Comparison between dried blood spot and plasma sampling for therapeutic drug monitoring of antiepileptic drugs in children with epilepsy: A step towards home sampling

CLB-09449; No. of pages: 7; 4C: Clinical Biochemistry xxx (2016) xxx–xxx

Contents lists available at ScienceDirect

Clinical Biochemistry journal homepage: www.elsevier.com/locate/clinbiochem

Comparison between dried blood spot and plasma sampling for therapeutic drug monitoring of antiepileptic drugs in children with epilepsy: A step towards home sampling Camilla Linder a,b,⁎, Katarina Wide c,d, Malin Walander c,d, Olof Beck a,b, Lars L Gustafsson a,b, Anton Pohanka a,b a

Department of Laboratory Medicine, Division of Clinical Pharmacology, Karolinska Institutet, Stockholm, Sweden Department of Clinical Pharmacology, Karolinska University Hospital, Stockholm, Sweden c Department of Clinical Science, Technology and Intervention (CLINTEC), Karolinska Institutet, Stockholm, Sweden d Astrid Lindgren Children Hospital, Karolinska University Hospital, Stockholm, Sweden b

a r t i c l e

i n f o

Article history: Received 13 October 2016 Received in revised form 12 December 2016 Accepted 22 December 2016 Available online xxxx Keywords: Red blood cell/plasma ratio Conversion factor Estimated plasma concentrations Hematocrit LC-MS/MS

a b s t r a c t Objectives: To investigate if dried blood spots could be used for therapeutic drug monitoring of the antiepileptic drugs, carbamazepine, lamotrigine and valproic acid in children with epilepsy. Methods: Fingerprick blood samples from 46 children at a neuropediatric outpatient clinic was collected on filterpaper at the same time as capillary plasma sampling. A validated dried blood spot liquid chromatography tandem mass spectrometry method for carbamazepine, lamotrigine and valproic acid was compared with the routine plasma laboratory methods. Method agreement was evaluated and plasma concentrations were estimated by different conversion approaches. Results: Strong correlation was shown between dried blood spot and plasma concentrations for all three drugs, with R2 values N 0.89. Regression analysis showed a proportional bias with 35% lower dried blood spot concentrations for valproic acid (n = 33) and concentrations were 18% higher for carbamazepine (n = 17). A ratio approach was used to make a conversion from dried blood spots to estimated plasma for these two drugs. Dried blood spot concentrations were directly comparable with plasma for lamotrigine (n = 20). Conclusions: This study supports that dried blood spot concentrations can be used as an alternative to plasma in a children population for three commonly used antiepileptic drugs with the possibility to expand by adding other antiepileptic drugs. Clinical decisions can be made based on converted (carbamazepine, valproic acid) or unconverted (lamotrigine) dried blood spot concentrations. Dried blood spot sampling, in the future taken at home, will simplify an effective therapeutic drug monitoring for this group of patients who often have concomitant disorders and also reduce costs for society. © 2016 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

1. Introduction The most commonly used antiepileptic drugs (AEDs) for the treatment of children with epilepsy in Sweden and elsewhere are carbamazepine (CBZ), lamotrigine (LTG), valproic acid (VPA) and levetiracetam [1,2]. Therapeutic drug monitoring (TDM) can assist in individualizing dosages in pediatric care due to pharmacokinetic variations between Abbreviations: ADD, attention deficit disorder; ADHD, Attention deficit hyperactivity syndrome; AEDs, antiepileptic drugs; b/p, blood-to-plasma ratio; CBZ, carbamazepine; CI, confidence interval; DBS, dried blood spots; EMA, European Medicines Agency; HCT, hematocrit; LC-MS/MS, liquid chromatography-tandem mass spectrometry; LTG, lamotrigine; RBC, red blood cell; RBC/p, red blood cell-to-plasma ratio; TDM, therapeutic drug monitoring; VPA, valproic acid. ⁎ Corresponding author at: Department of Clinical Pharmacology C1:68, Karolinska University Hospital, SE-141 86 Stockholm, Sweden. E-mail address: [email protected] (C. Linder).

children on antiepileptic drug polytherapy, drug-drug interactions, subtle adverse drug reactions or co-morbidities in chronically disabled children [3,4]. Presently TDM is usually performed using venous or capillary blood sampling at out-patient clinics, measuring the trough drug concentrations in plasma before the morning dose [3]. Blood sampling at a clinic is time-consuming and sometimes stressful, especially for children with chronic diseases or disabilities. Self-collection at home of a capillary dried blood spot (DBS) sample should be of advantage for many of these patients and their guardians [5,6]. Although DBS has been used in different qualitative assays since the 60's, it is a new matrix for the TDM laboratory [7,8]. In quantitative bioanalysis based on DBS, hematocrit (HCT), spot homogeneity, blood-to-plasma ratio of the drug and sampling a correct volume are factors that may affect the result [9]. There has been a rapid development of assay technologies with much improved sensitivity such as liquid chromatography-tandem

http://dx.doi.org/10.1016/j.clinbiochem.2016.12.008 0009-9120/© 2016 The Canadian Society of Clinical Chemists. Published by Elsevier Inc. All rights reserved.

