Analytical Method Validation for Estimation of Neurotransmitters (Biogenic Monoamines) from Cerebrospinal Fluid Using High Performance Liquid Chromatography

Abstract

Biogenic amine neurotransmitters such as serotonin and dopamine are essential for signaling in both central and peripheral nervous system. Their metabolism is a multistep pathway and any defect in this results in alteration in metabolites of serotonin 5-Hydroxyindole acetic acid (5HIAA) and dopamine homovanillic acid (HVA) and 3-O-Methyl Dopa (3-OMD). Estimation of these metabolites in cerebrospinal fluid (CSF) assists in diagnosis of neurotransmitter defects. Their estimation is technically demanding and is currently available only in referral centers. We aimed to optimize a method for analysis of 5HIAA, HVA and 3-OMD. A high performance liquid chromatography (HPLC) method with electro chemical detector (ECD) was standardized for estimation. Analysis for method validation, reference range verification and clinical correlation was performed. Linearity obtained for 5-HIAA, HVA and 3-OMD was 65.35–2615.0 nmoles/l, 68.62–2745.0 nmoles/l and 236.5–4730.0 nmoles/l respectively. The coefficient of variation for internal quality controls ranged from 5 to 14% and the external proficiency testing samples (n = 16) were within peer group range. CSF metabolite levels of samples for reference range analysis overlapped with age matched ranges reported in literature. Among the 40 suspected patients analyzed for clinical testing four were found to have a neurotransmitter defect. These patients were then confirmed with molecular testing and clinical correlation. The method is validated and can be adapted in a clinical laboratory with analytical competence in HPLC.

Introduction

The biogenic monoamine neurotransmitters [1] are essential for signaling in both central & peripheral nervous system and are involved in the regulation of movement, basal muscle tone, activity levels, mood etc. [2]. An inherited defect of these monoamine neurotransmitter metabolism forms a well-recognized group of inherited metabolic disorders in pediatric neurology [3]. Their presentation is often overlapping with other frequently occurring neurological conditions like cerebral palsy, primary movement disorders, epileptic encephalopathies etc. [4]. Their diagnosis is best made by the estimation of biogenic amine metabolites in CSF & other body fluids and respective enzyme assays [2]. Many of these disorders are treatable and the neurological deficits can be reverted if treated on timely basis. Variation in CSF neurotransmitter Homovanillic acid (HVA), 5-Hydroxyindole acetic acid (5-HIAA) & 3-O-Methyl Dopa (3-OMD) collectively known as biogenic monoamines indicate most of the neurotransmitter disorders and hence are vital for diagnosis.

The estimation of serotonin and dopamine metabolites is preferably done from CSF as their concentration in blood and urine may be uninformative for detection of inherited neurotransmitter defects. [5]. An abnormal CSF neurotransmitter profile is one of the most important indicator of a neurotransmitter disorder. Diagnostic facilities for inherited metabolic disorders are scarce in India and are restricted to amino acid estimation, organic acid analysis, acyl carnitine analysis, enzyme assays for storage disorders and other biomarkers. However, to the best of our knowledge there are no centers offering analysis of CSF biogenic amines. Thus, these disorders are under diagnosed and an availability of CSF biogenic amine analysis service in India will help to diagnose these patients.

The biogenic amine metabolites can be analysed using several different techniques such as HPLC with electrochemical (EC) detection or fluorescent detection (FD) [6] and liquid chromatography coupled with tandem mass spectrometry (LC–MS/MS) [7]. These techniques are available in several analytical and pharmaceutical laboratories however their availability in clinical laboratories is limited. In this study we used the HPLC with ECD in our laboratory for optimizing the estimation of 5HIAA, HVA & 3 –OMD.

Materials and Methods

The study was conducted in the Biochemistry Department of our hospital. The samples for neurotransmitter analysis in suspected cases were referred by the pediatric neurology department of our hospital.

