Comparison of Ahlstrom Grade 226, Munktell TFN, and Whatman 903 Filter Papers for Dried Blood Spot Specimen Collection and Subsequent HIV-1 Load and Drug Resistance Genotyping Analysis

Abstract

Dried blood spots (DBS) collected onto filter paper have eased the difficulty of blood collection in resource-limited settings. Currently, Whatman 903 (W-903) filter paper is the only filter paper that has been used for HIV load and HIV drug resistance (HIVDR) testing. We therefore evaluated two additional commercially available filter papers, Ahlstrom grade 226 (A-226) and Munktell TFN (M-TFN), for viral load (VL) testing and HIVDR genotyping using W-903 filter paper as a comparison group. DBS specimens were generated from 344 adult patients on antiretroviral therapy (ART) in Botswana. The VL was measured with NucliSENS EasyQ HIV-1 v2.0, and genotyping was performed for those specimens with a detectable VL (≥2.90 log₁₀ copies/ml) using an in-house method. Bland-Altman analysis revealed a strong concordance in quantitative VL analysis between W-903 and A-226 (bias = −0.034 ± 0.246 log₁₀ copies/ml [mean difference ± standard deviation]) and W-903 and M-TFN (bias = −0.028 ± 0.186 log₁₀ copies/ml) filter papers, while qualitative VL analysis for virological failure determination, defined as a VL of ≥3.00 log₁₀ copies/ml, showed low sensitivities for A-266 (71.54%) and M-TFN (65.71%) filter papers compared to W-903 filter paper. DBS collected on M-TFN filter paper had the highest genotyping efficiency (100%) compared to W-903 and A-226 filter papers (91.7%) and appeared more sensitive in detecting major HIVDR mutations. DBS collected on A-226 and M-TFN filter papers performed similarly to DBS collected on W-903 filter paper for quantitative VL analysis and HIVDR detection. Together, the encouraging genotyping results and the variability observed in determining virological failure from this small pilot study warrant further investigation of A-226 and M-TFN filter papers as specimen collection devices for HIVDR monitoring surveys.

INTRODUCTION

In response to the increased coverage of antiretroviral therapy (ART) in resource-limited settings, the World Health Organization (WHO) has developed a strategy for HIV drug resistance (HIVDR) prevention and assessment (1). This strategy provides guidelines for monitoring the development and transmission of HIVDR variants at the population level by assessing the viral load (VL) and HIVDR genotype of HIV-1-infected patients commencing ART in a prospective cohort (2). Such surveys are essential for maintaining the efficacy of antiretroviral drug regimens within the population in areas where individualized patient monitoring is not available.

Plasma is the gold-standard specimen type for HIVDR surveys and the only specimen type that is currently recommended by WHO for monitoring patients on ART (3). Due to the increased cost and logistical challenges of separating and maintaining frozen plasma specimens, several studies investigated alternative specimen types for VL analysis and HIVDR genotyping (reviewed in reference 4). Dried blood spots (DBS) require minimal technical skills to collect and do not require cold-chain transportation (3) and therefore have been the most widely studied alternative specimen type (5–15). DBS have produced results similar to those of plasma for quantitative VL analysis (4, 16–18) and HIVDR genotyping analysis (5–15, 19, 20) for both patients on ART and treatment-naive patients, indicating that DBS could be a viable alternative to plasma for HIVDR monitoring surveys (19).

Whatman 903 (W-903) filter paper is the predominant filter paper used for DBS collection in studies analyzing HIV-1 loads (4, 16, 18) and the only filter paper used for HIVDR genotyping (4–15, 19–21). With ART rapidly expanding, the number of HIVDR monitoring surveys is increasing in resource-limited countries where DBS are the preferred specimen type, thus creating a greater demand for filter paper. Having only one source of recommended filter paper for HIVDR monitoring surveys may lead to shortages if unforeseen manufacturing difficulties occur. Diversifying the type of filter papers available for HIVDR monitoring surveys could also decrease costs through price competition and increase the availability of filter paper, thus positively impacting low-income countries. We therefore selected two test filter papers, Ahlstrom grade 226 (A-226) and Munktell TFN (M-TFN), and conducted a small pilot evaluation study. Ahlstrom grade 226 filter paper is registered with the U.S. Food and Drug Administration (FDA) as a class II medical device, and Munktell TFN filter paper meets the Clinical and Laboratory Standards Institute (CLSI) 2007 requirements for blood collection onto filter paper (22). This study describes a direct comparison of the utilities of DBS specimens prepared by using the standard (W-903) and test (A-226 and M-TFN) filter papers for VL determination and HIVDR genotyping analysis for HIV-1-infected patients on ART.

