Translate this page into:
Evaluation of dried blood spot bioassay for estimation of HbA1c, high-sensitivity C-reactive protein, creatinine, and high-density lipoprotein-cholesterol
, Ragib Alam1, Anshul Tiwari1, Ransi Ann Abraham1, Umesh Kapil2, Tripti Khanna3, Ramakrishnan Lakshmy1,
*Corresponding author: Ramakrishnan Lakshmy, Department of Cardiac Biochemistry, All India Institute of Medical Sciences, New Delhi, India. lakshmy_ram@yahoo.com
-
Received: ,
Accepted: ,
How to cite this article: Tarik M, Alam R, Tiwari A, Abraham RA, Kapil U, Khanna T, et al. Evaluation of dried blood spot bioassay for estimation of HbA1c, high-sensitivity C-reactive protein, creatinine, and high-density lipoprotein-cholesterol. J Lab Physicians. 2025;17:333-41. doi: 10.25259/JLP_287_2024
Abstract
Objectives:
Utilization of dried blood spot (DBS) assay provides an alternative approach to conventional assays in blood or serum, particularly for neonates, older populations, and large-scale population-based studies. This study aimed to standardize the DBS-based bioassay for the estimation of glycated hemoglobin (HbA1c), high-sensitivity C-reactive protein (hs-CRP), creatinine, and high-density lipoprotein-cholesterol (HDL-C), and to validate the levels of these analytes between venous and DBS.
Materials and Methods:
Blood samples were collected on filter paper, dried at room temperature, and eluted. HbA1c, hs-CRP, creatinine, and HDL-C levels were estimated in DBS and serum/plasma (whole blood in the case of HbA1c).
Statistical analysis:
DBS-to-serum equivalency was evaluated by linear regression analysis and Bland–Altman plot analysis to assess the agreement and bias between both assays.
Results:
The DBS assay was linear, sensitive, accurate, and precise with acceptable recovery and matrix effects for HbA1c, hs-CRP, creatinine, and HDL-C. Within- and between-batch precision was within an acceptable limit (<10% for all analytes). A strong correlation and agreement in Bland–Altman analysis were revealed between liquid and DBS measures for HbA1c (r2 = 0.955), hs-CRP (r2 = 0.973), and creatinine (r2 = 0.953). However, a weak association was observed for HDL-C levels (r2 = 0.572) between DBS and liquid bioassay.
Conclusions:
We conclude that the HbA1c, hs-CRP, creatinine, and HDL-C can be measured from DBS samples and are readily transferable to a liquid phase for analysis, offering a convenient alternative for monitoring cardiovascular disease risk factors.
Keywords
Bioassay development
Biomarkers
Dried blood spot
INTRODUCTION
Dried blood spot (DBS), in which drops of capillary whole blood are collected from a finger prick, offers a minimally invasive alternative to venipuncture. Despite its advantages, of being less invasive, requiring less volume of blood collection, ease of collection, requiring no processing, and being less infectious due to drying, etc., the use of DBS for biochemical analysis is limited. Validated assays for quantifying biomarkers in DBS samples are relatively low in comparison with serum or plasma, preventing their widespread use.
The use of DBS is of special relevance in low- and middle-income countries where the availability of good-quality diagnostic laboratories is limited in remote settings. The blood collected and dried on filter paper matrix can be transported through a simple mail with no requirement for a cold chain. The challenges of blood collection are deterrents for biochemical analysis in large-scale community-based research, particularly in remote field settings where access to a centrifuge, freezer, or even electricity may be limited. DBS is also a viable alternative in neonates, older adults, as well as bedridden patients, and disabled persons.[1,2]
A validated DBS protocol needs to demonstrate a reasonable level of accuracy, precision, and reliability in laboratory measurements These steps include deciding on reagents, preparing calibration and quality control material, evaluating elution protocols, optimizing sample quantity, and assessing multiple aspects of assay performance, such as intra- and inter-assay variation, lower limit of detection, accuracy, stability, and agreement between results from matched DBS and plasma samples.[3] The objective of this study was to establish a DBS-based assay for the measurement of cardiovascular disease (CVD) risk markers such as glycated hemoglobin (HbA1c), high-sensitivity C-reactive protein (hsCRP), creatinine, and high-density lipoprotein-cholesterol (HDL-C), and to validate the levels of these analytes between DBS and traditional venous approach.
MATERIALS AND METHODS
In this study, a protocol was established for calibration and validation of the DBS bioassay for the estimation of CVD risk biomarkers, including HbA1c, hs-CRP, creatinine, and HDL-C, according to the United States Food and Drug Administration (US FDA) guidelines.[4] For calibration and validation studies, DBS were prepared from freshly collected venous blood samples, analyzed in DBS as well as in serum (whole blood in case of HbA1c), and compared. After analysis, DBS were stored for different lengths of time at different temperatures to assess the storage stability of these analytes in the second part of the study. The storage stability of creatinine and HDL-C in DBS was assessed, whereas the storage stability of HbA1c and hs-CRP in DBS has already been reported in previous studies.[5,6] This study was approved by the institutional ethics committee from the All India Institute of Medical Sciences (AIIMS), New Delhi (Ref No.: IEC-293/June 01, 2018).