Chemicals and Materials

All reagents were of analytical and HPLC grade. Acetonitrile, Methanol, Potassium dihydrogen phosphate and 85% Orthophosphoric acid were purchased from Merck (Kenilworth, New Jersey, United States). Sodium 1-octanesulfonate monohydrate, 5-Hydroxyindole-3-acetic acid (5HIAA), Homovanillic acid (HVA) and 3-O-Methyl-L-DOPA Monohydrate (3-OMD) were purchased from SantaCruz Biotech (Dallas, Texas, United States), Disodium EDTA was purchased from Sigma Aldrich (St. Louis, Missouri, United States) and CSF matrix based assayed control from Randox (Crumlin, United Kingdom) was used to prepare internal quality control material. This control material had no 5HIAA, HVA & 3-OMD and was used by spiking with respective aqueous standards.

Instrumentation and Conditions

CSF monoamine neurotransmitters were separated using a reverse phase Symmetry C18 column (4.6 × 250 mm, 5 μm particle size) and detected with a Waters 2465 Electrochemical detector. Data was processed using EMPOWER software. The chromatographic separation was done in an isocratic mode at a flow rate of 1.3 ml/min, injection volume 20 µl and temperature of 35ºC. The mobile phase consisted of mixture of 50 mM potassium dihydrogen phosphate, 0.99 mM sodium 1-octanesulfonate monohydrate, 0.053 mM of di-sodium EDTA and 12% methanol. The pH was adjusted to 2.5 with 85% O–phosphoric acid. The electrochemical detection was achieved using 2 mm Glassy carbon (GC) working electrode and Insitu Ag/AgCl (ISSAC) reference electrode. The ECD parameter were set as follows: Range—10 nA, EC—+ 0.80 v, Filt—0.1 s, Offs—+ 0%, Polar—+ , Max Comp—2.5 μA, Spacer—50 μm, Mod—DC.

Preparation of Standards and Quality control (QC)

The stock solutions of 5-HIAA, HVA and 3-OMD were prepared individually in Milli-Q water to obtain a concentration of 100 mg/l each and were stored in 1 ml aliquots at -80ºC. At the time of analysis, working solution mixture was prepared for each batch and was used for calibration curve. Two level internal controls were prepared by spiking the assayed CSF matrix controls with aqueous standards to obtain the concentration of 130.7 nmoles/l and 653.8 nmoles/l for 5-HIAA and HVA. One level internal control was prepared by spiking the assayed CSF matrix control with aqueous standards to obtain the concentration of 591.3 nmoles/l for 3-OMD. These QC samples were stored at – 80ºC until analysis.

Sample Collection and Preparation

CSF biogenic amine reference values were verified in 50 subjects aged from neonates to 15 years. The CSF samples received by the laboratory for evaluation of bacterial or viral meningitis or any other neurological condition of non-metabolic origin were used for reference range verification. The samples were obtained in accordance with Helsinki Declaration [8]. The CSF samples were also referred for biogenic amine estimation from forty patients clinically suspected to have neurotransmitter defects. For analysis, second CSF fraction (1–1.5 ml) was collected and snap cooled prior to sending it to the laboratory. The sample was immediately stored at −80 °C until analysis [9]. Unlike plasma/tissue or blood samples which need sample preparation like protein precipitation, extraction, filtration etc. CSF samples require minimum sample preparation [6]. At the time of analysis CSF sample was thawed and centrifuged at 3000 rpm for 10 min. The supernatant was transferred to an auto sampler vial and 20ul of supernatant was injected for analysis.

Method Validation

Analytical Assessment

The separation of 5HIAA, HVA & 3OMD was verified by the retention times (RT) and area under curve (AUC) using commercially available standards. The selectivity and specificity of the method was assessed by injecting the reagent blank and CSF matrix respectively. The lowest limit of quantification (LLOQ) was determined as the lowest matrix standard which gave precision and accuracy within ± 20% from its defined values. Method linearity was assessed by injecting the higher concentrations above the LLOQ until the highest concentration standard gave the precision and accuracy within ± 20% from its defined values. Based on the LLOQ and the upper limit of the highest concentration five calibration standards of all three analytes were prepared. The concentration ranges for 5HIAA, HVA & 3-OMD were 65.35—2615.0 nmoles/l, 68.62—2745.0 nmoles/l and 236.5—4730.0 nmoles/l respectively. The calibration curve was established for each analyte using peak area of analyte v/s concentration. Slope, intercept and correlation coefficient (R²) were obtained by regression analysis. The inter-day and intra-day precision and accuracy were determined in five replicates on same as well as on different days at all concentration level of standard and QC. The variation of < 20% was considered to be acceptable. The recovery and matrix effect was assessed by spiking three level standards to CSF sample. A recovery between 80 to 120% was considered acceptable. Carryover testing was performed by injecting a reagent blank after the highest level standard and also by injecting same standard before and after analysis of CSF sample. The criteria for acceptance was no peaks in the reagent blank and < 10% variation in the AUC of the analytes in the standard injected before and after the CSF sample.

