Dried blood spots from finger prick facilitate therapeutic drug monitoring of adalimumab and anti‐adalimumab in patients with inflammatory diseases

Abstract

Aims

Development of a self‐sampling method for therapeutic drug monitoring (TDM) of biologicals will enhance TDM implementation in routine care and pharmacokinetic knowledge. The aim of this study was to compare adalimumab and anti‐adalimumab antibody (ADA) concentration measurements in dried blood spots (DBS) obtained from finger prick with measurements in serum obtained via venepuncture, from patients with rheumatic inflammatory diseases.

Methods

In this cross‐sectional study, 161 consecutive patients were included. For clinical validation, DBS from finger prick and serum from venepuncture were collected simultaneously and adalimumab and ADA concentration were assessed by ELISA and antigen binding test (ABT), respectively. To convert DBS eluate results to values which can be compared to serum concentrations, five different methods were investigated, using a marker protein or a volumetric approach.

Results

Adalimumab and ADA concentrations obtained from the finger prick/DBS method correlated well with serum results from the same patient (correlation coefficient > 0.87). Interestingly, antibody concentrations (either adalimumab, ADA or total immunoglobulin G) in DBS from finger prick, but not albumin, were systematically lower compared to serum. Spike experiments demonstrated a quantitative recovery for all tested proteins in DBS, suggesting a slightly different protein composition of blood collected via finger prick vs. venepuncture. We established a correction factor to relate finger prick/DBS values with serum values (approximately 1.2).

Conclusions

We show here for the first time that adalimumab and ADA serum concentrations can be satisfactorily estimated by measuring concentrations in DBS eluates, collected by finger prick. This method offers great opportunity to simplify TDM of adalimumab.

What is Already Known about this Subject

• Adalimumab concentrations in venepuncture serum are correlated with clinical efficacy.

• Dried blood spots (DBS) obtained by finger prick are used as a more convenient way to draw blood for diagnostic purposes compared to venepuncture.

• The use of DBS for quantitative detection of therapeutic antibodies in capillary blood for diagnostic purposes in a large cohort has not been published before.

What this Study Adds

• We show that the therapeutic antibody adalimumab and total IgG can be accurately measured in DBS obtained by finger prick of patients with rheumatic inflammatory diseases.

Introduction

Adalimumab standard dosing can result in insufficient clinical response or overtreatment in a large proportion of patients 1, 2, 3, 4. Several circumstances can affect clinical efficacy, ranging from undetectable drug concentrations due to immunogenicity, to serum trough concentrations substantially above the threshold necessary for complete target blockade 1, 2, 3, 4, 5, 6, 7, 8. Therefore, it seems logical to use a personalized dosing scheme based on drug concentration and disease activity. Indeed, therapeutic drug monitoring (TDM) can help to identify causes for insufficient response 9, and concentration–effect relationships have been identified in multiple studies 1, 2, 3, 4, 10. Preliminary treatment algorithms and tapering strategies, making use of TDM, are currently available 11, 12, 13, 14. In addition, preliminary results of a tapering study of adalimumab in rheumatoid arthritis patients in our hospital demonstrate the feasibility of concentration‐guided interval prolongation 15.

However, currently used methods for drug monitoring rely on blood collection by venepuncture. This requires qualified personnel and flexibility of the patient. Visiting an outpatient clinic for blood collection, and logistics of storing and shipping samples is cumbersome, in routine care as well as during clinical trials. Alternatively, development of a dried blood spot (DBS) obtained via a finger prick performed at home will enable self‐sampling, with the results ready for immediate decision making at consultation of the care giver when the next dose needs to be administered. Moreover, self‐sampling is easy and minimally invasive. Only a small volume on a filter paper is required; it is convenient for storage and for many analytes transportation can be performed by the regular mail service at ambient temperature according to WHO regulations 16, 17. Furthermore, development of a self‐sampling method is also an important step forward in gaining more pharmacokinetic (PK) knowledge, needed for the implementation of TDM. In particular, it enables convenient data collection at multiple time‐points, allowing, for example, the evaluation of drug concentrations at time points other than through, for which there is currently a paucity of data.

Detection of antibodies in DBS has been described for screening of metabolic diseases, allergies, viral infections and vaccination efficacy 18. These studies did not address quantitative measurements of monoclonal antibody concentrations. One (small) study described preliminary results of their developed DBS method for detection of adalimumab and infliximab concentrations 19.

Here, we describe how DBS/finger prick can be used in a controlled environment in patients with rheumatic inflammatory diseases treated with adalimumab to obtain reliable estimates of serum concentrations of adalimumab and anti‐adalimumab (ADA). To our knowledge this is the first extensive study in which a venepuncture and a finger prick are obtained simultaneously in patients, for clinical validation; and this is the first study to develop a DBS method for the measurement of ADA.