Sample collection and preparation of DBS
Left-over venous ethylenediaminetetraacetic acid anticoagulated blood samples from Centralized laboratory, Cardiothoracic Neurosciences Center, AIIMS, New Delhi, were used to prepare DBS. 10 μL of whole blood was pipetted onto Whatman®qualitative filter paper, Grade 3 (Cat No: #1003-125, Whatman, USA). Blood-spotted filter paper was kept to dry at room temperature (avoiding direct sunlight) for 3 h on a non-absorbent thermocol sheet. After drying, the DBS sample was transferred to a resealable plastic zip bag individually and stored at 4°C until the analysis. Similarly, sets of DBS were prepared to assess the storage stability and stored at −20°C and −80°C.
Preparation of calibrators and quality controls on filter paper
To eliminate matrix differences, calibrators and controls were also prepared on filter paper similar to the samples. Blood-based quality controls and calibrators were prepared by mixing the quality control samples and washed erythrocytes in the ratio 1:1. Human assayed multi-sera-level 3 (Cat No. HE10401, Randox, UK) was used as a creatinine calibrator with a target value of 4.3 mg/dL. Multiple calibrators were prepared by serial dilution of this calibrator with deionized water to get four different concentrations for creatinine (4.3, 2.15, 1.07. 0.053 mg/dL) and a blank consisting of only deionized water (as a zero calibrator) was used to plot a calibration curve for the estimation of actual creatinine level in DBS. Human assayed multi-sera-level 2 (Cat No.: HN10400, Randox, UK) (concentration: 1.45 mg/dL) and level 3 (Cat No.: HE10401, Randox, UK) (concentration: 4.3 mg/dL) were used as a control for creatinine. Lipid control level 2 (Cat. No. LE2669, Randox, UK) with target value 41.3 mg/dL (range 35.1–47.5 mg/dL) and lipid level 3 (Cat. No. LE2670, Randox, UK) with target value 62.1 mg/dL (range 52.9–71.3 mg/dL) were used as controls for HDL-C assay. Lipid control level 3 (62.1 mg/dL) was serially diluted with deionized water to get the different concentrations for HDL-C calibrators (62.1, 31.5, 15.7 mg/dL) and a blank consisting of only deionized water (as a zero calibrator) and used to generate the calibration curve. The calibrators and controls for hs-CRP were provided with an enzyme-linked immunosorbent assay (ELISA) kit, and the same was used for the calibration of these analytes in the DBS assay.
Estimation of analytes in DBS extract and liquid specimen
One spot (6 mm diameter) was punched out from the blood spot and pre-treated with 200 μL hemoglobin denaturant reagent (provided with kits) and incubated at 37°C for 20 min. The samples were centrifuged at 3000 rpm for 10 min, and the extracted supernatant was used for analysis. HbA1c in DBS samples and whole blood was estimated by the latex agglutination inhibition assay kit (Randox, UK) on an auto-analyzer (Beckman coulter).
For hs -CRP estimation, one spot (6 mm diameter) was punched out and eluted with 200 μL of hs-CRP diluent provided in the kit. Tubes were incubated on a shaker at 37°C for 1 h. Subsequently, the tubes were centrifuged at 3000 rpm for 10 min, and the supernatant was used for hsCRP estimation according to the manufacturer’s protocol. The hs-CRP in DBS and serum was analyzed using an ELISA assay using kits (BioCheck, Inc., Foster City, CA).
For creatinine estimation, two spots (6 mm diameter) of sample, calibrators, and controls were punched out into tubes, and 100 μL of methanol was added. Tubes were incubated on a shaker at 37°C for 1 h and further centrifuged at 3000 rpm for 10 min. Creatinine in DBS and serum samples was estimated by the Jaffe method manually using a kit from Roche (Ref No: 11875418216, Roche Diagnostics Ltd, Germany) with minor modifications. Reagent R-1 (0.32 mol/L Sodium Hydroxide) was freshly prepared in the laboratory, and reagent R2, containing 25 mmol/L picric acid (provided with the kit), was diluted to get a concentration of 20 mmol/L. 75 μL supernatant from the sample, calibrators (4.3, 2.15, 1.07, 0.053 mg/dL and a blank), and control extracted from dried spot were mixed with 125 μL reagent R1 in a 96-well plate, and after 60 s, reagent R2 was added. Absorbance (A1) was taken at 492 nm in a spectrophotometer plate reader immediately within 15 s, and then, the plate was incubated for 20 min at 37°C. The second absorbance (A2) was taken exactly after 20 min. ∆A = (A2–A1) for the sample or standard was calculated, and a standard curve was plotted with the values obtained for calibrators, and creatinine levels in samples were calculated using the standard curve.