Quality Assurance Plan

Internal quality control samples were analyzed with every batch of samples. The coefficient of variation of the concentration of all three metabolites for both controls was also calculated and was acceptable when < 20%. As a part of external quality assurance plan the laboratory enrolled for CSF neurotransmitter analysis program of the European Research Network for evaluation and improvement of screening, Diagnosis and treatment of Inborn Errors of Metabolism (ERNDIM). During the study period i.e. years 2018 & 2019, 16 samples i.e. 8 per year were received and analyzed by the laboratory. The results were reported to ERNDIM for peer group evaluation.

Clinical Assessment

The concentration of 5HIAA, HVA and 3-OMD obtained in the 50 CSF samples received in the laboratory were compared with the age matched ranges reported in literature. The CSF of 40 patients suspected to have neurotransmitter defects were analyzed and the results were correlated with clinical and other laboratory findings.

Result and Discussion

Optimization of Chromatographic and ECD Conditions

Chromatographic separation of analytes was performed by reverse-phase chromatography in an isocratic condition (Fig. 1). The method was standardized for various parameters for proper separation. The mobile phase and stationary phase interaction was optimized and the best results were obtained by adjusting the mobile phase pH to 2.5 and using ion pairing modifiers. Low concentration of analytes of interest and matrix interference due to direct injection of CSF sample warrants use of sensitive and reproducible column. In our experience Symmetry C18 column shows good column to column and batch to batch reproducibility and hence was used for estimation.

Fig. 1.

Chromatograms, A-Standard, B-Quality control, C-Dihydropteridine reductase deficiency (DHPR) patient sample with levodopa treatment, D-Dopamine transport defect (DTD) patient sample with levodopa treatment

The ECD is a special detector and not routinely available in analytical laboratories. In our laboratory we have a Waters 2465 detector which is an amperometric detector with a single electrode. The amperometric detector with a single electrode has a low sensitivity [10] and thus posed a challenge to optimize the method for CSF biogenic amine detection.. We customized detector parameters like voltage, range and spacers to reduce background noise and increase intensity.

Method Validation

Method validation was done as per the criteria for chromatography [11–13].

Peak Identification, Selectivity and Specificity

The RTs of 5HIAA, 3-OMD and HVA were obtained as 18.80, 24.95 & 31.59 min respectively. The reagent blank and the CSF matrix blank did not show interfering peaks at the identified RTs of 5HIAA, HVA & 3-OMD (Fig. 2). This suggests that the optimized method conditions are selective and specific for the estimation of biogenic amines.

Fig. 2.

Blank and carryover testing, A-Blank, B-Blank injected after highest standard, C- Standard injected before sample, D- Standard injected after sample

Limit of Quantification (LLOQ)

Different concentrations of the matrix standards were injected to determine the lowest quantifiable concentration which was then injected five times. LLOQ for 5-HIAA was 65.35 nmoles/l, for HVA was 68.62 nmoles/l and for 3-OMD was 236.5 nmoles/l. The %CV of AUC for 5HIAA, HVA & 3-OMD was 1.7, 3.8 and 7.3 respectively.

Linearity

Standards were analyzed at five different concentrations to determine the linearity range of the method. Calibration curves of analytes were regressed by plotting the peak AUC of analytes v/s corresponding concentration of standards. Linearity obtained for 5-HIAA, HVA and 3-OMD was 65.35—2615.0 nmoles/l, 68.62—2745.0 nmoles/l and 236.5—4730.0 nmoles/l respectively. The correlation coefficient (R²) for all analytes was 0.99 (Fig. 3).

Fig. 3.

Linearity of 5HIAA, HVA and 3OMD

Precision and Accuracy

Different concentrations of the standards and QC were processed in replicate on same as well as different days. Intraday % CV at various concentrations spanning the linearity for 5-HIAA, HVA and 3-OMD was between 0.6–5.3, 1.1–5.5 and 2.6–4.6 respectively while the inter day % CV was between 3.2–7.5, 2.4–7.2 and 3.8–10.4 respectively.