Patients and Methods

Study design and patients

A cross‐sectional study was performed in patients with rheumatoid arthritis (RA) (n = 96), psoriatic arthritis (PsA) (n = 31) and ankylosing spondylitis (AS) (n = 34) treated with adalimumab (AbbVie Inc.). All patients started adalimumab 40 mg once every 2 weeks, with our without synthetic disease‐modifying antirheumatic drugs (DMARDs), non‐steroidal anti‐inflammatory drugs (NSAIDS) and/or prednisone. Dose adaptation could be made based on the expert opinion of the treating rheumatologist. The study was approved by the Medical Ethics Committee of the Slotervaart hospital and Reade Amsterdam, the Netherlands. All patients gave written informed consent according to the Helsinki declaration.

Collection material by venepuncture and finger prick

Venepuncture (Vp) and finger prick were performed at the same visit by a trained laboratory assistant at the hospital. Capillary blood from the finger prick was adsorbed on a piece of filter paper (Whatman 903, Whatman Germany) and air dried.

Elution of blood from DBS

Whole DBS was eluted from the filter by overnight incubation in phosphate‐buffered saline (PBS) containing 0.05% Tween and 0.05% NaN₃ gently shaking at room temperature. For the recovery experiment, adalimumab was spiked at a serum concentration of 10 mg l⁻¹. Stability of DBS on filter cards kept at room temperature was tested with anti‐adalimumab (clone 2.2; IgG1) 20 spiked blood and no effect in elution efficiency was seen for up to 3 months (Supplemental Figure S1). Eluates were kept at 4°C until further measurements were performed.

Haematocrit (Hct) measurements

Hct level of whole blood obtained from venepuncture was measured with the XN9000 from Sysmex (OLVG, Amsterdam) or with the ADVIA 2120 (Sanquin Diagnostics). The coefficient of variation (CV) for this measurement was 1.5% (Supplemental Table S1).

DBS area measurements

The blood area of the DBS was quantified by scanning the filter paper, and measuring DBS area with the image processing program Image J 21. The average area of the front and the back was taken as representative for the average extent of the blood spot within the filter paper. The CV of the area measurements was 2% (Supplemental Table S1).

DBS haemoglobin (Hb) concentration measurements

Hb concentrations were measured by Sanquin Diagnostics based on absorbance at 540 nm of the DBS eluates with a CV of 3.9% (Supplemental Table S1).

Immunoglobulin G (IgG) and albumin concentration measurements

IgG and albumin concentrations in DBS eluates and Vp‐serum were measured by nephelometry (Behring Nephelometer II) by Sanquin Diagnostics. CV of these assays is 5.4% and 5.96%, respectively (Supplemental Table S1).

Adalimumab and ADA concentration measurements

Adalimumab and ADA concentrations were measured using an ELISA and antigen‐binding test (ABT), respectively, both described previously 1, 6; Accuracy (precision in % CV) for adalimumab concentration is: 108% (4%) for 10 μg l⁻¹; 92% (5%) for 2 mg l⁻¹; 92% (7%) for 4 mg l⁻¹; 90% (6%) for 20 mg l⁻¹. The accuracy and precision of adalimumab spiked DBS (n = 36) is 101% and 8% respectively. The lower limit of quantification (LLOQ) DBS‐serum concentration is dependent on the blood spotted on the DBS and hct value of the spotted blood, but has a maximum of 0.411 mg l⁻¹, with an accuracy of 99% and a precision of 5% CV. Accuracy (precision in % CV) for ADA in the ABT is: 94% (14%) for 30 AU ml⁻¹; 99% (4%) for 50 AU ml⁻¹; 97% (5%) for 200 AU ml⁻¹; 101% (7%) for 700 AU ml⁻¹; 96% (6%) for 975 AU ml⁻¹. ADA spiked DBS (n = 22) has an accuracy of 107% and a precision of 7% CV. DBS eluates were tested starting with a fivefold dilution instead of the usual 50 times dilution in the ADA ABT, therefore the limit of detection (LOD) in DBS eluates is 1.2 AU ml⁻¹ and, depending on the hct and amount of spotted blood, has a maximum of 33 in the corresponding DBS‐serum.

Conversion of DBS values into DBS‐serum concentrations

Five different methods were investigated to convert values measured in DBS eluates (total IgG, albumin, adalimumab and ADA) into a value that can be compared with Vp‐serum concentrations (the so‐called DBS‐serum concentrations). An overview is provided in Table 1. Rather than performing a fixed area punch, the methods are based on elution of the entire blood spot from which its surface area is measured. The calculations are given in the Supplemental Methods.

Table 1.

Description and formula of five different methods to calculate serum fraction in order to convert DBS values into DBS‐serum concentrations

