Busulfan (Bu) is a bifunctional alkylating agent currently used as a component of the conditioning regimen before pediatric hematopoietic stem cell transplantation (HSCT). A high interindividual pharmacokinetic variability is generally observed during intravenous (IV) Bu treatment, specifically in children,1,2 and this, together with its narrow therapeutic window, necessitates its therapeutic drug monitoring (TDM).3–5 Several studies indicated that over half of the pediatric patients undergoing Bu treatment had off-target steady-state concentration levels, which were improved (brought within target range) by applying TDM during treatment.6–8
Several analytical methods (for Bu quantitative analyses) have been described, including immunoassays,9 liquid chromatography coupled with ultraviolet10 or fluorescence detectors,11 and gas chromatography with mass spectrometry.12 The use of liquid-chromatography tandem-mass spectrometry (LC-MS/MS) offers the high specificity and sensitivity required for Bu TDM. Given the therapy's short duration, faster turnaround times are important to achieving the target window as early as possible.
Finding alternative methods for decreasing the blood volumes required for TDM in pediatric populations also seems desirable. An alternative approach for quantifying blood Bu would be applying the dried blood spot (DBS) method, which requires a lower blood volume (mL), is fast, has easy sample preparation and rapid turnaround times. We previously reported a validated DBS method for quantifying Bu with good trueness (relative SD <14%), precision (relative SD <10%), and 100% recovery.13 A more recent study conducted with 10 adult patients indicated good correlation between the plasma and DBS methods for quantifying Bu.14 However, considering that several factors may affect the differences between plasma- and DBS-based Bu quantification, further DBS method validation is desirable before its implementation in pediatric clinical practice.15 Electrophilic compounds such as Bu get partitioned into red blood cells (RBCs) easily. About 46.9% of Bu binds irreversibly to blood cells,16 whereas 7.4%–32.4% of Bu binds to plasma proteins.16,17 Moreover, hematocrit (Hct) changes will change a drug's plasma fraction. In essence, Bu is equally distributed to the cellular and plasma compartments. The Hct remains one of the most important parameters affecting the DBS assay's performance because it affects the spot size present in the specific DBS paper, which could produce inconsistent results, not only between individuals, but also between venous and capillary blood.18 Validating DBS sampling against a reference plasma assay is a crucial procedure before introducing a DBS assay into routine clinical practice for Bu TDM.
The aim of this study was, therefore, to evaluate the feasibility of clinically applying the DBS method for pediatric Bu TDM. The impact of the Hct on DBS assay's performance was also evaluated using in vitro and in vivo clinical samples. Furthermore, we attempted to evaluate the influence of the conditioning day on the correlation of Bu levels in DBS and dried plasma spot (DPS) in pediatric patients.
MATERIAL AND METHODS
Chemicals and Reagents
Busulfan and ammonium formate were purchased from Sigma-Aldrich (Steinheim, Germany). Busulfan-d8 from Toronto Research Chemicals Inc (Toronto, Canada) was used as an internal standard (IS). High-performance liquid chromatography-grade methanol and acetonitrile were procured from Merck (Darmstadt, Germany). Formic acid was procured from Honeywell Riedel-de Haën (Hanover, Germany). Methazolamide was purchased from Abcam (ab145585; Cambridge, United Kingdom). In vitro diagnostic Whatman 903 sample collection cards were purchased from GE Healthcare (Dassel, Germany). Autosampler glass vials (0.3 mL capacity) were procured from BGB Analytik SA (Geneva, Switzerland). Six healthy, drug-free, informed volunteers, after consenting to give their blood for use in anonymized research, had it collected in tubes containing heparin at Geneva University Hospitals (Geneva, Switzerland). Bu stock was prepared in methanol at a concentration of 1 mg/mL. Busulfan-d8 (IS) in acetonitrile was used to prepare a stock solution at a concentration of 2.5 mg/mL. The working solutions of IS were obtained by adding the prepared stock solution to methanol to achieve a concentration of 100 ng/mL. Calibrators (standards), working solutions, and quality control samples were prepared as previously described.13
