher concentration in the fifth compared to the first fraction was found for dopamine (+ 67%:), DOPAC (+ 26%), 3-MT (+ 50%), HVA (+ 29%), noradrenaline (+ 9%), normetanephrine (+ 24%) and 5-HIAA (+ 35%). Although a similar pattern was observed for CSF MOPEG, these differences across CSF fractions did not reach statistical significance (p = 0.054). There were also no significant concentration gradients across fractions for L-DOPA and 3-OMD. For the concentrations of most molecules there was a large degree of inter-individual variability, but a much smaller intra-individual variability, with clear concentration trends across fractions in several cases (Additional file 1: Fig. S1).

Although serotonin concentrations were detectable, we do not consider them sufficiently reliable for publication and do not report them. Samples from this study underwent a maximum of three freeze thaw cycles which resulted in degradation of serotonin. The serotonin concentrations that were measured were lower than published values [41] and were 5—tenfold lower than concentrations measured in a larger sample of the ABC biobank study (data not shown here). The samples from the ABC biobank study were analysed after only one thaw cycle.

CSF 3-OMD concentrations were strongly correlated with plasma concentrations (all CSF fractions p < 0.001) (Additional file 2: Table S1). There was no significant correlation between plasma and CSF concentrations for the remaining molecules (p = 0.051–0.992), except for MOPEG levels of the second CSF fraction (p = 0.012).

The results of routine assays are summarised in Table 2 and Fig. 2. Albumin concentrations decreased significantly from 0.21 g/L in the first 2 mL to 0.19 g/L in the last 2 mL of CSF (p < 0.001). Total protein concentrations showed a similar decrease, from 0.34 g/L in the first CSF fraction to 0.31 g/L in the last fraction (p < 0.001). For glucose concentration, erythrocyte and leucocyte count, no significant effect of CSF fraction was found. Erythrocyte count exceeded 500/μL in eight samples (8% of total), of which four samples of the first fraction, two of the fifth fraction and one sample of the second and third fraction. There was a moderate, positive correlation between plasma and CSF glucose concentrations (p = 0.002–0.004 for all five CSF fractions; Additional file 2: Table S1).