MATERIALS AND METHODS

Specimen collection and storage.

Between October and November 2010, DBS specimens were collected from 344 HIV-positive patients who were reported to be on three-drug-combination ART. DBS were collected from remnant whole-blood specimens that were sent to the Nyangabgwe HIV Reference Laboratory in Francistown, Botswana, for clinical CD4 testing. No personal (including duration on ART) or demographic information was collected from these patients for this study. Blood was stored at an ambient temperature (median temperature = 31°C; median humidity = 33%) for an average of 1.30 ± 0.62 days (range, <1 to 3 days) prior to DBS preparation. DBS were prepared by pipetting 100 μl of whole blood onto Whatman 903 (Whatman, Springfield Mill, United Kingdom), Ahlstrom grade 226 (Ahlstrom Corporation, Helsinki, Finland), and Munktell TFN (Munktell Inc., Raleigh, NC) filter papers. DBS were allowed to dry overnight at an ambient temperature. Glassine paper was then folded around each DBS card, and 10 to 25 cards were packaged in a Bitran bag with desiccant packs and a humidity indicator card. DBS were stored at −80°C prior to shipment on dry ice to the WHO Specialized Drug Resistance Laboratory at the Centers for Disease Control and Prevention (CDC), Atlanta, GA. All specimens were stored at −80°C upon arrival at the CDC. Based on current recommendations by the WHO, DBS can be stored at an ambient temperature for up to 14 days after collection (3); however, we chose to store them at −80°C for this initial evaluation of new filter papers, as this is the optimal DBS storage condition for HIVDR genotyping (3).

Nucleic acid extraction and HIV-1 load analysis.

One DBS spot was cut out per specimen and placed into 2 ml of NucliSENS lysis buffer (Biomerieux, Durham, NC) for 30 min at room temperature with gentle rotation. Nucleic acid was then extracted from all specimens by using the NucliSENS EasyMag (Biomerieux, Durham, NC) automated extraction system according to the manufacturer's instructions. Nucleic acid was eluted in 25 μl of NucliSENS extraction buffer 3 and either immediately used for HIV-1 load or HIVDR genotyping analysis or stored at −80°C until use. The HIV-1 load was determined by the NucliSENS EasyQ automated system using NucliSENS EasyQ HIV-1 v2.0 RUO test kits (Biomerieux, Durham, NC), according to the manufacturer's instructions. The linear range of this assay is 500 to 21,000,000 copies/ml for DBS specimens, and a lower detection limit is 802 copies/ml when a DBS spot containing 100 μl of whole blood is used (23).

HIV-1 drug resistance genotyping.

Genotyping of the protease and reverse transcriptase (RT) regions of the HIV-1 pol gene was performed by using a broadly sensitive in-house genotyping assay described in detail previously (13, 20). Briefly, a 1,084-bp segment of the 5′ region of the pol gene was generated by RT-PCR followed by nested PCR. This fragment was purified, sequenced by using the BigDye Terminator v3.1 cycle sequencing kit (Applied Biosystems, Foster City, CA), and analyzed on an ABI Prism 3730 genetic analyzer (Applied Biosystems). The ReCALL software program was used to edit the raw sequences and generate consensus sequences (24). HIVDR mutations and drug susceptibility profiles (HIVdb algorithm) were determined by using the Stanford University Drug Resistance Database (Palo Alto, CA).

Statistical analysis.

All HIV-1 load values were log₁₀ transformed prior to analysis. Quantitative variables are expressed as means ± standard deviations (SD), unless otherwise stated, and were compared by the Wilcoxon signed-rank test. Qualitative VL for determining virological failure was assessed using a kappa statistic. Kappa values were categorized as poor (<0.40), good (0.4 to 0.75), or excellent (>0.75) (25). Concordance between the VL on the test filter papers (A-226 or M-TFN) and the VL on the gold standard (W-903) was determined by Bland-Altman analysis (26). Statistical calculations were performed by using GraphPad Prism software (version 5.0; GraphPad Software, La Jolla, CA).

RESULTS

Quantitative HIV-1 load analysis.