Please cite this article as: C. Linder, et al., Comparison between dried blood spot and plasma sampling for therapeutic drug monitoring of antiepileptic drugs in children with ep..., Clin Biochem (2016), http://dx.doi.org/10.1016/j.clinbiochem.2016.12.008

2

C. Linder et al. / Clinical Biochemistry xxx (2016) xxx–xxx

mass spectrometry (LC-MS/MS), allowing the measurements of multiple of exogenous small compounds in a drop of blood [10]. Breakthroughs with prelaboratory preparation of microliter whole blood samples to dried plasma and the design of micro-devices for collection of a defined sample volume, makes it likely that DBS will become an attractive sampling technique for quantitative analysis [11–14]. Drugs vary in their distribution between the cell fraction and plasma in whole blood, described as the specific blood-to-plasma ratio (b/pratio) or red blood cell/plasma ratio (RBC/p-ratio) of the drug [6,9,15– 17]. Apart from the RBC/p-ratio, which can be concentration dependent, the patients individual HCT should in theory also affect the relation between DBS and plasma concentrations. In order to compare DBS concentrations with traditional plasma concentrations, a conversion algorithm may be required to calculate an estimated plasma value based on the DBS concentration. Different approaches, with experimental data of paired plasma and DBS concentrations, corrected for sample HCT [18] or not corrected for HCT [19,20] have been used to calculate a ratio which is then used as a conversion factor. Theoretical approaches constructing algorithms considering RBC/p ratios and individual patient HCT levels have also been presented [21]. VPA is mainly distributed in plasma and yields lower concentrations in whole blood, e.g. using DBS [20–22]. In CBZ and LTG, the distribution between whole blood and plasma is more equal since the RBC/p ratio is close to one and direct comparisons have been proposed for these drugs [21,23–26]. Clinical validations are needed to introduce DBS as an alternative matrix in routine TDM [27]. We present an approach on how TDM decisions can be made with DBS concentrations. In this study on children with epilepsy and concomitant neurological diagnoses, we evaluated capillary DBS concentrations of three commonly used AEDs with concentrations of simultaneously collected capillary plasma. DBS concentrations were analyzed using a recently developed liquid chromatography tandem mass spectrometry (LC-MS/MS) method while plasma concentrations from capillary blood were analyzed with the routine laboratory methods. We also investigated if conversion factors were needed to calculate estimated plasma concentrations from DBS concentrations for the use in clinical practice TDM. 2. Material & method 2.1. Patients This study was approved by the local research ethics committee (Regionala etikprövningsnämnden EPN at Karolinska Institutet, Stockholm, 2012/2146-3) and the work conducted in accordance with the Declaration of Helsinki. Patients and/or their guardians approved participation in the study by informed consent prior to blood sampling. Inclusion criteria for the study population were children and adolescents aged 2 to 18 years and treated for epilepsy with CBZ, LTG or VPA as a single or combined drug therapy at Department of Neuropediatrics, Karolinska University Hospital, Huddinge. 2.2. Sample collection Samples were collected by pediatric nurses at the neuropediatric clinic from April 2013 to March 2014. DBS samples were collected at the same occasion as routine TDM samples for plasma and did not require extra visits to the clinic or additional fingerpricks. All samples reflected trough drug levels. Time for sampling, reported prescriptions of antiepileptic drug and last intake of drug were recorded. Fingerprick blood (three to five drops) was collected on filterpaper (Whatman 903 protein saver card, GE Healthcare, Westborough, MA) using a Microtainer Lansett, 1.8 or 2.0 (Becton, Dickinson and Company, Franklin Lakes, NJ). Guidelines from CLSI, Procedures and Devices for the Collection of Diagnostic Capillary Blood Specimens, was followed [28]. Hands were always warm before sampling and first drop of blood was wiped away. From the same fingerprick, approximately 300–500 μL of

capillary blood was collected in a Li-heparin capillary tube for routine plasma analysis. An additional 300–500 μL was collected in a capillary tube for HCT measurement (K2 EDTA Microvette, both tubes from Sarstedts, Nümbrecht, Germany). HCT was analyzed on a SYSMEX XE-5000 (Sysmex, Kobe, Japan), Department of Clinical Chemistry, Karolinska University Hospital, Huddinge. Whatman cards were left drying at room temperature for at least 3 h and then stored in zip-lock bags (Joka 11-68, VWR, Radnor, PA) kept at 4 °C with desiccant packages (Millipore, Darmstadt, Germany) until analysis. On arrival in the laboratory DBS samples were visually inspected and blood spots with a diameter of b7 mm, corresponding to an approximate volume of b15 μL were excluded since they were not within validation ranges [29]. 2.3. Analytical methods CBZ, LTG and VPA concentrations in DBS samples were measured using a recently developed LC-MS/MS method [29]. The method could be applied in the hematocrit range 0.30–0.60 in CBZ and LTG, VPA 0.35–0.60 and for volumes between 15 and 50 μL. In the analysis of plasma concentrations, LTG and CBZ were analyzed by immunochemical methods, QMS LTG and CEDIA CBZ II on an Indiko Plus analyzer (all from Thermo Scientific, Waltham, MA). The methods were performed according to the manufacturer's protocols and kit-inserts. An accredited in-house LC/MS method was used for routine VPA analysis [22]. 2.4. Statistics Capillary DBS concentrations and plasma concentrations were compared using Passing and Bablok regression analysis. No constant bias between the methods was defined as when the 95% confidence interval (CI) of the intercept of the regression line included zero. In analogy, no proportional bias was defined as when the 95% confidence interval for the slope of the regression line included one. Bland-Altman plots were used to identify outliers or tendencies. Criteria for cross validation from European Medicines Agency (EMA) guidelines on bioanalytical method validation (≥67% of the samples should have a difference within ±20% of the mean) [30], were applied on unconverted DBS concentrations. Drugs that needed conversion were evaluated with these criteria a second time, after being converted to estimated plasma concentrations. All calculations, analyses and figures were made using Microsoft Excel 2013 and Addinsoft XLSTAT, 2016. 2.5. Conversion of DBS concentrations to estimated plasma concentrations Two different approaches were used to calculate estimated plasma concentrations based on DBS concentrations. The first approach was to use the average ratio between measured plasma concentrations and DBS concentrations and multiply the DBS concentration with this ratio to achieve an estimated Cplasma. This simple ratio-approach has been used in earlier TDM comparison studies [19,20]. The second was partly a theoretical approach taking into account RBC/plasma ratio (= K in the equation below) and patient HCT and then converting it based on linear regression (21). The following equation was used to create a theoretical plasma concentration. Theoretical C plasma ¼ CDBS =½1−HCTð1−KÞ