All experiments were performed using a previously validated rapid method developed in our laboratory.13 In brief, the method was linear in the calibration range of 100–2000 ng/mL. This calibration range is sufficient and covers the Bu levels when administered in a 4-times daily dosing schedule. This method was validated with quality controls of 300, 600, and 1400 ng/mL, fulfilling the acceptance limits for trueness and precision (relative SD <15%). No matrix or carryover effects were observed. An API 4000TM triple quadrupole mass spectrometer (AB Sciex, Concord, Canada) controlled using Analyst 1.5.1 software was used for method development and analyses. The MS was operated in the multiple reaction monitoring mode with positive ion electrospray ionization.13 The multiple reaction monitoring transitions were 264.1→151.1 for Bu, and 272.1→159.1 for IS Bu‐D8. Chromatography was performed on a Kinetex C18 analytical column (50 mm × 2.1 mm, 2.6 µm; Phenomenex, Torrance, CA) preceded by a KrudKatcher ultra in‐line filter (0.5 µm). Ammonium formate (10 mM), 0.1% (vol/vol) formic acid, and 5% (v/v) acetonitrile in water were used as mobile phase A, and mobile phase B contained acetonitrile; the mobile phase was run in gradient with a flow rate at 0.5 mL/min and had a total run time of 7 minutes13 Fresh standards were prepared for each run, and quality control samples at 300, 600, and 1400 ng/mL were included.13
Analyses of Blood-to-Plasma Ratio
The blood-to-plasma ratio protocol was obtained using a previously described method.19 The Hct was measured using a Hettich Hematocrit 210 centrifuge (Hettich AG, Bäch, Switzerland) and calculated as the packed cell volume percentage of total whole blood volume. Bu (final concentration, 0.5 μM; final dimethyl sulfoxide concentration, 0.005%) was incubated separately with fresh heparinized whole blood, reference RBC, and reference plasma for 60 minutes at 37°C (n = 3 replicates per incubation). After incubation, the experimental whole blood samples were centrifuged for 5 minutes at 5000g at 4°C. The spiked reference plasma was stored on ice during this period. The spiked reference RBCs were quickly freeze-thawed thrice to assist RBC lysis. After centrifugation of the experimental whole blood samples, aliquots were sampled from the plasma and RBC layers for analysis. Like before, the RBC layer was quickly freeze-thawed thrice to assist RBC lysis. After protein precipitation and centrifugation, the experimental and reference sample supernatants were analyzed by LC-MS/MS. The blood-to-plasma ratio was calculated using the following equation:
where: Kb/p = blood-to-plasma ratio.
where: Ke/p, RBC-to-plasma partition coefficient; IRBC, LC-MS/MS response (peak-area ratio to IS) for the RBC fraction; IRBCREF, LC-MS/MS response (peak-area ratio to IS) for reference RBCs; IPL, LC-MS/MS response (peak-area ratio to IS) for plasma fraction; and IPLREF, LC-MS/MS response (peak-area ratio to IS) for reference plasma.
Impact of Hematocrit on Bu Levels in Whole Blood and Plasma In Vitro
For in vitro analyses, Hct levels that would cover the range expected in patients were prepared. Whole blood (10 mL) was collected from 6 healthy volunteers, and Hct levels were determined. To prepare 3 different Hct levels (Table 1), 3 tubes containing 100 µL of whole blood from each volunteer was diluted with 50, 100, and 150 µL of plasma, respectively. Working solutions of Bu were then added to the blood with different Hct levels to obtain 3 different concentrations (300, 600, and 1400 ng/mL). All the samples were prepared in duplicates. DBS and DPS sampling was performed using calibrated volumetric pipettes. The samples were incubated for 1 hour at 37°C. After DBS sampling in triplicates (5 µL each), samples were centrifuged, and plasma from their respective sample was used to prepare DPS in triplicates (5 µL each). Circular discs of 6 mm for blood and 8 mm for plasma, including the entire spot, were cut and placed into individual autosampler vials of 0.2 mL with a clamp. An IS of 100 µL d8-Bu at 100 ng/mL was added to each vial and sealed to prevent evaporation. The vials were vortexed for 1 minute, and 40 µL of the solution containing IS and the paper were pipetted to the new vials containing 40 µL of milliQ water, sealed, and vortexed. Finally, 10 µL of the aliquots were injected into the LC-MS/MS system.