	Method	Description	Formula
Reference methods	Vp A	Serum fraction in DBS eluate is calculated with albumin marker protein concentrations in both serum and DBS eluate	$c_{\mathrm{s}}=c_{\mathrm{e}}\cdot \frac{{\mathrm{alb}}_{\mathrm{s}}}{{\mathrm{alb}}_{\mathrm{e}}}$
	Vp H	Serum fraction in DBS eluate is calculated by DBS area in combination with the Vp Hct	$c_{\mathrm{s}}=c_{\mathrm{e}}\cdot \frac{1}{1-\mathrm{Hct}}\cdot \frac{V_{\mathrm{e}}}{A\left(\mathrm{Hct}\right)\cdot \left(v_0+v_1\cdot \mathrm{Hct}\right)}$
Diagnostic methods	DBS A42	Serum fraction in DBS eluate is estimated based on albumin marker protein concentration in DBS eluate and a fixed albumin concentration of 42 g l−1 in serum	$c_s=c_{\mathrm{e}}\cdot \frac{42}{{\mathrm{alb}}_{\mathrm{e}}}$
	DBS H0.42	Serum fraction in DBS eluate is based on DBS area in combination with a fixed Hct factor of 0.42	$c_s=c_{\mathrm{e}}\cdot \frac{1}{1-0.42}\cdot \frac{V_{\mathrm{e}}}{A\left(\mathrm{Hct}\right)\cdot \left(v_0+v_1\cdot 0.42\right)}$
	DBS Hcomp	Serum fraction in DBS eluate is based on DBS area in combination with a Hct factor computed from Hb measurements	$c_{\mathrm{s}}=c_{\mathrm{e}}\cdot \frac{1}{1-\mathrm{Hct}}\cdot \frac{V_{\mathrm{e}}}{A\left(\mathrm{Hct}\right)\cdot \left(v_0+v_1\cdot \mathrm{Hct}\right)}$ $\mathrm{Hct}=\frac{-v_0+\sqrt{v_0^2+4\cdot \frac{\mathrm{y}\cdot {\mathrm{Hb}}_{\mathrm{e}}\cdot V_{\mathrm{e}}}{A\left(\mathrm{Hct}\right)}\cdot v_1}}{2\cdot {\mathrm{v}}_1}$

A(Hct) = area of the DBS; alb_e = albumin concentration in eluate; alb_s = albumin concentration in serum; c _e = concentration in DBS eluate; c _s = serum concentration; DBS = dried blood spot; Hb_e = Hb concentration in eluate; Hct = haematocrit; v₀ and v₁= parameters determined by Hct dependent spreading of blood on filter paper; V _e = volume of the eluate; Vp = venepuncture; y = conversion factor between Hb and Hct

Statistical analyses

Statistical analyses were executed using Graphpad Prism 6.04. Linear regression was performed to test the relationship between two variables. Correlations were calculated as Spearman correlation coefficients. Deming regression analysis was performed to calculate the slope and intercept of the shown correlations. Kruskal–Wallis multiple comparisons were used to calculate differences in percentage deviation between the quartiles. Outliers were detected with Grubbs analysis. The threshold for significance was set at a P‐value of less than 0.05.

Nomenclature of targets and ligands

Key ligands in this article are hyperlinked to corresponding entries in http://www.guidetopharmacology.org, the common portal for data from the IUPHAR/BPS Guide to PHARMACOLOGY 22.

Results

Adalimumab measurements in DBS

To be able to compare adalimumab concentrations in Vp‐serum and finger pricks, the concentrations were measured in both DBS eluates and corresponding Vp‐sera. Measured adalimumab concentrations in DBS eluates were converted into DBS‐serum concentrations based on the five different methods described in Table 1 and Supplemental Methods. In short: the two reference methods use either albumin as an internal marker protein (Vp A) or a volumetric conversion in combination with whole blood Hct value (Vp H). Of the three diagnostic methods, two use either a fixed estimated albumin concentration of 42 g l⁻¹ (DBS A42), or a fixed estimated Hct of 0.42 in combination with the DBS area (DBS H0.42). These methods are expected to deliver results that may systematically deviate depending on the actual albumin or Hct values, respectively. Since Hct also affects the surface area of the DBS (Supplemental Figure S1A), the impact of Hct on the blood–plasma ratio is partly abolished. The third diagnostic method uses an Hct computed from the measured DBS Hb value in combination with the DBS area (DBS Hcomp), and should not yield such a deviation, but may suffer from a larger CV due to the additional Hb measurement. The correlation of DBS Hb and Hct was determined by measurement of Hb in DBS eluates of the same volume spotted blood with different Hct values (Supplemental Figure S2B).

With these methods the DBS obtained by finger prick and corresponding Vp‐serum were compared in a cohort of 161 patients with RA, PsA or AS treated with adalimumab. Baseline characteristics and response rates of the patients included in this study showed that this cohort is representative for patients treated in daily clinical practice (Table 2). A total of 27 (16.8%) patients had detectable ADA, and given these low numbers the clinical validation was done only with the adalimumab concentration data.

Table 2.