For HDL-C estimation, one disk of 6 mm diameter was punched out and eluted with 100 μL of reagent 1 (provided with the kit). Tubes were incubated on a shaker at 37°C for 1 h with intermittent mixing and centrifuged at 3000 rpm for 10 min, and the supernatant was used for HDL-C cholesterol estimation according to protocol. HDL-C in DBS and serum was measured by the enzymatic method using HDL-C direct assay kit (DIALAB Gmbh, Austria) based on immuno-inhibition methodology.
Analytical parameters for DBS bioassay
The standardization of the DBS assays was performed according to the US FDA guidelines of 2018.[4] The analytical parameters, including linearity, recovery, and between-batch coefficient of variation (CV), were applied to assess the performance of the DBS-based assay, and storage stability of proposed analytes was also assessed in dried blood on filter paper.
Linearity
The linearity of the calibration curve generated from a range of calibrators of HbA1c, hs-CRP, creatinine, and HDL-C eluted from filter paper was assessed and considered acceptable if the coefficient of determination (r2) was ≥0.990.
Recovery
The recovery of the HbA1c, hs-CRP, creatinine, and HDL-C from DBS was calculated by comparison of the observed concentration of the particular analytes with the actual concentration in whole blood using the following equation: recovery (%) = ([observed concentration in DBS/concentration in liquid specimen]) × 100.
Precision and accuracy
The precision of the bioassay was determined by the intra-assay and inter-assay CV% of repeated measures of the analytes. For the inter-assay CV, quality control samples were run on different days, and for intra-assay CV, quality control samples repeated multiple times in the same assay were taken. Mean and standard deviation of values were calculated, and CV was computed by dividing the standard deviation by the mean and multiplying by 100 to express as a percentage. Accuracy or percentage bias was computed by subtracting the observed value in serum from the observed value in DBS and dividing this by the observed value in serum, and multiplying the value obtained by 100.
Validation of DBS assay
The estimations of analytes were performed in serum (whole blood in case of HbA1c) as well as in DBS and compared. Creatinine and HDL-C were assessed in 31 samples in serum as well as DBS assay. HbA1c and hs-CRP were done in 22 samples.
Storage stability
Freshly prepared DBS were stored at three different temperatures including 4°C, −20°C, and −80°C, and analytes were assessed at baseline, 3 months, and 6 months, and validated with baseline values of the same samples. The percentage deviation /bias in the analyte value was computed by subtracting values at baseline (T0) from those at other time periods (Tx) using the following formulae: percentage deviation /bias = ([Tx - T0]/T0) × 100.
Statistical analysis
Statistical analysis was carried out using IBM Statistical Package for the Social Sciences, version 22.0. P < 0.05 was considered significant. To evaluate whether the bias and average absolute error of the DBS method were within acceptable limits, we calculated the acceptable total change limit (TCL) using the following formula: TCL = √([2.77 × CV1]2+ [0.5 × CV2]2) as reported in a previous study.[7] CV1 was derived from the coefficient of analytical variation (inter-assay CV) for the DBS assay for each assay, and biological variation (CV2) was taken from values of biological variations (CVb) compiled by Ricos et al.[8]
RESULTS
The analytical performance of the DBS assay for HbA1c, hs-CRP, creatinine, and HDL-C is depicted in Table 1. The coefficient of determination (r2) of the standard curve was >0.990 for all analytes, which indicates an excellent linear relationship between the absorbance and concentration of standards for each analyte. The observed mean recovery was >95% for HbA1c, hs-CRP, and creatinine, respectively, indicating a satisfactory recovery of analytes from the dried spot. Intra-assay CV and the inter-assay CV were <10% for HbA1c, hs-CRP, creatinine, and HDL-C. The recovery (94%) was lower for HDL-C as compared to other analytes. The accuracy of the proposed assay was assessed as the bias percent within a day. Within-day bias was 0.78%, 0.69%, 4.76%, and 8.50% for HbA1c, hs-CRP, creatinine, and HDL-C, respectively.
| Characteristic | HbA1c | hsCRP | Creatinine | HDLC |
|---|---|---|---|---|
| Linearity (r2) | 0.999 | 0.999 | 0.999 | 0.991 |
| Recovery (%) | 103 | 99.3 | 109 | 94 |
| Intraassay CV% | 5 | 6.9 | 5.1 | 7.2 |
| Interassay: CV% | 7.6 | 8.1 | 6.5 | 8.7 |
| Accuracy (bias%) | 0.78 | 0.69 | 4.76 | 8.5 |
HbA1c: Glycated hemoglobin, hs-CRP: High-sensitivity C-reactive protein, HDL-C: High-density lipoprotein-cholesterol, CV: Coefficient of variation, DBS: Dried blood spot
Validation of liquid and dried bioassay
The results obtained with DBS and liquid assays were compared through a linear correlation and Bland–Altman analysis for agreement, as shown in Figure 1.