The primary objective of this pilot study was to compare two test filter papers (M-TFN and A-226) to the current gold standard (W-903) for the purpose of HIVDR monitoring surveys. We first compared quantitative HIV-1 load results obtained from DBS specimens spotted onto the two test filter papers with those obtained from DBS specimens collected onto the control W-903 filter paper. Of the 344 specimens, we identified 24 specimens with a detectable VL (VL of ≥2.90 log₁₀ copies/ml) in all three filter paper types. Such low numbers of specimens with detectable VLs were expected, as the majority of patients sampled for this study were virally suppressed due to ART. Of these 24 specimens, the mean log₁₀ VLs ± SD for W-903, A-226, and M-TFN filter papers were 4.10 ± 0.77, 4.13 ± 0.76, and 4.13 ± 0.70, respectively. The mean log₁₀ VLs for A-226 and M-TFN filter papers were not statistically different from the VL for W-903 filter paper by the Wilcoxon signed-rank test, with P values of 0.4838 and 0.4652, respectively. Bland-Altman analysis also demonstrated that A-226 and M-TFN filter papers were comparable to W-903 filter paper for VL measurements, with all of the data points lying within the 95% limits of agreement and minimal biases of −0.034 ± 0.246 log₁₀ copies/ml (mean difference ± SD) for A-226 filter paper and −0.028 ± 0.186 log₁₀ copies/ml for M-TFN filter paper (Fig. 1A and B).

Fig 1.

Bland-Altman analysis of HIV-1 load quantification between A-226 (A) or M-TFN (B) and W-903 filter papers. Nucleic acid was extracted from DBS specimens by using the NucliSENS EasyMag automated system, and NucliSENS HIV-1 v2.0 RUO kits were used to determine the HIV-1 load using the NucliSENS EasyQ analyzer. Twenty-four specimens with a detectable viral load (VL) (defined as a VL of ≥2.90 log₁₀ copies/ml) with all three types of filter paper were included in the analyses. The solid line represents the mean difference between W-903 and A-226 (−0.034 ± 0.246 log₁₀ copies/ml) or W-903 and M-TFN (−0.028 ± 0.186 log₁₀ copies/ml) filter papers, and the dotted lines represent the 95% limits of agreement (mean difference ± 1.96 standard deviations) for A-226 (−0.517 and 0.449; width = 0.965) and M-TFN (−0.393 and 0.336; width = 0.728) filter papers.

Qualitative HIV-1 load analysis.

We next evaluated whether DBS collected onto A-226 and M-TFN filter papers could accurately identify patients experiencing virological failure for the purpose of HIVDR monitoring surveys by comparing A-226 and M-TFN filter papers to W-903 filter paper with 2-by-2 analyses. Virological failure was defined as a VL of ≥3.00 log₁₀ copies/ml, as recommended by WHO (2). Specimens collected onto A-226 filter paper correctly identified 25 out of 35 patient samples with a VL of ≥3.00 log₁₀ copies/ml, resulting in a sensitivity of 71.43% (Table 1). Similarly, specimens collected onto M-TFN filter paper accurately identified virological failure in 23 out of 35 patient samples (65.71% sensitivity). Kappa statistic values (means ± standard errors [SE]) of 0.682 ± 0.067 (A-226) and 0.677 ± 0.069 (M-TFN) demonstrated good agreement between the test filter papers and the gold standard of W-903 filter paper (Table 1).

Table 1.

Qualitative HIV-1 load performance of DBS collected onto A-226 and M-TFN filter papers compared to W-903 filter paper for determining virological failure^a

Filter paper	VL (log10 copies/ml)	No. of specimens on W-903 filter paper			Kappa value ± SE
		VL ≥ 3.00	VL < 3.00	Total
A-226	≥3.00	25	10	35	0.682 ± 0.067
	<3.00	10	298	308
Total		35	308	343b
M-TFN	≥3.00	23	7	30	0.677 ± 0.069
	<3.00	12	297	309
Total		35	304	339c

^aVirological failure is defined as a VL of ≥3.00 log₁₀ copies/ml. For the comparison of A-226 filter paper and W-903 filter paper, the sensitivity, specificity, positive predictive value, and negative predictive value were 71.43, 96.75, 71.40, and 96.75%, respectively; for the comparison of M-TFN filter paper and W-903 filter paper, the sensitivity, specificity, positive predictive value, and negative predictive value were 65.71, 97.70, 76.64, and 96.13%, respectively.

^bOne specimen was excluded from the analyses due to invalid VL results.

^cFive specimens were excluded from the analyses due to invalid VL results.

HIV drug resistance genotyping analysis.