Geneva University Hospitals' Ethics Committee approved this observational study, and recruitment occurred from December 2013 to January 2019 in the Onco-Hematology Unit, Department of Pediatrics, HUG, Geneva (Approval No: CER-13-225; PB_2017-00198). Fifteen children scheduled for HSCT and about to begin an IV Bu-based conditioning regimen were enrolled after obtaining written informed consent from the patients and/or their legal representatives. Bu was intravenously administered for 2 hours, 4 times per day for 16 doses. The first dose was age-/weight-based, calculated according to hospital guidelines.4 Blood samples were collected at 6 time points before initiation and at 120, 130, 150, 180, and 360 minutes after IV Bu initiation at the first dose level.7
Additional TDM was performed at dose 2 or 3 (day 1) and then from dose 5 (day 2) to dose 9 (day 3), when guidelines suggested that verification of Bu exposure was required. Patients' demographic characteristics and comedication were extracted from their medical files. DBS and DPS samples were prepared on the day of sample collection, and analyzed on the same day or the following days, in a few cases.
The correlation between DBS and DPS Bu levels was evaluated from the in vitro experiments and pediatric cohort using Deming-regression model analyses. The percentage differences between the DBS Bu levels and DPS Bu levels were compared. Criteria for the acceptance of DBS levels were set within ±20% of the DPS concentration based on analytical method and clinical significance acceptance limits. This analytical variability limit is equivalent to European Medical Agency guidelines on bioanalytical method evaluation.20 Acceptable method agreement was considered if the differences between DBS and DPS methods were <±20% of the average difference for more than >67% of the paired samples. Clinical acceptance limit was also set <±20% as area under the curves (AUCs, interpreted from the measured concentrations) divergence below 20% would not result in a significant change in the dose recommendation. The divergence between DBS and plasma (DPS) was calculated using the following formula:14
The Bland–Altman analysis was performed for multiple observations per individual, to calculate the mean difference between DBS and DPS measurements (bias) and the mean difference (±1.96 SD).21 The 95% confidence interval for limits of agreement (LoA) were estimated using the MOVER method, as described earlier.22 A Passing–Bablok regression analysis was performed to test the influence of Hct and day of conditioning, subject to linearity between DBS and DPS Bu levels. Statistical analyses were performed using SPSS Statistics for Windows (version 25.0, IBM, Armonk, NY) and Medcalc Statistical Software (version 16.4.3) from MedCalc Software bvba (Ostend, Belgium; https://www.medcalc.org; 2016).
Blood-to-Plasma Ratio and In Vitro Analyses
The mean ± SD Bu blood-to-plasma ratio was 0.93 ± 0.05 (n = 3). The (mean ± SD) ratio observed for methazolamide (positive control) was 23.5 ± 3.02. Assessments of the relationship between DPS and DBS at different Hct levels indicated a significant correlation between both matrices (Pearson correlation coefficient = 0.99). As the Bu concentration increased, the variability between DBS and DPS tended to become higher. The Bland–Altman plots (Fig. 1A) demonstrated that the mean Bu concentrations obtained from DBS were 5.4% lower than those obtained from DPS. The data points fell within the predefined LoA and ranged from −15.3% to 4.6%. The relationship between the Hct values and divergence in Bu levels from DBS and DPS is shown in Figure 1B. The divergence between the 2 matrices ranged from −6.69% to 15.21%, within the ±20% accepted value. No significant correlation was observed between the Hct levels and divergence in Bu concentrations measured using DBS and DPS (P = 0.41, Fig. 1B).