- Comparison between analytes levels assessed in liquid and dried bioassay: Correlation and Bland–Altman agreement. For HbA1c, (a) represents the linear regression plots and solid line represents the data best fit (r2 = 0.955); and (b) shows Bland–Altman plot for agreement analysis of dried blood and venous blood. The bias was −0.096% and limit of agreement (LoA) ranged from −2.21 to 2.02%. (c and d) represent the linear regression plots (r2 = 0.973) and Bland–Altman plot for agreement analysis (mean bias: 0.11 mg/L; LoA: −0.66–0.44 mg/L) between high-sensitivity C-reactive protein levels in dried and liquid bioassay, respectively. (e and f) represent the linear regression plots (r2 = 0.953) and Bland–Altman plot for agreement analysis (mean bias: 0.08 mg/dL; LoA: −0.29–0.45 mg/dL) between creatinine levels in dried and liquid bioassay, respectively. (g and h) represent the linear regression (r2 = 0.572) and Bland–Altman analysis (mean bias: 1.97 mg/dL; LoA: −12.4–16.3 mg/dL) between high-density lipoprotein-cholesterol levels in dried and liquid bioassay, respectively. hs-CRP: high-sensitivity C-reactive protein, HDL-C: high-density lipoprotein-cholesterol, DBS: Dried Blood Spot.
Validation of HbA1c levels between whole blood and DBS bioassay
To validate the measurement of HbA1c, DBS of fresh whole blood samples from 22 subjects was prepared and assessed in whole blood as well as DBS. No significant difference was observed in mean (±Standard deviation [SD]) HbA1c levels in fresh whole blood (7.8 ± 3.2%) and dried blood (7.7 ± 2.3%). The average bias and absolute error rate were −1.2% and 11.0% between liquid and dried blood, respectively [Table 2], which was within acceptable TCL. The linear correlation between DBS and whole blood HbA1c is depicted in Figure 1a. The coefficient of determination (r2) was 0.955, which indicates a strong linear relationship between both assays (P < 0.0001). Bland–Altman analysis shows good agreement between the DBS and whole blood HbA1c levels [Figure 1b]. The mean difference in HbA1c levels between both assays was −0.096% (±1.08), and the limit of agreement (LoA) varied from −2.21 to 2.02%.
| Analytes | HbA1c (%) | hsCRP (mg/L) | Creatinine (mg/dL) | HDLC (mg/dL) | ||||
|---|---|---|---|---|---|---|---|---|
| Temperatures | 4°C | 4°C | 4°C | −20°C | −80°C | 4°C | −20°C | −80°C |
| n | 22 | 12 | 31 | 31 | 31 | 31 | 31 | 31 |
| Baseline (Liquid) mean (±SD) | 7.8±3.2 | 1.5±1.6 | 1.04±0.7 | 1.04±0.7 | 1.04±0.7 | 36.7±9.9 | 36.7±9.9 | 36.7±9.9 |
| Baseline (DBS) | ||||||||
| Mean (±SD) | 7.7±2.3 | 1.6±1.7 | 1.11±0.8 | 1.11±0.8 | 1.11±0.8 | 37.2±9.1 | 37.2±9.1 | 37.2±9.1 |
| Bias (%) | 1.2 | 4.6 | 7.4 | 7.4 | 7.4 | 1.5 | 1.5 | 1.5 |
| AER (%) | 11 | 18.9 | 15.8 | 15.8 | 15.8 | 15.2 | 15.2 | 15.2 |
| 3 months (DBS) | ||||||||
| Mean (±SD) | - | - | 1.02±0.7 | 1.12±0.7 | 1.08±0.7 | 36.6±10.3 | 37.2±11.7 | 36.6±9.5 |
| Bias (%) | - | - | 1.2 | 7.7 | 4.4 | 0.3 | 1.5 | 0.1 |
| AER (%) | - | - | 11.4 | 17.1 | 17.3 | 19.7 | 18.7 | 15.2 |
| 6 months (DBS) | ||||||||
| Mean (±SD) | - | - | 0.98±0.7 | 1.04±0.7 | 1.06±0.7 | 33.5±8.1 | 36.7±7.9 | 35.3±8.6 |
| Bias (%) | - | - | 4.8 | 0.1 | 5.6 | 8.8 | 0.1 | 0.2 |
| AER (%) | - | - | 13.7 | 17.8 | 13.4 | 14.4 | 16.0 | 12.6 |
| TCL (%) | 21.07 | 33.41 | 18.25 | 24.37 | ||||
Table shows the comparison of mean (±SD) levels of HbA1c, hs-CRP, and Creatinine, HDL-C and the bias and AER in their levels (DBS) at baseline, 3 months, and 6 months in DBS stored at 4°C, −20°C, and−80°C from baseline levels (Liquid). DBS: Dried blood spot, SD: Standard deviation, HbA1c: Glycated hemoglobin, hs-CRP: High-sensitivity C-reactive protein, HDL-C: High-density lipoprotein-cholesterol, AER: Average error rate, TCL: Total change limit
Validation of hs-CRP between serum and DBS bioassay
Hs-CRP was assessed in 20 samples in both serum and DBS to validate results in both matrices. The mean hs-CRP levels were 1.5 ± 1.6 mg/L and 1.6 ± 1.7 mg/L in plasma and DBS, respectively. The average bias (4.6%) and absolute error rate (18.9%) between plasma and dried blood were within the TCL limit [Table 2]. We observed a linear relationship (r2 = 0.973) and a strong positive correlation (r = +0.986, P < 0.0001) between both assays [Figure 1c]. The mean difference in hs-CRP levels between both assays was 0.11 mg/L, and the LoA varied from −0.66 to 0.44 mg/L; and a good agreement between hs-CRP levels analyzed in DBS and serum was observed in Bland–Altman analysis [Figure 1d].