ð1Þ

CDBS is the drug concentration in DBS, HCT is the individual hematocrit value and K is the specific drug RBC concentration/plasma concentration ratio derived from in-house in vitro tests, see S3. In calculations, the individual HCT was also replaced with the mean HCT of the patient group for evaluation of individual HCT measurement necessity.

Please cite this article as: C. Linder, et al., Comparison between dried blood spot and plasma sampling for therapeutic drug monitoring of antiepileptic drugs in children with ep..., Clin Biochem (2016), http://dx.doi.org/10.1016/j.clinbiochem.2016.12.008

C. Linder et al. / Clinical Biochemistry xxx (2016) xxx–xxx

This theoretical plasma value was used to plot theoretical plasma concentrations (x-values) versus measured plasma concentrations (y-values). The linear regression equation was used to calculate an estimated theoretical plasma concentration from the DBS concentration. Estimated Theoretical C plasma ¼ ðm  CDBS =½1−ðHCT  KÞÞ þ c

ð2Þ

where “m” is the slope and “c” is the intercept from the regression of theoretical plasma concentration plotted against plasma concentrations, CDBS is the concentration in DBS, HCT is the individual hematocrit value and K is the specific drug RBC/p ratio derived from in vitro experiments (S3). The ratio-approach and the theoretical approach of calculating estimated plasma concentrations from DBS concentrations were compared and evaluated. If both approaches showed similar results, the ratio approach was chosen.

2.6. Clinical evaluation of DBS versus plasma concentrations Based on results of suggested conversion to estimated plasma concentrations or no conversion, the recommended approach for evaluating DBS concentrations were used in the comparison (Section 2.5). Every single DBS sample concentration was compared to the plasma concentration. Each concentration was categorized into one of three groups depending on recommended therapeutic ranges for the respective drugs [3]. The categories were; risk of adverse effect, within therapeutic range and low. The clinical implications of each concentration were assessed by the responsible pediatric neurologist.

3. Results 3.1. Patient characteristics Forty-six neuropediatric patients aged 2 to 18 years old with a mean age of 9 years on CBZ, LTG or VPA treatment or on combination therapy, were included, see Table 1 and S1 for epilepsy characteristics. In the patient group 67% of the patients had concomitant diagnoses such as cerebral palsy and mental retardation, autism and mental retardation, ADHD/ADD (S2). Collected samples from these patients are summarized in Table 2.

3.2. Sample collection A total of 68 paired DBS-plasma samples from 46 patients were collected, generating in total 83 pairs of concentrations, due to that several patients were on polytherapy. 67 samples for HCT determinations were collected. One VPA sample was excluded from the study due to lack of preanalytical documentation. Spots with small volumes (b 15 μL) were found in 12 samples and were not included in the results since they were outside the criteria for the bioanalytical method. Samples lacking HCT values were included in the results and an average HCT was applied for these samples in calculations, Table 2.