Recovery and Matrix Effect

The recovery and matrix effect was analyzed by spiking the patient sample with the analyte standards concentration of 65.35 nmoles/l, 261.5 nmoles/l and 1307.5 nmoles/l. For all analytes recovery was between 90 to 110%.

Carryover Testing

Carryover evaluation performed to ensure the accuracy and precision of our method showed no peaks at the RT of the respective analytes in reagent blank injected after highest standard. Also, the standards injected before and after the CSF analysis showed < 10% variation in their AUC for all 3 analytes (Fig. 2).

Quality Assurance Program

The low level control with 5HIAA and HVA concentration of 130.7nmole/l each showed a %CV of 12.2 and 7.6 respectively while the high level control of 653.8 nmoles/l each showed a % CV of 13.5 & 5.6 respectively. Since the sensitivity of 3-OMD in our method was higher than its reference range only high control with concentration of 591.3 nmoles/l was used as an internal quality control. The %CV of this control was 8.2. ERNDIM proficiency testing results were within ± 2SD of the peer group mean except for 3-OMD wherein 10 samples had results < 236.5 nmoles/L, 6 samples had 5HIAA level < 65.3 nmoles/L and 4 samples had HVA < 68.6 nmole/L which is lower than LLOQ of our method for respective analytes (Table 1).The LLOQ of participant laboratories appear to be almost 10 times lower than that of our method. This may be due to the amperometric detector used by us.

Table 1.

External Quality Assurance Scheme (EQAS) results of 2018 and 2019

Cycle	5-HIAA (nmoles/l)			HVA (nmoles/l)			3-OMD (nmoles/l)
	Lab value	Peer group mean	Z score	Lab Value	Peer group mean	Z score	Lab Value	Peer group mean	Z Score
2018
1	124	125	0.1	238	301	1.4	< 236.5	34	NA
2	470	455	0.3	843	900	0.7	< 236.5	199	NA
3	211	165	1.7	1079	1142	0.5	< 236.5	22	NA
4	< 65.3	29.8	NA	< 68.6	68.6	NA	< 236.5	8	NA
5	185	161	1.4	1127	1136	0.3	< 236.5	20	NA
6	501	439	1.4	899	898	0.1	< 236.5	204	NA
7	136	126	0.8	274	296	0.7	< 236.5	30	NA
8	< 65.3	29.4	NA	< 68.6	70.7	NA	< 236.5	7.82	NA
2019
1	183	158	1.3	374	391	0.3	< 236.5	36.9	NA
2	79	77.8	0.1	167	170	0.2	880	850	0.1
3	< 65.3	15.6	NA	< 68.6	28.5	NA	490	449	0.8
4	< 65.3	31.8	NA	101	80.2	1.8	1185	1205	0.2
5	159	155	0.6	337	364	0.7	< 236.5	40	NA
6	< 65.3	14.6	NA	< 68.6	27.8	NA	447	430	0.3
7	82	75.9	0.8	176	171	0.5	914	842	1.1
8	< 65.3	29.3	NA	84	78.3	0.7	1123	1134	0.2

Z scores for the analytes wer not applicable the concentration of 5-HIAA, HVA and 3-OMD in respective cycles of 2018–19 were below the lowest level of quantification (LLOQ) of our method

NA Not applicable

Reference Range Verification

The age matched reference range was obtained by analyzing CSF samples of 50 subjects aged neonates to 15 years with no neurotransmitter defects (Fig. 4). This range showed an overlap with literature reported reference ranges [14].

Fig. 4.

Reference range verification

The level of 3-OMD in all age groups is < 200 nmoles/l. Our method’s lowest quantification limit for 3-OMD is 236.5 nmoles/l and hence we did not detect 3-OMD in the control group. This low sensitivity may not affect the patient diagnosis as in pathologic conditions the 3-OMD level is much higher than our LLOQ for 3-OMD [2].