Validation of creatinine levels in serum and DBS bioassay
Fresh blood samples were collected from 31 subjects, and creatinine was assessed in serum and dried spot at baseline to validate the DBS bioassay. Mean (±SD) creatinine levels were comparable in serum (1.04 ± 0.67 mg/dL) and dried blood (1.11 ± 0.8 mg/dL). The average bias and absolute error rate were 7.4% and 15.8% between liquid and dried blood, respectively [Table 2]. The linear correlation and Bland–Altman analysis for creatinine levels between serum and DBS are depicted in Figure 1e and f. The coefficient of determination (r2) was 0.935, which indicates a strong linear relationship between both assays, and a positive correlation (r = +0.967, P < 0.0001) was observed between both matrices. Bland–Altman analysis shows a good agreement between the creatinine levels of the DBS and serum sample (mean bias: 0.08 mg/dL; LoA: −0.29–0.45 mg/dL).
Validation of HDL-C between serum and DBS bioassay
HDL-C validation was performed in 38 samples. The mean HDL-C levels were (37.2 ± 9.1) in the DBS assay, which was comparable to mean serum HDL-C levels (36.7 ± 9.9 mg/dL) with acceptable average bias (1.5%) and absolute error rate (15.2%) between both matrices [Table 2]. The linear correlation between HDL-C levels between serum and DBS is depicted in Figure 1g. The coefficient of determination (r2) was observed to be 0.572, and a positive correlation (r = +0.756, P < 0.001) was observed. Bland–Altman analysis shows a good agreement between HDL-C levels of DBS and serum samples [Figure 1h].
Storage stability of creatinine and HDL-C levels in DBS
All blood spots used for validation of creatinine and HDL-C in DBS (n = 31) were also stored at 4°C, −20°C, and −80°C to check the storage stability in DBS after 3 months and 6 months. No significant difference was observed in mean creatinine and HDL-C levels after 3 months and 6 months of storage at all temperatures, and average bias and absolute error rate were within the acceptable change limit [Table 2].
DISCUSSION
In this study, we have developed the DBS assays for the estimation of analytes considered as cardiovascular risk biomarkers, including HbA1c, hs-CRP, creatinine, and HDL-C. We have established the analytical protocol and performed the validation of these analytes between dried and liquid specimens. Analytical parameters such as linearity, recovery, and accuracy intra- and inter-assay CV (<10%) for the DBS assay were within the accepted criteria as proposed by the US FDA.[4] A good recovery was obtained for HbA1c, hsCRP, creatinine, and HDL-C levels from filter paper. A direct comparison between levels of these analytes measured in DBS and liquid specimens demonstrated a good positive correlation in regression analysis, and a good agreement was obtained between HbA1c, hs-CRP, and creatinine values in dried and liquid samples in Bland–Altman analysis. However, a moderate correlation (r2 = 0.572) was revealed for HDL-C levels between liquid and DBS samples. The percentage bias and absolute error were within acceptable limits for all four analytes. The finding of our study suggests good suitability of the DBS assay for the measurement of these analytes.