Baseline demographics and characteristicsa

	RA	PsA	AS
	96 (59.6%)	31 (19.3%)	34 (21.1%)
Demographics
Age, years, median (IQR)	59.22 (49.3–67.5)	52.9 (44.0–57.7)	50.5 (41.0–56.8)
Female, no (%)	78 (81.3)	13 (41.9)	12 (35.3)
BMI, median (IQR)	24.3 (22.3–28.8)	26.8 (24.2–29.9)	24.4 (22.2–26.5)
Disease characteristics
Disease duration, years, median (IQR)	8 (5–14)	8 (4–13)	7 (4–14)
RF, no (%)	60 (62.5)	—	—
ACPA, no (%)	48 (50)	—	—
HLA‐B27, no (%)	—	—	29 (85.3)
DAS28, mean ± SD	2.3 ± 1.0	—	—
PASI, median (IQR)	—	0 (0.04)	—
BASDAI, mean ± SD	—	—	3.3 ± 2.2
Laboratory parameters (serum)
ESR mm h−1, median (IQR)	9 (5–21)	5 (2–10)	7 (2–11)
CRP mg l−1, median (IQR)	1.8 (0.7–4.4)	1.1 (0.6–2.2)	1.9 (0.7–4.1)
Haematocrit (l l−1), mean ± SD	0.41 ± 0.03	0.43 ± 0.03	0.43 ± 0.03
Haemoglobin (mmol l−1), mean ± SD	8.4 ± 0.6	8.9 ± 0.7	8.9 ± 0.7
Albumin (g l−1) mean ± SD	41 ± 3.3	43.5 ± 2.9	42.9 ± 3.2
Total IgG (g l−1), mean ± SD	12.6 ± 3	12.4 ± 3.1	12.5 ± 2.9
Medication use
Duration adalimumab, years, median (IQR)	4 (2–8)	3 (1–7)	4 (2–7)
Adalimumab concentration, median, mg l−1, (min.–max.)	5.9 (0.01–18.28)	5.3 (0.01–18.48)	6.6 (0.01–17.17)
ADA+, AU ml−1, no (%)	16 (16.7)	3 (9.7)	8 (23.5)
Methotrexate use, no (%)	76 (79.2)	25 (80.6)	—
NSAID use, no (%)	—	—	14 (41.2)
Clinical response b
Remission, MDA or inactive disease, no (%)	61 (63.5)	20 (64.5)	17 (50)

^aNormally distributed continuous variables are represented by mean values ± standard deviation (SD) and non‐normally distributed continuous variables are represented by median values (interquartile range, IQR); dichotomous variables are represented by numbers (percentages of total).

^bClinical response was defined according to disease‐specific criteria. Remission for RA was defined as a DAS28 < 2.6 31; MDA for PsA was defined by criteria drafted by Group for Research and Assessment of Psoriasis and Psoriatic Arthritis (GRAPPA) 32; and inactive disease for AS was defined as a BASDAI < 4 33.

ACPA = anticyclic citrullinated peptide; ADA = anti‐drug antibody; AS = ankylosing spondylitis; BASDAI = bath ankylosing spondylitis disease activity index; BMI = body mass index; CRP = C‐reactive protein; DAS28 = disease activity score using 28 joint count; ESR = erythrocyte sedimentation rate; HLA‐B27 = human leukocyte antigen B27; IgG = immunoglobulin G; MDA = minimal disease activity; NSAID = non‐steroid anti‐inflammatory drugs; PASI = psoriasis area severity index; PsA = psoriatic arthritis; RA = rheumatoid arthritis; RF = rheumatoid factor.

Figure 1A shows the correlation of adalimumab concentrations in Vp‐serum samples and corresponding DBS‐serum concentrations calculated with the DBS H0.42 method. There was a good correlation between the two values (r = 0.9342). However, adalimumab DBS‐serum concentrations were consistently lower compared to their corresponding Vp‐serum concentrations, which was seen for all five different methods (Supplemental Figure S3). By plotting the percentage of deviation in adalimumab DBS‐serum vs. Vp‐serum concentration against the adalimumab Vp‐serum concentration (Figure 2), a median deviation of around −16% was seen for all methods across the full range of adalimumab concentrations. Similar patterns were observed when plotting the results in a standard Bland–Altman plot (Supplemental Figure S4).

Figure 1.

Correlation of adalimumab (ADL) (mg l⁻¹) (A) and anti‐ADL (AU ml⁻¹) (B) measured in Vp‐serum as reference value vs. the DBS‐serum concentration calculated with the DBS H0.42 method. Data were analysed with Spearman correlation and Deming regression. ADL: r = 0.9342; slope (95% confidence interval (CI)): 0.8160 (0.774–0.8580); intercept (95% CI): 0.0722 (−0.2419–0.3864); anti‐ADL: r = 0.8741; slope (95% CI): 0.8488 (0.8416–0.8560); intercept (95% CI): −2.05 ± (−34.65–30.55); four DBS‐serum concentrations were low positive (13.3–17.1 AU ml⁻¹) whereas antibodies were not detected above the cut‐off of 12 AU ml⁻¹ in the corresponding sera

Figure 2.