Correlation of DBS and DPS Methods in the Pediatric Cohort
The characteristics of all the patients included in the study are summarized in Table 2. We analyzed clinical samples (n = 153) from 15 pediatric patients who were on a 4-times daily Bu dosing regimen before allogeneic HSCT. The mean age at the time of HSCT was 7.2 years (SD = 4.9). The indication for HSCT was a malignant disease in all but one child, who had a nonmalignant condition. In most patients, sampling for the determination of Bu concentration was conducted using dose levels 1, 2, and 9 (Table 2). Hct levels were measured on consecutive days, starting from the day of the first Bu dose. Reported Hct values ranged from 21.7% to 34.7%. The mean Hct level on the first day of Bu administration was 28.3% (SD = 3.8), and 26.8% (SD = 3.6) on the second or third day. Bu concentrations ranged from 123.5 to 1695 ng/mL in DPS, and 112.5 to 1970 ng/mL in DBS. There were higher absolute differences between DBS and DPS at higher measured Bu concentrations detected in vitro, although within the defined ±20% limit. A strong correlation between DBS and DPS Bu concentrations was observed in clinical samples, with a Pearson correlation coefficient of 0.96 and a slope of 1.0 (see Figure 1A and Table 1, Supplemental Digital Content 1, http://links.lww.com/TDM/A351). The mean bias observed in DBS samples was 5.4% ± 9.6% (range: −17.7% to 18.9%), all within the defined acceptance limits for clinical use (Fig. 2A). The Bland–Altman plots for multiple measurements per individual showed agreement between the Bu levels measured by DPS and DBS sampling, with only 5.8% (9/153) of the paired samples exceeding the LoA (±1.96 SD; Fig. 2B). A good level of agreement was observed between the Bu levels measured by DBS and DPS sampling in the paired sample Bland–Altman analyses. Over 93% of the paired measurements fell within the LoA (Fig. 2C). Neither the Hct nor conditioning day altered the linear relationship between DBS and DPS levels (see Table 1 and Figure 1B, Supplemental Digital Content 1, http://links.lww.com/TDM/A351). There was no significant correlation between Hct levels and the divergence in Bu concentrations measured by DBS and DPS sampling among the clinical samples (Fig. 2D). The Bu blood-to-plasma ratios measured in vitro were confirmed from the in vivo samples of the 15 children (mean ± SD: 0.943 ± 0.099).
In this study, the use of DBS sampling for Bu TDM was clinically validated because the differences in Bu levels measured using both DPS and DBS sampling were within the predefined LoA. A good linear correlation (Spearman rank correlation coefficient = 0.97) was observed between Bu concentrations in DBS and DPS at different Hct levels. In patients undergoing HSCT for either malignant or nonmalignant diseases, Hct levels are usually below the normal levels in healthy individuals. Thus, during in vitro experiments, we attempted to prepare samples with Hct levels as close as possible to those of real patients. The assessment of agreements between Bu concentrations measured in vitro revealed higher concentrations of Bu in DBS than in DPS, with the intervals of agreement always within the predefined limit (±20%). This may be explained by the mean blood-to-plasma ratio measured in vitro, which was close to 1, indicating an almost equal plasma and RBC Bu distribution. We also observed a similar Bu blood-to-plasma ratio (0.94) among the 15 children in the clinical validation cohort. The in vitro analyses indicated that Hct had no significant influence on the Bu concentration divergence between the 2 matrices. Similar to in vitro analyses, the Hct levels had no effect on the divergence observed between Bu concentrations in clinical samples after DBS and DPS measurements. This study indicated that for compounds with equal blood and plasma distribution, Hct may not influence the linear relation between whole blood and plasma when the entire spot (volumetric) is used for analysis.23,24 We did not evaluate the effect of higher Hct values (>45%) on variations between methods because high Hct levels are uncommon in malignant patients. Nevertheless, given that increased Hct leads to decreased Bu plasma fraction,16 it could result in a change in the drug blood-to-plasma ratio and consequently a higher bias. However, considering that the effect of Hct is not significant when the entire spot is used for analysis, it is expected that increased Hct levels will not significantly affect the variations between methods. This conclusion is supported by previous studies that consider the use of the entire spot for reducing the effect of Hct on the DBS method's performance.23
These observations indicated the function of direct Bu concentration measurements using DBS sampling, without an initial estimation of plasma levels from DBS. This is in accordance with the Hct level having no influence on Bu concentrations measured using DBS or DPS sampling. However, 5.8% of the paired samples exceeded the acceptable limits in Bland–Altman analyses. Samples (only 9 out of 153) outside the acceptable limits were not restricted to a single individual or a specific time point, but were more frequent at higher Bu concentrations. Therefore, any differences after comparisons of single paired sample concentrations will result in a higher influence on method agreement, but may not significantly affect the AUC estimations from several concentrations measured per dose level. No significant deviations were observed between DBS and DPS sampling, indicating that similar AUC profiles would most likely be obtained using Bu measurements of either sampling method. However, small differences in sample collection time, time of analyses (because some of the DBS samples were analyzed 1–5 days after DPS analyses), and interpersonnel pipetting may have contributed to the variability observed between DBS and DPS. Minor differences in Bu distribution between blood cells and plasma may also lead to different drug concentrations in both matrices. Moreover, plasma protein binding may change depending on the patient's albumin level.24 Low albumin levels or protein binding saturation at very high drug concentrations in blood can increase the unbound Bu fraction, resulting in a greater distribution of the drug in the RBCs.23,24 It was observed that neither conditioning regimen, age, nor conditioning day had an influence on the correlation between Bu level measurements using DBS and DPS sampling. Samples were processed by 2 trained personnel using calibrated pipettes, thus minimizing the sample handling-induced variability. The linear relation between DBS and DPS sampling was persistent, with a slope of 0.94–1.0 (see Table 1, Supplemental Digital Content 1, http://links.lww.com/TDM/A351).