3

3.3. Plasma and DBS comparisons with Passing and Bablok regression and Bland-Altman plots Passing and Bablok regression were constructed for regression analysis between plasma and DBS concentrations, Fig. 1. DBS concentrations showed strong correlations (N0.89) compared to capillary plasma concentrations (Fig. 1). Passing and Bablok analysis revealed a proportional bias for VPA since the slope was 0.67 (95% CI 0.58 to 0.76) and the CI did not include 1. Also for CBZ a small proportional bias could be seen since the slope was 1.32 and the 95% CI was 1.01 to 1.45, just outside the limits of including 1, see Fig. 1. This proportional bias suggests the need to use a conversion factor in VPA and CBZ. No proportional bias was detected in LTG. No constant bias was found for any of the drugs. Bland-Altman plots for VPA and CBZ were analyzed with the ratio approach converted concentrations, see Fig. 2. The differences in VPA were evenly distributed on both sides of the mean with a tendency of higher scatter in the upper concentration range. Also in CBZ the differences were evenly spread except for one patient with a concentration of 7.1 μg/mL measured in plasma and 5.5 μg/mL with DBS, Fig. 2. LTG concentrations were not corrected and the plot revealed some uneven distribution around the mean. LTG mean concentrations were 6% higher for DBS compared to plasma. One patient outside the 95% CI had a concentration of 17.2 μg/mL measured in plasma and 14.9 μg/mL in DBS, Fig. 2. 3.4. Results using different conversion approaches In-house in vitro estimation of RBC/plasma ratio was made at three different HCT levels, 0.30, 0.43 and 0.55 L/L. This experiment was performed to cover possible variations in RBC/plasma ratio for these drugs at different concentrations and at different HCT levels. RBC/plasma ratio (K-values), was found to vary depending on concentration and HCT which made it complicated to choose a K-value for the theoretical approach. Average K-values derived from three concentration ranges at HCT 0.43 were used in the Eqs.s (1) and (2) (theoretical approach). Results from the experiment are shown in S3. RBC/plasma ratios at HCT 0.43 L/L in CBZ was 0.88, in LTG 1.15 and in VPA 0.06, see S3. CBZ DBS concentrations were on average 18% higher than in plasma (unconverted Bland-Altman plots, S5). For CBZ, conversion with a factor of 0.84 from the ratio-approach (range 0.66–1.06) 82% of the samples were within the ± 20% EMA limit. Conversions by the theoretical approach also reached 82% within ±20% (S4, S6, S7). The mean LTG ratio between plasma and DBS concentrations was close to 1 (0.96) implying that a correction factor was not needed. A total of 60% of unconverted LTG concentrations were within the ±20% limit, i.e. the EMA method comparison criterion was not accepted for LTG, supplementary S7. Samples in the lower measuring range had a high scatter which decreased when reanalyzing plasma samples with an LC-MS/MS method for LTG in plasma. Only 18 samples were compared with the LC-MS/MS method due to shortage of plasma for two samples. In this comparison 78% of the samples were within the ±20%, supplementary S7. LTG corrected with the theoretical approach resulted in 75% of the samples within ±20%, but the conversion made the bias higher at lower concentrations, (Passing & Bablok and BlandAltman plots theoretical approach in S4, S6 and data from comparison in S7).

Table 1 Patient characteristics in study population (n = 46). Drug therapy for epilepsy Patient characteristics

Valproic acid (VPA)

Carbamazepine (CBZ)

Lamotrigine (LTG)

Combinations

Female/male (n = 17/29)

6/10

4/12

6/1

Age (mean, range)

9 (3–18)

9 (2–17)

13 (1–18)

VPA VPA VPA VPA

+ + + +

CBZ: 0/1 LTG: 1/5 CBZ: 13 LTG: 9 (2–17)

Please cite this article as: C. Linder, et al., Comparison between dried blood spot and plasma sampling for therapeutic drug monitoring of antiepileptic drugs in children with ep..., Clin Biochem (2016), http://dx.doi.org/10.1016/j.clinbiochem.2016.12.008

4

C. Linder et al. / Clinical Biochemistry xxx (2016) xxx–xxx

Table 2 Description of collected samples from study population (n = 46).

Number of collected samples (n = 83)a Excluded, sample volume too small (n = 13) Number analyzed (n = 70) Number of HCT values Hematocrit (%, mean and range) Plasma levels (μg/mL, mean and range) DBS levels (μg/mL, mean and range) Estimated plasma levels (μg/mL, mean and range)c a b c

Valproic acid

Carbamazepine

Lamotrigine

42 9b 33 33 37 (32–42) 73 (27–111) 47 (17–71) 74 (27–113)

20 3 17 16 37 (35–42) 5.7 (1.9–10.4) 6.9 (2.8–13.6) 5.7(2.3–11.3)

21 1 20 18 37 (34–40) 6.7 (1.5–17.2) 7.4 (1.1–15.0) Not corrected

11 patients of the total 46 contributed with several samples throughout the study period. One VPA sample was excluded due to lack of documentation. DBS concentrations were multiplied with an average ratio between measured plasma concentrations and DBS concentrations.

Uncorrected VPA DBS concentrations were on average 35% lower than plasma concentrations, showed as Bland-Altman plot in supplementary S5. With the ratio-approach, a conversion factor of 1.58 (range 1.27–2.16) was calculated, resulting in 97% of estimated VPA within ±20% in the cross validation (comparison data in S7). With the use of the theoretical approach, a conversion including the HCT and the drug RBC/p ratio with a K of 0.06 in the equation, the same 97% within ±20% was reached in the EMA cross validation, see supplementary S3, S4, S6 and S7. 3.5. Individual HCT measurements versus mean HCT Estimated theoretical plasma concentrations calculated with a mean HCT from the population (0.37) compared to theoretical plasma concentrations calculated with individual HCT showed no significant differences for CBZ and LTG, the biggest difference was 2.2%, see S7. For VPA differences in HCT had greater impact and the biggest difference was −11.2% resulting in an estimated concentration (theoretical approach) of 32.6 μg/mL with individual HCT and 36.5 μg/mL with the average HCT. This difference did not affect the clinical decision, Supplementary S7. 3.6. Clinical evaluation of DBS versus plasma concentrations Unconverted LTG DBS concentrations and ratio-approach converted CBZ and VPA DBS concentrations were used for clinical evaluation. Every single converted (CBZ and VPA) or unconverted (LTG) DBS concentration as well as every single plasma concentration were put into three different categories, risk of adverse effects, within therapeutic