Percentage deviation ( $\frac{\mathrm{DBS}‐\text{serum}‐\mathrm{Vp}‐\text{serum}}{\mathrm{Vp}‐\text{serum}}•100\%$) in adalimumab (ADL) concentration (mg l⁻¹) calculated from the DBS eluates in the five different ways (DBS‐serum concentrations) compared to the corresponding Vp‐serum concentrations as reference value. The dashed line represents the median deviation, the dotted lines represent the ±1.96 SD of the mean. The mean, % coefficient of variation (CV), estimated %CV and 95% CI of the median are given in the embedded box

To evaluate whether adalimumab DBS‐serum concentrations may systematically deviate from the ‘true’ Vp‐serum concentrations as a function of Hct or albumin, minimum to maximum deviation of quartiles in adalimumab concentrations were plotted against these parameters (Supplemental Figure S5). For the Vp A method, less deviation in adalimumab concentrations was present in the last vs. first quartile of Vp‐serum albumin concentrations; and vice versa for the DBS A42 method (Supplemental Figures S5F and G). For the volumetric methods, no correlation was seen for the Vp‐concentrations with Hct (nor albumin or adalimumab, Figures 3A and B; Supplemental Figures S5C–E; S5H–J and S5M–O), although a non‐significant downward trend was observed for deviation in adalimumab concentration with increasing Hct (Figure 3B) for DBS H0.42.

Figure 3.

Percentage deviation in adalimumab (ADL) concentration (mg l⁻¹) calculated from the DBS eluates with the diagnostic method DBS H0.42 compared to the corresponding Vp‐serum concentration plotted against the quartiles of the ADL concentration (mg l⁻¹) (A) or the Hct values (l l⁻¹) (B). Box plots with minimum to maximum deviation are shown. Data were analysed with a Kruskal–Wallis test and no significant differences (P < 0.05) between the different quartiles were found

Besides the systematic lower antibody concentration in DBS eluates, there was also a certain level of random variation. This random variation can be almost completely accounted for by the expected collective %CV of the combination of assays needed to calculate the deviation of DBS‐serum over Vp‐serum. For DBS H0.42 the estimated %CV is 10.10 and the %CV observed in this cohort was 13.41 (Figure 2). The separate CVs of the different assays are given in Supplemental Table S1. If actual Hct values were estimated by measuring Hb, and these were taken into account when calculating adalimumab concentrations (DBS Hcomp), the CV increases from 13.4% to 15.0% (Figure 2).

Given that adalimumab is an IgG antibody, we wondered whether the aforementioned lower DBS‐serum adalimumab concentration reflects a general difference in antibody concentration in Vp‐serum compared to DBS‐serum concentrations. Indeed, IgG DBS‐serum concentrations were also lower compared to Vp‐serum concentrations (Figure 4A), although the deviation was less pronounced compared to adalimumab (−11.43% for IgG; −16.26% for adalimumab). Notably, this decreased concentration of antibodies in DBS eluates was not an overall protein effect, since the albumin concentration was not significantly lowered in DBS‐serum concentrations compared to serum (median deviation: 2.01; Figure 4B). To exclude an effect of loss of antibody protein during extraction from DBS and/or an inability to detect the protein in the eluates, we spotted whole Vp‐blood spiked with adalimumab on DBS cards, made eluates after 1 week storage, calculated the precise amount of plasma present in DBS eluates and compared the concentrations of adalimumab, IgG and albumin in the DBS‐plasma with the plasma concentrations, which were not spotted. There was no significant loss in adalimumab, IgG or albumin with blood spotted on DBS cards (Figure 4C). Since the lower antibody concentration in the DBS‐serum is consistently observed over the entire range of concentrations, the discrepancy between finger prick blood and Vp‐blood can be easily corrected by multiplying the calculated concentrations with a constant factor, based on the average decrease in antibody concentration in DBS‐serum compared to Vp‐serum. This factor was calculated to be 1.19 (= 1/(1–16.26%)) in this study. After conversion of DBS‐serum concentrations with this correction factor, the slope of the regression line between DBS‐serum and Vp‐serum concentration becomes close to 1. In addition, the median percentage deviation in ADL concentration and the bias in percentage difference in ADL concentration in the Bland–Altman plot become close to 0 (Supplemental Figure S6A, C and D). Moreover, converting the adalimumab concentration obtained by both Vp‐serum and DBS H0.42 method (with correction factor) to categories based on previously reported cut‐off values 1, 15, an overall similar pharmacokinetic profile was found, respectively, 59 vs. 60, 64 vs. 58, and 43 vs. 38 patients presented with adalimumab levels <5, 5–8 and >8 mg l⁻¹.

Figure 4.

Percentage deviation in IgG concentration (g l⁻¹) calculated from the DBS eluates with the diagnostic method DBS H0.42 compared to the corresponding Vp‐serum concentration plotted against the Vp‐serum IgG concentration. One outlier in percentage deviation in IgG concentration was identified by Grubbs analysis (black triangle) and removed from the data set. The dashed line represents the median deviation, the dotted lines represent the ±1.96 SD of the mean (A). Percentage deviation in albumin concentration (g l⁻¹) calculated from the DBS eluates with the diagnostic method DBS H0.42 compared to the corresponding Vp‐serum concentration plotted against the Vp‐serum albumin concentration. The dashed line represents the median deviation, the dotted lines represent the ±1.96 SD of the mean (B). Adalimumab (ADL) spiked Vp‐blood was spotted on DBS cards; eluates were made and ADL, IgG and albumin concentrations were measured. The DBS‐plasma concentration was calculated based on the spotted amount of blood in combination with the measured Hct value. DBS‐plasma concentration is presented as percentage recovery compared to Vp‐plasma concentration (C)