The analytical parameters of DBS-based HbA1c assay include linearity of the calibration curve (r2 = 0.999), recovery (103%), intra-assay (5%), inter-assay precision (7.6%), and accuracy (bias = 0.78%), indicating an excellent performance. The results of the validation study between whole blood and DBS-based HbA1c bioassay revealed an excellent linear relationship (r2 = 0.955, P < 0.0001) and good agreement in Bland–Altman analysis (mean bias = −0.096%; LoA: −2.21–2.02%). The results of this study are consistence with the previous studies. A systematic review and meta-analysis of 16 studies by Affan et al.[9] described the comparative analysis based upon standard methods (affinity chromatography, high-performance liquid chromatography, immunoassay, immunoturbidimetric assay, liquid chromatography-mass spectrometry [LC-MS/MS], and ion-exchange chromatography) of HbA1c and lipid levels assessed in DBS and venous blood samples. The author reported a close agreement between the assays for HbA1c levels based on DBS and venous specimens (DBS = 0.9858 venous +0.3809), except for assays based upon affinity chromatography. The author speculated that the source of the variation and basis between the various analytic methods could be due to variability in the preparation of the DBS samples, such as storage, transportation, and the analytes extraction procedure between the analytic assays. The validation of DBS-based HbA1c assays was also performed on multiple chemistry analyzer platforms, and a high correlation was reported between DBS and venous values for the Bio-Rad, Olympus, and Roche platforms (correlation coefficients, all above r2 = 0.940). The author concluded that DBS-based HbA1c assays may be successfully implemented in community-based surveys.[10]
Hs-CRP is estimated using ELISA in the previous studies as well as in the present study, and has shown excellent performance in DBS-to-serum equivalency. An acceptable performance was revealed for the DBS-based hs-CRP assay in the present study, as evidenced by the analytical parameters. In the validation analysis for hs-CRP levels between dried and venous serum, a good correlation (r2 = 0.973) and agreement in Bland–Altman analysis (mean bias: 0.11 mg/L; LoA: −0.66–0.44 mg/L) confer the potential of DBS-based hs-CRP assay to be implemented in clinical and population-based studies. The DBS-based hsCRP assay has been reported in several studies. Schakelaar et al.[11] assessed C-reactive protein (CRP) levels in plasma and DBS prepared from capillary blood on a chemistry auto-analyzer. The author reported a good correlation in capillary DBS to plasma (r2 = 0.986) CRP levels with acceptable bias (0.045 mg/L) but with a large range of LoA (95% confidence interval [CI] = −3.78–3.87 mg/L) as indicated in Bland– Altman plot.
The DBS-based creatinine assay showed adequate analytical performance and we found a good correlation (r2 = 0.935, P < 0.0001) and agreement in Bland–Altman analysis (mean bias: 0.08 mg/dL; LoA: −0.29–0.45 mg/dL) for creatinine levels between serum and DBS in the present study. The validation of DBS-based creatinine assay has been reported in earlier studies. The results of our study are in accordance with previous studies. In a recent study, Scribel et al.[12] estimated vancomycin and creatinine levels in DBS and plasma concentrations by LC-MS/MS. The author reported a good analytical performance, indicated by accuracy (94.4–102.6%), intra-assay (2.1–5.6%), and inter-assay (3.5–7.0%) precision. In a pilot study, the creatinine levels in capillary serum and DBS collected by fingerpick were estimated using Mass Spectrometry,[13] and a good agreement (Spearman: r = 0.77, P < 0.01; Pearson: r = 0.56, P = 0.04; Bland Altman: Mean bias = 0.01 mg/dL; lower LoA = −2.18; upper LoA = 2.22 mg/dL) was reported with acceptable variance (standard error of measurement [SEM]: 88.7; CV: 10.7%; intraclass correlation coefficient (ICC): 0.57 [95% CI: 0.06–0.84]). Mathew et al.[14] reported that the mean (95% CI) bias was 1.00 (−2.73–4.72)% between venous and DBS for creatinine levels estimated by LC-MS/MS assay and validated with traditional venous sampling and 69.4% of patient values of creatinine were within the limits of clinical acceptance (within 15% agreement against the venous samples). In comparison to other studies, a strong correlation and agreement were revealed for creatinine levels between dried and liquid-based assays in our study.