range and low, S8. Recommended therapeutic ranges from Patsalos et al. [3] were 4–12 μg/mL for CBZ, 1–13 μg/mL for LTG, and 50–100 μg/mL for VPA. For LTG the decision was to use a smaller recommended range of 2–10 μg/mL as concentrations below 2 μg/mL for LTG may indicate compliance problem or being in risk of low or no effect. The upper limit of 13 μg/mL was lowered to 10 μg/mL due to scatter in the comparison of plasma and DBS concentrations to be sure of not missing any patient who might be in risk of adverse effects for TDM decisions from DBS. Nine pairs of samples differed in the way they were categorized, see Table 3 and S8. Clinical decisions made by the pediatric neurologist on drug treatment from these concentrations were the same as clinical decisions based on the plasma concentrations since the absolute differences between these samples were small, see S8. 4. Discussion 4.1. Sample collection and bioanalysis Small volume spots may influence accuracy according to the validated analytical method [29] and was found to be a significant problem since 12 of 83 samples generated small volume spots. This stresses the importance of training and education for patients and their guardians as well as nurses when applying DBS sampling at home or at the clinic [10,31]. In relation to this, the project also included the production of text-based and audiovisual learning material that is now used in part two of this study where guardians are collecting samples from their children [32]. A future improvement to avoid sampling errors is to use

Fig. 1. Scatter plots with Passing and Bablok fit. Correlations of plasma measured with routine method and DBS measured with an LC-MS/MS method for lamotrigine, carbamazepine and valproic acid. Dotted lines are identity lines, continuous lines representing Passing & Bablok regression. R2 values from simple linear regression and Passing & Bablok equation for carbamazepine: R2 = 0.892 (n = 17) y = 1.32x − 0.44 with a 95% CI for slope: 1.01 to 1.45,intercept: −1.29 to 0.78, lamotrigine: R2 = 0.978 (n = 20) y = 1.18x − 0.39 with a 95% CI for slope: 0.32 to 1.32, intercept: −1.01 to 0.96 for valproic acid R2 = 0.899 (n = 33) y = 0.67x − 1. 55 with a 95% CI for slope: 0.58 to 0.76, intercept: −7.70 to 4.16.

Please cite this article as: C. Linder, et al., Comparison between dried blood spot and plasma sampling for therapeutic drug monitoring of antiepileptic drugs in children with ep..., Clin Biochem (2016), http://dx.doi.org/10.1016/j.clinbiochem.2016.12.008

C. Linder et al. / Clinical Biochemistry xxx (2016) xxx–xxx

5

Fig. 2. Bland-Altman plots. Bland-Altman plots for carbamazepine and valproic acid; x-axis shows the mean concentrations of the two methods while y-axis shows differences between ratio approach estimated plasma concentrations from DBS and analyzed plasma concentrations. For lamotrigine unconverted DBS and plasma concentrations are used. Blue bold line representing the mean difference between methods, blue dotted lines are 95% CI of the mean and red dotted lines are 95% CI. Mean bias for CBZ is 0.087 and SD is 0.68 CI −1.25 to 1.42. Mean bias for LTG is 0.47 and SD is 1.29 95% CI −2.06 to 3.00. Mean bias for VPA is −0.001 and SD is 8.57 95% CI −16.8 to 16.8. (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of this article.)

filter paper devices where the volume is automatically measured [11, 12]. Multiple reaction monitoring LC-MS/MS analysis makes it possible to add more drugs to the method so that several AEDs can be included. Levetiracetam has now been added to the validated method by Linder et al. [29]. 4.2. Conversion of DBS to estimated plasma values For CBZ the slope of 1.32 indicated a conversion factor and it was needed to fulfill EMA criteria. Published methods on clinical validation of CBZ reported divergent results when comparing plasma and DBS concentrations. One study used fingerprick capillary samples and presented a slope of 1.32 which is the same as in this study [25]. Other studies compared DBS samples taken from venous blood and test tubes, or it was not clear how the DBS were sampled. No significant difference between plasma and DBS samples (slopes of 0.97 to 1.13) has also been reported [21,26]. The RBC/plasma ratio in capillary blood might be slightly different than in venous blood, resulting in higher CBZ concentrations in capillary DBS than with venous DBS. For this reason it is important to investigate the relationship between capillary blood and plasma since capillary blood is collected in home sampling. 17 samples are not enough to conclude a difference between DBS capillary and venous samples and more data will be gathered in the ongoing part two of this study to evaluate this finding. The two different methods of conversion showed similar results; hence the ratio approach was used. Since venous blood collection is the traditional way of collecting samples from adults it is also important to study the relationship between venous DBS and capillary DBS from adults before DBS can be offered as home sampling for these patients. A conversion factor in VPA was derived from the ratio between measured plasma and DBS concentrations (n = 33). We used the ratio approach with a correction factor of 1.58 for these patients since it was comparable with results from theoretical approach. Conversion factors of 1.88 and 1.96 have been reported by others [20,21]. For LTG, the Table 3 Comparison of plasma and DBS concentrations in relation to therapeutic ranges. CBZ n = 17