ADA measurements in DBS

In addition to adalimumab concentration, we also investigated whether ADA could be detected in DBS eluates. There was a decent correlation (r = 0.8741) between DBS‐serum concentrations calculated with the DBS H0.42 method and corresponding sera for 16.8% (27 of 161) of serum samples with detectable ADA in serum samples (Figure 1B). A lower ADA concentration, similar to the above‐mentioned result for adalimumab concentration, was seen in DBS eluates. After correction for this systematic deviation with the same factor as used for adalimumab concentration (Supplemental Figure S6B), two patients with low adalimumab concentrations (< 5 mg l⁻¹) and two patients with adalimumab concentrations in the therapeutic range (5–8 mg l⁻¹) 1 were negative for ADA in serum samples and slightly positive for ADA (between 16 and 20.5 AU ml⁻¹) in the corresponding finger prick sample. Conversely, four patients with low adalimumab concentrations (< 5 mg l⁻¹) and three patients with adalimumab concentrations in the therapeutic range (5–8 mg l⁻¹) were negative for ADA in finger prick sample and slightly positive for ADA (between 12 and 28 AU ml⁻¹) in the corresponding serum sample.

Discussion

Here we present a method that supports the usage of DBS obtained from finger prick as alternative for Vp‐serum in the measurement of adalimumab and ADA concentrations in patients with rheumatic inflammatory diseases treated with adalimumab. This method would simplify the process of TDM, thereby increasing its possibility for implementation in routine care. However, the advantages of a DBS method (e.g., less burden for patient and physician) should outweigh the potential disadvantages compared to the conventional method (such as loss in precision and accuracy), and the amount of labour intensity, potentially prolonged turnaround time and costs should be within acceptable limits 16, 23, 24.

To compare the DBS values with serum concentrations, we used five different methods to convert these values to DBS‐serum concentrations. All methods showed a good correlation with Vp‐serum results for adalimumab and the precision of all assays remained below 15%, in line with the European Medicine Agency (EMA) and Food and Drug Administration (FDA) guidelines of method validation 25, 26. The precision of the final DBS assay to test adalimumab concentrations can be expected to be even lower, since no comparison with serum measurements (CV 7%) is needed.

One of the most important parameters that needs to be considered when optimizing a DBS method, according to the White Papers of the European Bioanalysis Forum (EBF), is Hct 23, 24. Hct influences recovery, spot size and blood‐to‐plasma ratio 16. The effect of Hct on these parameters was investigated in this study. Recovery was not affected by Hct, as Hct levels did not significantly alter the deviation of DBS‐serum over Vp‐serum. Spot size and blood‐to‐plasma are clearly affected by Hct levels (Supplemental Figure S2A); however, these effects are completely (Vp H and DBS Hcomp) or partially (DBS H0.42) accounted for in the calculations to convert DBS eluate concentrations into DBS‐serum concentrations. Theoretically, Hct values deviating from the average of 0.42 will result in a proportional deviation in DBS‐serum concentrations calculated with the DBS H0.42 method from the ‘true’ value (Supplemental Figure S2D). However, since almost 80% of the patients in this study had a very limited Hct range from 0.38 to 0.45, the Hct effect was small (max. 5%) compared to the random variation. This may explain why a significant influence of Hct on the calculated adalimumab concentration was not observed. Therefore, the DBS H0.42 method is a satisfactory alternative to Vp‐serum adalimumab concentrations for this patient group. Nevertheless, because a reasonably accurate Hct value was obtained in the DBS Hcomp method (Supplemental Figure S2C), this latter method is most likely preferred for patient populations with a known wider or aberrant Hct distribution, despite a slightly increased overall %CV for this method. An accurate estimation of Hct levels can also be acheived using potassium concentrations in DBS eluates instead of the Hb concentrations used in this study (27; and unpublished results).

Other parameters that could affect DBS results are spot homogeneity and DBS stability 23, 24. Because we eluted the whole DBS, the issue of spot homogeneity is excluded. Stability of anti‐adalimumab (IgG1 clone) in blood spotted DBS cards was tested for 3 months at room temperature and no loss of recovery was observed. This simplifies shipment of material and allows sending of DBS by regular mail if kept at room temperature during shipment.

The DBS method that uses albumin as marker protein to predict the amount of serum present in the DBS eluates (DBS A42) showed a small, but significant, dependency of DBS‐serum concentrations of adalimumab on actual albumin concentrations, as expected. In addition, a reverse effect of the albumin serum concentration on the deviation in adalimumab concentration in the Vp A method was observed and needs further investigation.

The systematically lower antibody concentration in DBS eluates compared to Vp‐serum suggests that there is a discrepancy in protein concentrations in capillary compared to venous blood or that the capillary blood is diluted with fluids present between the capillary and the paper. The difference in molecular weight of IgG and albumin or the presence of albumin in dilution fluids (e.g. exudate) could explain why this lower concentration in DBS eluates was not observed for albumin. Although a definitive explanation for the lower concentrations of antibody in DBS when compared to serum is lacking, it can easily be corrected by multiplying the calculated DBS concentration with a constant factor of 1.19, based on the adalimumab concentration data of the present study. This resulted in a good correlation between DBS and the Vp‐serum values of adalimumab and ADA, also when categorized in groups based on clinically relevant adalimumab concentration cut‐off values 1, 15. Although our data suggest that this factor is similar for different IgG molecules, we cannot exclude the possibility that for specific monoclonal antibodies, a different factor will be found in practice, especially when the elution procedure results in less than quantitative recovery.