These findings demonstrate the feasibility and reliability of DBS sampling for routine Bu TDM. DBS sampling offers the advantage of decreased quantities of venous blood at multiple time point collections (only 5 µL required each time), specifically in infants and young children. Combining DBS sampling with a limited sampling strategy will require fewer collections and smaller blood quantities to provide reliable estimations of Bu clearance. Furthermore, DBS sampling can be improvised with the use of microfluidic devices to avoid potential volumetric pipetting–induced differences. DBS sampling analyses are quite rapid, and shipping across laboratories (for Bu-TDM) can be convenient, provided samples are not left at room temperature for >6 hours (analyses can be performed after 1 week if stored at −20°C).13 However, in this study, DBS samples analyzed on the same day as paired plasma samples had fewer differences than those analyzed a few days after the plasma samples. These findings are clearly indicative of the importance of the day of analyses in the quantification of Bu using DBS. Although stability studies from our laboratory have shown that Bu remains stable for up to a month when stored at −20°C,13 variations in stability may be a challenge when samples are tested after longer periods.6 DBS sampling should be feasible with the help of precise blood collection methods such as microfluidic devices, along with implementation of limited sampling strategy.
The potential limitation of our study is that our small sample size could have an impact on the statistical power to identify the interindividual variability. The sample size was estimated using data of 17 paired samples from the first 4 patients, during method validation. In this data set, the mean and SD difference between DBS and DPS were both 15 ng/mL, and the maximum allowed difference between methods was set at 55 ng/mL (<±7% difference). Considering an alpha level of 0.05 and a power of 80%, we would require 51 paired samples to test the stated difference (<±7%), which is lower than our defined limit (<±20), suggesting adequate power to establish method agreement. Although we had a considerable number of paired samples, the interindividual variability would have been clearer if a higher number of individuals were analyzed. All but one of the pediatric patients were diagnosed with malignancies, and received Bu 4 times daily; hence, the results and conclusions drawn are applicable to a similar setting. We also observed a slight but nonsignificant method bias at high Bu concentrations, which requires further testing, especially in a once-daily dosing schedule.
In conclusion, we demonstrated that Bu DBS concentrations correlated with Bu DPS concentrations. The study demonstrated the utility of DBS sampling for Bu TDM in pediatric patients with malignancies in whom Bu is administered 4 times daily. DBS sampling offers unique benefits to infants and younger children because only minute blood quantities are required. Moreover, changes in Hct levels, conditioning day, and age had no impact on the assay's performance, enabling the implementation of this simple method without the need for prior Hct level determination. Clinical validation criteria were adequately met despite the fact that a few of the paired samples fell outside the predefined clinical acceptance limits. DBS is a promising potential alternative to plasma sampling for Bu TDM in pediatric patients.
1. Tran H, Petropoulos D, Worth L, et al. Pharmacokinetics
and individualized dose adjustment of intravenous busulfan
with advanced hematologic malignancies undergoing allogeneic stem cell transplantation. Biol Blood Marrow Transpl. 2004;10:805–812.
2. Ansari M, Curtis PH, Uppugunduri CRS, et al. GSTA1 diplotypes affect busulfan
clearance and toxicity in children
undergoing allogeneic hematopoietic stem cell transplantation: a multicenter study. Oncotarget. 2017;8:90852–90867.
3. Bartelink IH, Bredius RG, Belitser SV, et al. Association between busulfan
exposure and outcome in children
receiving intravenous busulfan
before hematologic stem cell transplantation. Biol Blood Marrow Transpl. 2009;15:231–241.
4. Ansari M, Théoret Y, Rezgui MA, et al. Association between busulfan
exposure and outcome in children
receiving intravenous busulfan
before hematopoietic stem cell transplantation. Ther Drug Monit. 2014;36:93–99.