LTG n = 20

VPA n = 33

Categories

Plasma

DBS

Plasma

DBS

Plasma

DBS

Low Within range Risk of adverse effects Different categories

3 14 0 1

2 15 0

4 13 3 2

4 11 5

4 25 4 6

6 21 6

decision to keep the concentrations unconverted was made since the regression analysis did not show proportional bias. Also, the theoretical approach resulted in larger % differences for some samples, see S7. Replacing the individual HCT with a mean HCT in the theoretical approach did not affect the clinical decision for CBZ, LTG or VPA. The recommendation is to use the ratio approach for conversion to estimated plasma concentrations, where no HCT is included in the conversion. However, extreme individual HCT will generate bias on measured concentrations. The conclusion is that patients with expected HCT within bioanalytical validated ranges could be suitable for home sampling. A way to control that the HCT is within validated ranges is to measure it from the same DBS [33,34]. 4.3. Clinical application of DBS in TDM in children Earlier comparisons between plasma and DBS for these drugs were often based on venous and not capillary DBS samples and do not propose recommendations for the use of DBS concentrations in TDM [25, 26]. Our aim is to suggest how the TDM laboratory can make decisions when both plasma and DBS concentrations are used. An important advantage with this study is that it is performed on capillary DBS, which is the true matrix for home sampling. Despite the matrix differences in DBS compared to plasma the clinical decision can be made directly from DBS (LTG) or after conversion to estimated plasma concentrations (CBZ, VPA) and used in the future as an alternative to traditional sampling for these children. The result from the cross validation showed that there were concentrations that differed more than ±20% and in some individuals, especially for LTG, even N 30% difference (S7). For TDM of antiepileptic drugs the ranges are recommended ranges. Every patient should have an individual dose that is suitable depending on type of epilepsy and combinations of other drugs. If a plasma or DBS concentration will be in a range around the critical concentration where there is risk of adverse effects, the clinician will always contact the patient for an evaluation and health status. Unexpected deviations from earlier concentration levels or levels being close to be at risk of serious adverse effects should always result in considering a new sample for verification, and in the case of DBS, a new sample in plasma is recommended. We conclude that the differences noticed for CBZ and VPA are acceptable from a clinical point of view. The differences in LTG are large in some of the compared concentrations and the decision is to be careful with LTG and all concentrations above 10 μg/mL should be considered as being in risk of adverse effects. When each pair of DBS concentrations and plasma concentrations were evaluated no patient were identified

Please cite this article as: C. Linder, et al., Comparison between dried blood spot and plasma sampling for therapeutic drug monitoring of antiepileptic drugs in children with ep..., Clin Biochem (2016), http://dx.doi.org/10.1016/j.clinbiochem.2016.12.008

6

C. Linder et al. / Clinical Biochemistry xxx (2016) xxx–xxx

to have a different recommendation of dose adjustment, even though some fell into different categories. On the other hand DBS is a new matrix and more data is needed to be sure of how much DBS concentrations may differ from plasma concentrations when using a traditional filter paper for collection. Although the population studied is small, we consider it large enough to go on with the next part of the study where guardians are educated to collect DBS samples from their children or older children collect the samples themselves. The data gathered will be added to these DBS data and in this way enlarge the study population. Preliminary results show that guardians can collect qualitative DBS samples from their children. Using DBS in a home based setting may facilitate for children who also have concomitant neurological and/or cognitive disabilities and their guardians The calculation of theoretical plasma concentrations aims at applying therapeutic ranges and decision limits based on plasma concentrations on DBS data. Still, there is an added uncertainty in treating the calculated theoretical plasma values as normal plasma values in a TDM context. The obvious alternative to DBS estimated plasma concentrations from DBS would be to construct therapeutic ranges and decision limits for uncorrected DBS data instead of taking the detour over plasma concentrations. This could be done but calls for much more DBS data than is present at this stage as the power of clinical data from plasma over DBS is indisputable. The need for a straightforward evaluation and decision making in the clinical situation also supports the notion of translatability between plasma and DBS concentrations as the main clinical scenario in the future is that plasma and DBS sampling will be used alternately in monitoring the individual patient. After finalization of the second part of the study the ratio approach with estimated plasma concentrations for CBZ, VPA and for LTG unconverted DBS concentrations is suggested to be used in TDM for this patient group. It is of importance that DBS home sampling will be offered as a service to this patient group and that there is a possibility to choose traditional collection at a clinic. DBS collection will not be the most convenient solution for all patients. 5. Conclusion DBS has a great potential as a valuable tool in the treatment and evaluation of epilepsy in children especially among children with disabilities, for whom a visit to clinic is an effort. This clinical validation shows that the time is right to go on and investigate whether it is feasible for this patient group to collect DBS samples in their home environment. The need to monitor other AEDs can be met by expanding the method. With careful TDM of these patients, DBS is suggested to be used as an alternative to the traditional venous or capillary sampling at a clinic for patients with normal HCT and no signs of other severe problematic interactions between drugs. Funding This study was supported by grants from Swedish Research Council (VR-2011/3440) and from Stockholm Healthcare Region (ALF/SLL 2013/ 0398). Disclosure of conflict of interests None of authors has any conflict of interest. Acknowledgements The authors wish to thank research nurse Mirja Neideman who participated in the research and LicMed, Clinical Laboratory Scientist Gerd Ackehed for bioanalytical expertise.

Appendix A. Supplementary data Supplementary data to this article can be found online at http://dx. doi.org/10.1016/j.clinbiochem.2016.12.008.