A moderate association was revealed between venous and DBS measures for HDL-C in the present study (r2 = 0.572). The results of our study are in accordance with previous studies[15] but poor performance of DBS-based HDL-C assay was reported in some studies.[1,16-18] Samuelsson et al.[1] reported a poor correlation for HDL-C (r2 = 0.118). In other studies, Lacher et al.[16] reported poor correlation for total cholesterol (r2 = 0.34) and HDL-C (r2 = 0.30) and a strong correlation for hemoglobin A1c (r2 = 0.92), CRP (r2 = 0.84) and glucose (r2 = 0.81) between DBS and venous methods. Crimmins et al.[17] reported a little better correlation between venous and DBS samples for HDL-C levels from two sites of study, the University of Washington (r2 = 0.49) and the Heritage site (r2 = 0.314). In a recent study by Kumari et al.[18] CVD risk markers, including HbA1c, C-reactive protein, and lipids (total cholesterol, HDL-C, triglycerides), were measured in venous and DBS samples. A significant relationship between venous and DBS measures was observed for HbA1c (r2 = 0.41), CRP (r2 = 0.87), triglycerides (r2 = 0.37), and total cholesterol (r2 = 0.49), (P < 0.0001 for all,) in bivariate correlations and good agreement in Bland– Altman plots. However, this association was not revealed for HDL-C (Adjusted r2 = 0.005, p = 0.08) between venous and DBS measures. The author concluded that less successful measurement of HDL-C between venous and DBS measures may be related to the stability of HDL-C in the DBS samples. A poor correlation was also observed for other analytes except CRP, indicating that proper instructions may not be followed for the preparation of DBS, such as over-drying of spots during self-collected DBS sampling, resulting in lower recovery of analytes from DBS. Cholesterol measures (total cholesterol and HDL-C) have been shown to be sensitive to many pre-analytical factors, including small spots, shipping time, high temperature, and humidity.[19,20] We have found a good correlation and agreement in Bland–Altman analysis between venous and DBS measures for total cholesterol and triglyceride in our previously reported studies[21,22] wherein methanol was used to elute the analytes from DBS. HDL-Care complex particles are composed of a central hydrophobic core (60% lipids: Cholesterol ester and triglyceride), which is surrounded by a hydrophilic membrane consisting of multiple proteins (40%). The elution of these molecules from the DBS may be critical in a complex form. This could be a main reason for the weak correlation between venous and DBS measures for HDL-C. However, we tried a number of combinations of elution buffer/eluent, such as 20–80% methanol with PBS or water, ×1 PBS with 0.1% tween 20. The highest yield was obtained with ×1 PBS with 0.1% tween 20 and with reagent 1 (provided with kit). There is still a need to improve the analytical assay for estimation of HDL-C in DBS specimens.
We further observed that the creatinine was stable in DBS samples for 6 months at 4°C, −20°C, and −80°C. However, HDL-C was less stable in DBS during 6 months of storage at 4°C, compared to DBS samples stored at −20°C and −80°C. The HDL-C levels were reduced by −8.8% from baseline at 4°C during 6 months. However, deviation/bias and absolute error were within the acceptable limit (TCL = ±24.37%). Several studies have reported that creatinine is unaffected by storage at <10°C[23] and to the best of our knowledge, the storage stability of HDL-C in DBS has not been reported previously.
Limitations and strengths
The limitation of the study was that the blood samples in the present study were collected and DBS prepared in a well-controlled environment in a laboratory, not a field setting, and therefore do not mimic situations involving variability in spot size, prolonged shipping, storage, and delayed transportation. The strength of the study was that we followed the standard guidelines for the validation of the bioanalytical assay.
Translational potential
Transportation of blood samples from the site of collection to the laboratory is a challenge in epidemiological studies, particularly when they are multicentric, involve distant locations, and include large numbers of subjects, especially older adults and children. The measurement of biochemical analytes provides more objective measures as compared to questionnaire-based assessments, which are semi-quantitative for the risk assessment of chronic diseases. The measurement of these analytes in DBS, as attempted in this study, opens up the possibility of including biochemical analytes in large-scale studies, as these are easy to collect and transport.
CONCLUSIONS
We conclude that for measurement of HbA1c, hs-CRP, creatinine, and HDL-C, blood samples can be collected on filter paper and dried and are readily transferable to a liquid phase for analysis, offering a convenient alternative for monitoring of CVD risk factors.
Author contribution:
MT, RAA, RL, UK and TK: Contributed to the conception and design of the research; MT, RA, and AT: Conducted research; MT: Analyzed data and wrote the paper; RL, RAA, UK and TK: Evaluated the manuscript. All authors read and approved the final manuscript.
Ethical approval:
The research/study was approved by the Institutional Review Board at All India Institute of Medical Sciences, approval number IEC-293/June 01, 2018, dated 1st June 2018.
Declaration of patient consent:
The authors certify that they have obtained all appropriate patient consent.
Conflicts of interest
There are no conflicts of interest.
Use of artificial intelligence (AI)-assisted technology for manuscript preparation
The authors confirm that there was no use of artificial intelligence (AI)-assisted technology for assisting in the writing or editing of the manuscript and no images were manipulated using AI.
Financial support and sponsorship: This work was supported by the Indian Council of Medical Research, New Delhi, grant no. 54/1/GER/2017/-NCD-II, dated 27th August 2018.
References
- Validation of biomarkers of CVD risk from dried blood spots in community-based research: Methodologies and study-specific serum equivalencies. Biodemography Soc Biol. 2015;61:285-97.
- [CrossRef] [PubMed] [Google Scholar]
- Role of dried blood spots in health and disease diagnosis in older adults. Bioanalysis. 2014;6:3121-31.