The current study was a cross‐sectional analysis of patients with RA, PsA or AS treated with adalimumab. The number of patients with detectable ADA was limited (26, or 16.8%) in this study. Possible explanations might be that non‐responders due to ADA formation discontinued treatment prematurely 1, 2, 3, 4, 6; moreover, an ABT was used to measure ADA against adalimumab which is a drug‐sensitive assay 28, 29, 30. Because of the low number of ADA‐positive patients (27 of 161 patients), we only clinically validated our assay based on the adalimumab concentration data. However, ADA concentrations in DBS and serum correlate well (r = 0.8741) and the patients with discrepancies between finger prick and Vp‐serum had low ADA concentrations (max. 28 AU ml⁻¹; threshold: 12 AU ml⁻¹), which could be partly explained by an altered drug tolerance in DBS eluates compared to Vp‐serum samples.

Based on the similar behaviour of the three different antibodies in DBS tested in this study (adalimumab, ADA and IgG), we expect that TDM using DBS from finger prick can be used for patients treated with other biologicals as well. In the current study, finger pricks (as well as venepunctures) were taken by the nurse in a controlled environment. Self‐sampling by the patient at home could introduce some more variation, and proper instruction of patients will be a prerequisite for successful implementation. In the study by Vande Casteele and colleagues 19, infliximab‐ and adalimumab‐treated patients were included via a self‐taken finger prick. These results together with our own results described in this paper support the feasibility of TDM studies of biologicals via self‐sampling. Moreover, it might enhance the possibility to prospectively study clinically relevant adalimumab concentration cut‐off values and development of treatment algorithms, which are currently under investigation 1, 2, 9, 10, 11, 12, 13, 14, 15. Although, these cut‐off values and proposed algorithms are preliminary, one adalimumab concentration measurement at trough level can be sufficient for dose adaptations 15, depending on which algorithm is used.

In conclusion, we have shown in this cross‐sectional study of 161 patients with rheumatic inflammatory diseases treated with adalimumab, who are representative of patients treated in daily clinical practice, that adalimumab concentrations can be measured well in DBS samples obtained via finger prick. Precision and accuracy were within acceptable limits as described by EMA and FDA guidelines 25, 26. In addition, although low numbers of ADA‐positive samples were detected in this study, ADA concentrations also seem to be measured reasonably well in DBS samples from finger prick. Moreover, DBS can be stored at room temperature for 3 months which is convenient for shipment and only a limited amount of blood is needed. In addition, DBS will reduce costs and time of physicians or nurses and patients, compared to serum withdrawal with Vp. Implementing this DBS method simplifies the TDM process and can provide more insight into PK of adalimumab, as frequent sampling within one dosing interval can easily be performed with an finger prick taken at home.

Competing Interests

E.L.K. reports having received payment for lectures from Pfizer. G.J.W. reports having received a research grant from Pfizer (Wyeth) (paid to the institution) and payments for lectures from Pfizer, Amgen, AbbVie, UCB and BMS. M.F.P. has no disclosures. T.S. has no disclosures. M.T.N. reports having received consultancy fees from Abbott, Roche, Pfizer, MSD, UCB, SOBI and BMS, payment for lectures from AbbVie, Roche and Pfizer. A.D.V. has no disclosures. T.R. reports having received payment for lectures from AbbVie and Pfizer. K.B. has no disclosures.

Pfizer Aspire (competitive) grant. The funding party had no involvement in the study design; in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. K.B. is supported by the MRC grant: Psoriasis Stratification to Optimise Relevant Therapy (PSORT), MR/L011808/1.

The authors are grateful to research nurses and medical doctors of Reade, Amsterdam, The Netherlands, for performing clinical assessments and the technicians of the Department of Immunopathology and Blood Coagulation and the Laboratory for Red Blood Cell Diagnostics, Sanquin Diagnostics Services for performing the assays.

Contributors

All the authors were responsible for the study concept and design. E.L.K., G.J.W., T.R. and K.B. were responsible for the analysis and interpretation of the data. The study was supervised by G.J.W., T.R. and K.B. All the authors were responsible for the clinical revision and drafting of the manuscript for important intellectual content, and all authors gave final approval for the manuscript.