References [1] R. Ackers, M.L. Murray, F.M.C. Besag, I.C.K. Wong, Prioritizing children's medicines for research: a pharmacoepidemiological study of antiepileptic drugs, Br. J. Clin. Pharmacol. 63 (6) (2007) 689–697. [2] P. Mattsson, T. Tomson, K. Edebol Eeg-Olofsson, L. Brännström, G. Ringbäck Weitoft, Association between sociodemographic status and antiepileptic drug prescriptions in children with epilepsy, Epilepsia 53 (12) (Dec 2012) 2149–2155 Available from: http://www.ncbi.nlm.nih.gov/pubmed/23061699. [3] P.N. Patsalos, D.J. Berry, B.F.D. Bourgeois, J.C. Cloyd, T.A. Glauser, S.I. Johannessen, et al., Antiepileptic drugs – best practice guidelines for therapeutic drug monitoring: a position paper by the subcommission on therapeutic drug monitoring, ILAE Commission on Therapeutic Strategies, Epilepsia 49 (7) (Jul 2008) 1239–1276 Available from: http://www.ncbi.nlm.nih.gov/pubmed/18397299. [4] A. Verrotti, G. Loiacono, G. Coppola, A. Spalice, A. Mohn, F. Chiarelli, Pharmacotherapy for children and adolescents with epilepsy, Expert. Opin. Pharmacother. 12 (2) (Feb 2011) 175–194 Available from: http://www.ncbi.nlm.nih.gov/pubmed/ 21208135. [5] G. Lawson, S. Tanna, Self-sampling and quantitative analysis of DBS: can it shift the balance in over-burdened healthcare systems? Bioanalysis 7 (16) (2015) 1963–1966. [6] A.J. Wilhelm, J.C.G. den Burger, E.L. Swart, Therapeutic Drug Monitoring by Dried Blood Spot: Progress to Date and Future Directions, Clin. Pharmacokinet. 53 (11) (Sep 10 2014) 961–973 Available from: http://www.ncbi.nlm.nih.gov/pubmed/ 25204403. [7] Li W, Tse FLS. Dried blood spot sampling in combination with LC-MS/MS for quantitative analysis of small molecules. Biomed. Chromatogr. John Wiley and Sons Ltd (Southern Gate, Chichester, West Sussex PO19 8SQ, United Kingdom); Jan 2010; 24(1):49–65. Available from: http://www.interscience.wiley.com/cgi-bin/fulltext/ 123215196/PDFSTART\nhttp://ovidsp.ovid.com/ovidweb.cgi?T= JS&PAGE= reference&D=emed9&NEWS=N&AN=2010068470 [8] P.A. Demirev, Dried blood spots: analysis and applications, Anal. Chem. 85 (2013) 779–789. [9] A. Sharma, S. Jaiswal, M. Shukla, J. Lal, Dried blood spots: concepts, present status, and future perspectives in bioanalysis, Drug Test. Anal. 6 (5) (May 2014) 399–414 Available from. (http://www.ncbi.nlm.nih.gov/pubmed/24692095). [10] M. Wagner, D. Tonoli, E. Varesio, G. Hopfgartner, The use of mass spectrometry to analyze dried blood spots, Mass Spectrom. Rev. (Sep 22 2014) Available from: http://www.ncbi.nlm.nih.gov/pubmed/25252132. [11] P. Denniff, N. Spooner, Volumetric absorptive microsampling: a dried sample collection technique for quantitative bioanalysis, Anal. Chem. 86 (2014) 8489–8495. [12] G. Lenk, S. Sandkvist, A. Pohanka, G. Stemme, O. Beck, N. Roxhed, A disposable sampling device to collect volume-measured DBS directly from a fingerprick onto DBS paper, Bioanalysis 7 (16) (2015) 2085–2094. [13] L.A. Leuthold, O. Heudi, J. Déglon, M. Raccuglia, M. Augsburger, F. Picard, et al., New microfluidic-based sampling procedure for overcoming the hematocrit problem associated with dried blood spot analysis, Anal. Chem. 87 (4) (2015) 2068–2071. [14] R. Sturm, J. Henion, R. Abbott, P. Wang, Novel membrane devices and their potential utility in blood sample collection prior to analysis of dried plasma spots, Bioanalysis 7 (16) (2015) 1987–2002. [15] G. Emmons, M. Rowland, Pharmacokinetic considerations as to when to use dried blood spot sampling, Bioanalysis 2 (11) (2010) 1791–1796. [16] P.H. Hinderling, Red blood cells: a neglected compartment in pharmacokinetics and pharmacodynamics, Pharmacol. Rev. 49 (3) (1997) 279–295. [17] P. Paixão, L.F. Gouveia, M. JAG, Prediction of drug distribution within blood, Eur. J. Pharm. Sci. 36 (4–5) (2009) 544–554. [18] M.V. Antunes, S. Raymundo, V. de Oliveira, D.E. Staudt, G. Gössling, G.P. Peteffi, et al., Ultra-high performance liquid chromatography tandem mass spectrometric method for the determination of tamoxifen, N-desmethyltamoxifen, 4-hydroxytamoxifen and endoxifen in dried blood spots, Talanta 132 (2015) 775–784. [19] C.M. Nijenhuis, A.D.R. Huitema, S. Marchetti, C. Blank, J.B.A.G. Haanen, J.V. van Thienen, et al., The use of dried blood spots for pharmacokinetic monitoring of i treatment in melanoma patients, J. Clin. Pharmacol. (2016) 1–6. [20] L. Rhoden, M.V. Antunes, P. Hidalgo, C.Á.D. Silva, R. Linden, Simple procedure for determination of valproic acid in dried blood spots by gas chromatography–mass spectrometry, J. Pharm. Biomed. Anal. 96C (Apr 2 2014) 207–212 Elsevier B.V. Available from: http://www.ncbi.nlm.nih.gov/pubmed/24769300. [21] S.T. Kong, S.-H. Lim, W.B. Lee, P.K. Kumar, H.Y.S. Wang, Y.L.S. Ng, et al., Clinical validation and implications of dried blood spot sampling of carbamazepine, valproic acid and phenytoin in patients with epilepsy, PLoS One 9 (9) (Jan 2014), e108190. Available from: http://www.ncbi.nlm.nih.gov/pubmed/25255292. [22] A. Pohanka, M. Mahindi, M. Masquelier, L.L. Gustafsson, O. Beck, Quantification of valproic acid in dried blood spots, Scand. J. Clin. Lab. Invest. 74 (7) (Jul 25 2014) 648–652 Available from: http://www.ncbi.nlm.nih.gov/pubmed/25059925. [23] S. AbuRuz, M. Al-Ghazawi, Y. Al-Hiari, A simple dried blood spot assay for therapeutic drug monitoring of lamotrigine, Chromatographia 71 (11−12) (Apr 11 2010) 1093–1099 Available from: http://www.springerlink.com/index/10.1365/s10337010-1569-y.