5. Huezo-Diaz P, Uppugunduri CR, Tyagi AK, et al. Pharmacogenetic aspects of drug metabolizing enzymes in busulfan
based conditioning prior to allogenic hematopoietic stem cell transplantation in children
. Curr Drug Metab. 2014;15:251–264.
6. Choong E, Uppugunduri CRS, Marino D, et al. Therapeutic drug monitoring of busulfan
for the management of pediatric patients: cross-validation of methods and long-term performance. Ther Drug Monit. 2018;40:84–92.
7. Ansari M, Lauzon-Joset JF, Vachon MF, et al. Influence of GST gene polymorphisms on busulfan pharmacokinetics
. Bone Marrow Transpl. 2010;45:261–267.
8. Bartelink IH, Lalmohamed A, van Reij EM, et al. Association of busulfan
exposure with survival and toxicity after haemopoietic cell transplantation in children
and young adults: a multicentre, retrospective cohort analysis. Lancet Haematol. 2016;3:e526–e536.
9. Courtney JB, Harney R, Li Y, et al. Determination of busulfan
in human plasma using an ELISA format. Ther Drug Monit. 2009;31:489–494.
10. Rifai N, Sakamoto M, Lafi M, et al. Measurement of plasma busulfan
concentration by high-performance liquid chromatography with ultraviolet detection. Ther Drug Monit. 1997;19:169–174.
11. Peris JE, Latorre JA, Castel V, et al. Determination of busulfan
in human plasma using high-performance liquid chromatography with pre-column derivatization and fluorescence detection. J Chromatogr B Biomed Sci Appl. 1999;730:33–40.
12. Quernin MH, Poonkuzhali B, Montes C, et al. Quantification of busulfan
in plasma by gas chromatography-mass spectrometry following derivatization with tetrafluorothiophenol. J Chromatogr B Biomed Sci Appl. 1998;709:47–56.
13. Ansari M, Uppugunduri CR, Déglon J, et al. A simplified method for busulfan
monitoring using dried blood spot
in combination with liquid chromatography/tandem mass spectrometry. Rapid Commun Mass Spectrom. 2012;26:1437–1446.
14. Matsumoto K, Uchida N, Sakurai A, et al. Clinical application of the dried blood spot
method in the measurement of blood busulfan
concentration. Biol Blood Marrow Transpl. 2016;22:1968–1973.
15. Wilhelm AJ, den Burger JC, Swart EL. Therapeutic drug monitoring by dried blood spot
: progress to date and future directions. Clin Pharmacokinet. 2014;53:961–973.
16. Ehrsson H, Hassan M. Binding of busulfan
to plasma proteins and blood cells. J Pharm Pharmacol. 1984;36:694–696.
17. Hassan M, Oberg G, Ehrsson H, et al. Pharmacokinetic and metabolic studies of high-dose busulphan in adults. Eur J Clin Pharmacol. 1989;36:525–530.
18. Denniff P, Spooner N. The effect of hematocrit on assay bias when using DBS samples for the quantitative bioanalysis of drugs. Bioanalysis. 2010;2:1385–1395.
19. Yu S, Li S, Yang H, et al. A novel liquid chromatography/tandem mass spectrometry based depletion method for measuring red blood cell partitioning of pharmaceutical compounds in drug discovery. Rapid Commun Mass Spectrom. 2005;19:250–254.
20. European Medicines Agency. Guideline on bioanalytical method validation. Available at: https://www.ema.europa.eu/en/documents/scientific-guideline/guideline-bioanalytical-method-validation_en.pdf
. Accessed May 25, 2019.
21. Bland JM, Altman DG. Agreement between methods of measurement with multiple observations per individual. J Biopharm Stat. 2007;17:571–582.
22. Zou GY. Confidence interval estimation for the Bland-Altman limits of agreement with multiple observations per individual. Stat Methods Med Res. 2013;22:630–642.
23. Enderle Y, Foerster K, Burhenne J. Clinical feasibility of dried blood spots: analytics, validation, and applications. J Pharm Biomed Anal. 2016;130:231–243.
24. Emmons G, Rowland M. Pharmacokinetic considerations as to when to use dried blood spot
sampling. Bioanalysis. 2010;2:1791–1796.