- [CrossRef] [PubMed] [Google Scholar]
- Development and validation of assay protocols for use with dried blood spot samples. Am J Hum Biol. 2014;26:1-9.
- [CrossRef] [PubMed] [Google Scholar]
- Guidance for industry: Bioanalytical method validation. 2018. Available from: https://www.fda.gov/files/drugs/published/bioanalytical-method-validation-guidance-for-industry.pdf [Last accessed on 2024 Oct 17]
- [Google Scholar]
- Measurement of HbA1c in gingival crevicular blood using a high-pressure liquid chromatography procedure. Lab Med. 2015;46:290-8.
- [CrossRef] [PubMed] [Google Scholar]
- Serum, plasma, and dried blood spot high-sensitivity C-reactive protein enzyme immunoassay for population research. J Immunol Methods. 2010;362:112-20.
- [CrossRef] [PubMed] [Google Scholar]
- Feasibility of measuring sodium, potassium and creatinine from urine sample on dried filter paper. Bioanalysis. 2019;11:689-701.
- [CrossRef] [PubMed] [Google Scholar]
- Desirable specifications for total error, imprecision, and bias, derived from intra- and inter-individual biologic variation. 2014. Database for biological variation and desirable specifications for allowable error. Available from: http://www.westgard.com/biodatabase1.html [Last accessed on 2024 Oct 17]
- [Google Scholar]
- Comparability of HbA1c and lipids measured with dried blood spot versus venous samples: A systematic review and meta-analysis. BMC Clin Pathol. 2014;14:21.
- [CrossRef] [PubMed] [Google Scholar]
- Validation and modification of dried blood spot-based glycosylated hemoglobin assay for the longitudinal aging study in India. Am J Hum Biol. 2015;27:579-81.
- [CrossRef] [PubMed] [Google Scholar]
- Analysis of C-reactive protein from finger stick dried blood spot to predict high risk of cardiovascular disease. Sci Rep. 2023;13:2515.
- [CrossRef] [PubMed] [Google Scholar]
- Vancomycin and creatinine determination in dried blood spots: Analytical validation and clinical assessment. J Chromatogr B Analyt Technol Biomed Life Sci. 2020;1137:121897.
- [CrossRef] [PubMed] [Google Scholar]
- Creatine and creatinine quantification in olympic athletes: Dried blood spot analysis pilot study. Biol Sport. 2022;39:745-9.
- [CrossRef] [PubMed] [Google Scholar]
- Analytical and clinical validation of dried blood spot and volumetric absorptive microsampling for measurement of tacrolimus and creatinine after renal transplantation. Clin Biochem. 2022;105-6:25-34.
- [CrossRef] [PubMed] [Google Scholar]
- Collection and laboratory methods for dried blood spots for hemoglobin A1c and total and high-density lipoprotein cholesterol in population-based surveys. Clin Chim Acta. 2015;445:143-54.
- [CrossRef] [PubMed] [Google Scholar]
- Comparison of dried blood spot to venous methods for hemoglobin A1c, glucose, total cholesterol, high-density lipoprotein cholesterol, and C-reactive protein. Clin Chim Acta. 2013;422:54-8.
- [CrossRef] [PubMed] [Google Scholar]
- Validation of blood-based assays using dried blood spots for use in large population studies. Biodemography Soc Biol. 2014;60:38-48.
- [CrossRef] [PubMed] [Google Scholar]
- A randomised study of nurse collected venous blood and self-collected dried blood spots for the assessment of cardiovascular risk factors in the Understanding Society Innovation Panel. Sci Rep. 2023;13:13008.
- [CrossRef] [PubMed] [Google Scholar]
- Dried blood spots: Effects of less than optimal collection, shipping time, heat, and humidity. Am J Hum Biol. 2020;32:e23390.
- [CrossRef] [PubMed] [Google Scholar]
- Dried blood spot collection, sample quality, and fieldwork conditions: Structural validations for conversion into standard values. Am J Hum Biol. 2021;33:e23517.
- [CrossRef] [PubMed] [Google Scholar]
- Utility of dried blood spots for measurement of cholesterol and triglycerides in a surveillance study. J Diabetes Sci Technol. 2010;4:258-62.
- [CrossRef] [PubMed] [Google Scholar]
- Measurement of cholesterol and triglycerides from a dried blood spot in an Indian Council of Medical Research-World Health Organization multicentric survey on risk factors for noncommunicable diseases in India. J Clin Lipidol. 2012;6:33-41.
- [CrossRef] [PubMed] [Google Scholar]
- Dried blood spot analysis of creatinine with LC-MS/MS in addition to immunosuppressants analysis. Anal Bioanal Chem. 2015;407:1585-94.
- [CrossRef] [PubMed] [Google Scholar]