Please cite this article as: C. Linder, et al., Comparison between dried blood spot and plasma sampling for therapeutic drug monitoring of antiepileptic drugs in children with ep..., Clin Biochem (2016), http://dx.doi.org/10.1016/j.clinbiochem.2016.12.008

C. Linder et al. / Clinical Biochemistry xxx (2016) xxx–xxx [24] N.M. Shah, A.F. Hawwa, J.S. Millership, P.S. Collier, J.C. Mcelnay, A simple bioanalytical method for the quantification of antiepileptic drugs in dried blood spots, J. Chromatogr. B 923–924 (2013) 65–73 Elsevier B.V. Available from: http:// dx.doi.org/10.1016/j.jchromb.2013.02.005. [25] E. Shokry, F. Villanelli, S. Malvagia, A. Rosati, G. Forni, S. Funghini, et al., Therapeutic drug monitoring of carbamazepine and its metabolite in children from dried blood spots using liquid chromatography and tandem mass spectrometry, J. Pharm. Biomed. Anal. 109 (2015) 164–170 Elsevier B.V. Available from: http://dx.doi.org/ 10.1016/j.jpba.2015.02.045. [26] G. Martins, S. De Lima, R.Z. Hahn, C. Rama, L. Rhoden, P. Hidalgo, et al., Determinação Simultânea de Carbamazepina, fenitoína e fenobarbital em sangue seco em papel por cromatografia líquida de alta eficiência, Quim Nova 37 (6) (2014) 1067–1071. [27] D. Milosheska, I. Grabnar, T. Vovk, Dried blood spots for monitoring and individualization of antiepileptic drug treatment, Eur. J. Pharm. Sci. 75 (2015) 25–39. [28] CLSI, Procedures and Devices for the Collection of Diagnostic Capillary Blood Specimens; Approved Standard, sixth ed., CLSI document H04-A6, 28, Clin Lab Stand Inst Doc, 2008 25 (28, no 25):viii. Available from: http://scholar.google.com/scholar? hl=en&btnG=Search&q=intitle: Procedures+and+devices+for+the+collection+of+diagnostic+capillary+blood+specimens+:+approved+standard#1. [29] C. Linder, M. Andersson, K. Wide, O. Beck, A. Pohanka, A LC–MS/MS method for therapeutic drug monitoring of carbamazepine, lamotrigine and valproic acid in DBS, Bioanalysis 7 (16) (2015) 2031–2039. [30] European Medicines Agency. Guideline on bioanalytical method validation. EMEA/ CHMP/EWP/192217/2009 Rev.1 Corr. 2011. Available from: www.ema.europa.eu [31] N. Yonan, R. Martyszczuk, A. Machaal, A. Baynes, B.G. Keevil, Monitoring of cyclosporine levels in transplant recipients using self-administered fingerprick sampling,

7

Clin. Transpl. 20 (2006) 221–225 Available from: http://www.ncbi.nlm.nih.gov/ pubmed/16640530. [32] Linder C, Gambell Barroso M. Blodprovtagning på filterpapper, Barn|Läkemedel. [cited Jun 2 2016]. Available from: https://www.youtube.com/watch?v= LpJeyfoWsgA [33] S. Capiau, V.V. Stove, W.E. Lambert, C.P. Stove, Prediction of the hematocrit of dried blood spots via potassium measurement on a routine clinical chemistry analyzer, Anal. Chem. 85 (1) (2013) 404–410. [34] S. Capiau, L.S. Wilk, M.C.G. Aalders, C.P. Stove, A novel, nondestructive, dried blood spot-based hematocrit prediction method using noncontact diffuse reflectance spectroscopy, Anal. Chem. 88 (12) (Jun 21 2016) 6538–6546 Available from: http://www.ncbi.nlm.nih.gov/pubmed/27206105.

Glossary DBS, dried blood spot sampling = collection, transport and storage of capillary blood on cellulose-based filter papers.: TDM, therapeutic drug monitoring = analysis and evaluation of drug concentrations for optimization of individual patient drug treatment, for which clinical experience or clinical trials have shown improved outcome.: HCT, hematocrit = the proportion of cells in whole blood by volume.: Red blood cell/plasma ratio = concentration of the test compound in red blood cells compared with plasma.:

Please cite this article as: C. Linder, et al., Comparison between dried blood spot and plasma sampling for therapeutic drug monitoring of antiepileptic drugs in children with ep..., Clin Biochem (2016), http://dx.doi.org/10.1016/j.clinbiochem.2016.12.008