Supporting information

Figure S1 Stability of blood spiked with anti‐adalimumab (clone 2.2; IgG1) in DBS on filter cards was tested at room temperature for 83 days. No decrease in elution recovery was seen between day 0 and day 83

Figure S2 Blood volume (μl) per area (mm²) plotted against the Hct values of the spotted blood. Average with standard deviation (SD) of five different volumes between 100 and 200 μl is shown. Slope (95% CI): 0.2195 (0.1923–0.2466); intercept (95% CI): 0.3582 (0.3469–0.3696); r ²: 0.7456; Sy|x: 0.01810 (A). Hb in blood is plotted against Hct values of spotted blood. Hb in blood is calculated from Hb measured in eluates multiplied by the dilution factor of spotted blood in elution buffer. Slope (95% CI): 21.43 (20.73–22.13); intercept (95% CI): −0.1341 (−0.4289–0.1607); r ²: 0.9769; Sy|x: 0.4680 (B). Computed Hct based on Hb with the DBS Hcomp formula given in Supplemental methods plotted against Hct values of spotted blood. Slope (95% CI): 0.9243 (0.7767–1.072); intercept (95% CI): 0.0237 (−0.0384–0.0857); r ²: 0.4903; Sy|x: 0.0302 (C). Theoretical percentage deviation of plasma proteins calculated with the DBS H0.42 method compared to concentration in Vp‐serum plotted against hypothetical Hct values. Formula is given in Supplemental methods. Slope (95% CI): −123.6 (−124.1–123); intercept (95% CI): 51.81 (51.58–52.04); r ²: 0.9992; Sy|x: 0.1117 (D)

Figure S3 Correlation of adalimumab (ADL) concentration (mg l⁻¹) measured in Vp‐serum and calculated from DBS eluates with the five different methods described in the Patients and Methods section. Data were analysed with Spearman correlation and Deming regression. Vp A: r = 0.9506; slope (95% CI): 0.8015 (0.7647–0.8389); intercept (95% CI): 0.1306 (−0.1495–0.4107); DBS A42: r = 0.9534; slope (95% CI): 0.7666 (0.7321–0.8012); intercept (95% CI): 0.3304 (0.07217–0.5887); Vp H: r = 0.9349 slope (95% CI): 0.8215 (0.7793–0.8637); intercept (95% CI): 0.0395 (−0.2763–0.3553); DBS H0.42: r = 0.9342; slope (95% CI): 0.8160 (0.7741–0.8580); intercept (95% CI): 0.0722 (−0.2419–0.3864); DBS Hcomp: r = 0.9168; slope (95% CI): 0.8171 (0.7684–0.8658); intercept (95% CI): 0.0399 (−0.3242–0.404)

Figure S4 Standard Bland–Altman plot of the five different methods with percentage difference in adalimumab (ADL) concentration between DBS‐serum and Vp‐serum plotted against the average ADL concentration in DBS‐serum and Vp‐serum (A–E). Dashed line shows bias and dotted lines show upper and lower limit of 95% confidence interval

Figure S5 Percentage deviation in adalimumab (ADL) concentration (mg l⁻¹) in the DBS‐serum concentrations calculated in the five different ways compared to Vp‐serum concentrations plotted against the quartiles of the ADL concentration (A–E), albumin concentration (g l⁻¹) (F–J) or haematocrit (Hct) values (l l⁻¹) (K–O) in Vp‐serum as reference value. Box plots with minimum to maximum deviation are shown. Data were analysed with a Kruskal–Wallis test and significant differences (P < 0.05) between the different quartiles are indicated with an asterisk

Supporting info item

Supporting info item

Figure S6 Correlation of adalimumab (ADL) (mg l⁻¹) (A) and anti‐ADL (AU ml⁻¹) (B) measured in Vp‐serum and calculated from DBS eluates with the DBS H0.42 method after correction with conversion factor. Data were analysed with Spearman correlation and Deming regression. ADL: r = 0.9342; slope (95% CI): 0.9829 (0.9324–1.033); intercept (95% CI): 0.0306 (−0.3478–0.409); anti‐ADL: r = 0.8741; slope (95% CI): 1.014 (1.005–1.022); intercept (95% CI): −2.50 (−41.43–36.43); four DBS‐serum concentrations were low positive (15.9–20.5 AU ml⁻¹), whereas antibodies were not detected above the cut‐off of 12 AU ml⁻¹ in the corresponding sera (A–B). Percentage deviation in ADL concentration (mg l⁻¹) after conversion of DBS‐serum concentration calculated with the DBS H0.42 method with correction factor plotted against the Vp‐serum concentration as reference value. The dashed line represents the median deviation, the dotted lines represent the ±1.96 SD of the mean (C). Standard Bland–Altman plot of the DBS H0.42 method with percentage difference in ADL concentration between DBS‐serum and Vp‐serum plotted against the average ADL concentration in DBS‐serum and Vp‐serum. Dashed line shows bias and dotted lines show upper and lower limit of 95% confidence interval (D)

 

Table S1 Coefficient of variation of the individual assays. ADL: adalimumab; anti‐ADL: anti‐adalimumab antibodies; Hb: Haemoglobin; Hct: Haematocrit; IgG: immunoglobulin G

Kneepkens, E. L. , Pouw, M. F. , Wolbink, G. J. , Schaap, T. , Nurmohamed, M. T. , de Vries, A. , Rispens, T. , and Bloem, K. (2017) Dried blood spots from finger prick facilitate therapeutic drug monitoring of adalimumab and anti‐adalimumab in patients with inflammatory diseases. Br J Clin Pharmacol, 83: 2474–2484. doi: 10.1111/bcp.13371.