Clinical Evaluation

An abnormal CSF biogenic amine concentration indicative of a neurotransmitter defect was obtained in 12 (30%) out of 40 patients studied. Of these 12 patients, a confirmed diagnosis either by molecular analysis or clinical management and drug response could be established in 4 patients (33.3%). Our results for 5-HIAA, HVA and 3-OMD in these 4 patients are consistent with the neurotransmitter defect diagnosis [Table 2]. The remaining 8 patients are being followed up for further evaluation. Kuster et al. [15] in their retrospective data analysis of CSF metabolite in 1200 patient samples found an abnormal biogenic amine pattern in 228 (19%) patients. Of these a confirmed genetic/clinical diagnosis suggestive of a NT defect was made in 54 (24%) patients. The variation in the % of abnormal biogenic amine concentration and the confirmed diagnosis in our study and Kuster et al.’s study could be due to the large difference in the sample size.

Table 2.

Patients with neurotransmitters defects

Age	Neurotransmitter results (nmoles/l)				Clinical history
	HVA	5HIAA	HVA/5HIAA	3-OMD
2 Y/F	215 (231–840)	< 65.3 (89–341)	–	2576 (< 50)	Dihydropteridine reductase deficiency (DHPR) Behavior and speech problem. Levodopa treatment
6 Y/M	< 68.6 (137–582)	< 65.3 (68–220)	–	< 236.6 (< 50)	Sepiapterin reductase deficiency (SPR) Developmental delay, motor > cognitive, not holding neck, not able to sit
1 Y/F	2046 (236–867)	255 (97–367)	8.02 (1–4)	1047 (< 50)	Dopamine transport defect (DTD) Infantile Parkinsonism, Parental consanguinity,3 previous sibling death, Levodopa treatment
1 Y/M	< 68.6 (236–867)	< 65.3 (97–367)	–	1262 (< 50)	Aromatic L-amino acid decarboxylase deficiency(AADC) Hypotonic/involuntary Dystonia, No head control, Oculogyric phenomena, Partial pediatric neurotransmitter disorder

The International working group on neurotransmitter related disorders (iNTD) in their registry and publication in 2016 [16] have reported 25% of NT patients with Aromatic L-amino acid decarboxylase (AADC) followed by 6-Pyruvoyl tetrahydropterin synthase (PTPS) in 23%, Tyrosine hydroxylase (TH) in 12%, Autosomal dominant GTP-cyclohydrolase (ADGTPCH) in 13%, Dihydropteridine reductase (DHPR) in 11% while other disorders are < 5% of the patients registered. Saadet et al. [17] in their retrospective laboratory data analysis of the CSF biogenic amine estimation obtained a confirmed inherited NT disorder in 6 out of 150 patients of which 1 had DHPR,1 PTPS,1 GTPCH,1 TH,1 Pyroxidine-dependent epilepsy (PDE) and 1 Pyridoxamine 5′-phosphate oxidase (PNPO). In our small group of 40 patients 4 had a confirmed NT disorder which included 1 AADC, 1 DHPR, 1 Sepiapterin reductase (SPR) and 1 Dopamine transport defect (DTD) resulting in 25% occurrence of each of these disorder.

Limitation

Literature reported methods for biogenic amine detection use coulometric detectors with two electrodes [18] or tandem mass spectrometry for estimation [7]. Coulometric detectors differ from amperometric detectors in several ways. The characteristic which makes coulometry more sensitive than amperometry is the geometry of working electrode and the way, and how much the analyte interacts with the electrode. In amperometric detector analytes flow over the working electrode surface and in coulometric electrodes analytes flow between the surfaces of working electrodes. This leads to greater conversion efficiency and more sensitivity of coulometric detectors [10]. Thus resulting in almost 10 times more sensitivity for 3-OMD with coluometric detectors as compared to amperometric detector. This variation in sensitivity limits the estimation of low to normal level of CSF 3-OMD level in our laboratory.

In EQAS results our laboratory was unable to report low level of 3-OMD, 5 HIAA and HVA (Table 1) and thus the z score could not be calculated in many samples. However the clinical interpretation for all the samples was acceptable thus suggesting that our laboratory method is able to diagnose patients with NT disorders even if the level of 5HIAA, HVA and 3-OMD are below the detection limit of the method.

Conclusion

The optimized method can easily be adapted by clinical laboratories for biogenic amine estimation. The test availability at our center is helping many patients to avail this service within India.

Funding

Development committee, National Health and Education Society.