We next evaluated the suitability of specimens collected onto the A-226 and M-TFN filter papers for use in HIVDR genotyping. The WHO recommends the use of a VL cutoff of 3.00 log₁₀ copies/ml to define virological failure for HIVDR monitoring surveys of patients on ART (2); therefore, only those 24 specimens with a VL of ≥3.00 log₁₀ copies/ml with at least one type of filter paper but having a VL of ≥2.90 log₁₀ copies/ml with all three types of filter paper (the lower detection limit of DBS VL measurements using the NucliSENS EasyQ HIV-1 v2.0 RUO test kit) were genotyped. The suitability of A-226 and M-TFN filter papers was assessed by comparing the genotyping efficiencies, nucleotide sequence identities, and mutation profiles obtained from specimens collected onto the test filter papers to those from specimens collected onto W-903 filter paper. Of the 24 specimens for which genotyping was performed, those collected onto M-TFN filter paper had the highest genotyping efficiency (24/24; 100.0%), followed by W-903 and A-226 filter papers (22/24; 91.7%) (Table 2). Among these 24 specimens, we identified 21 specimens from which we were able to obtain a genotype with all three types of the filter paper, which are referred to as matched specimens, while the remaining 3 specimens for which genotyping was successful for only one (specimen 23) or two (specimens 64 and 312) types of the filter paper are referred to as unmatched specimens (Table 3). Direct comparison of the nucleotide sequences obtained from these 21 matched specimens via nucleotide sequence identity calculations revealed a high level of concordance between W-903 and A-226 (98.99% ± 0.98%) or W-903 and M-TFN (99.14% ± 0.86%) filter papers (Table 2).

Table 2.

HIV-1 drug resistance genotyping performance of DBS collected onto W-903, A-226, and M-TFN filter papers with viral loads of ≥3.00 log₁₀ copies/ml in at least one type of the three filter papers

Filter paper	No. of DBS meeting genotyping criteria	No. of DBS successfully genotyped	Genotyping efficiency (%)	Mean % nucleotide sequence identity to W-903 result ± SD
W-903	24	22	91.67
A-226	24	22	91.67	98.99 ± 0.98
M-TFN	24	24	100	99.14 ± 0.86

Table 3.

HIV-1 drug resistance mutation profiles for DBS collected onto W-903, A-226, and M-TFN filter papers^a

Specimen	Mutation(s) identified
	W-903		A-226		M-TFN
	NRTI	NNRTI	NRTI	NNRTI	NRTI	NNRTI
32
35	D67N, K70R, M184V, T215ST, K219Q	Y181C	D67DN, K70KR, M184V, T215FIST, K219KQ	Y181C	D67N, K70R, M184V, T215FIST, K219Q	A98AG, Y181C
62	D67N, M184V	L100I, K103N	D67N, M184V	L100I, K103N, P225HP	D67N, M184V	L100I, K103N
69
90	M184V	K103N	M184V	K103N	M184V	K103N
102						V106MV
130
136	D67DN, M184V	K103N, V106M	D67DN, M184V	K103N, V106M	D67DN, M184V	K103N, V106M
155		K103KN		K103KN		K103KN
169	M184V	Y188L	M184V	Y188L	M184V	Y188L
182		K103N, V106MV		K103N, V106MV		K103N, V106MV
183	M184V	Y188L, K238N	M184V	Y188L, K238N	M184V	Y188L, K238N
228
255
263
269	M41L, M184V, T215Y	A98G, Y188L	M41L, M184V, T215Y	Y188L	M41L, M184V, T215Y	Y188L
287	M184V	V106M, V179D	M184V	V106M, V179D, F227FL	M184V	V106M, V179D
293
295	K65R, D67N, Y115F, K219EK	K103N, V106M	K65R, D67N, Y115F	K103N, V106M	K65R, D67N, Y115F	K103N, V106M
309
317	M184MV	A98AG, K103KN	M184V	A98G, K103N	M184V	A98G, K103N
23b	No PCR product		No PCR product		M184V	K101E, G190A
64b	D67N, K70R, M184V, T215F, K219Q	V106M, Y181C	No PCR product		D67N, K70R, M184V, T215F, K219Q	V106M, Y181C
312b	No PCR product

^aHIV-1 drug resistance genotyping analyses of the pol region were performed for all the DBS specimens with a detectable viral load (≥2.90 log₁₀ copies/ml) with all three types of the filter papers using the broadly sensitive genotyping assay (13, 20). Drug resistance mutations against nucleoside reverse transcriptase inhibitors (NRTI) and nonnucleoside reverse transcriptase inhibitors (NNRTI) were identified by using the HIVdb program, and HIV-1 drug resistance profiles were determined by the HIValg program using the HIVdb algorithm at the Stanford HIV Drug Resistance Database website. Discordant mutations that were identified in only one type of filter paper are shown in boldface type. Specimens that had a difference in drug susceptibility ratings with one of the filter paper types are shaded.

^bUnmatched DBS specimens.

The primary objective of an HIVDR monitoring survey is to determine the percentage of patients who are failing treatment due to acquired HIVDR mutations. To assess whether A-226 and M-TFN filter papers would perform similarly to W-903 filter paper for the purpose of these surveys, we directly compared the mutation profiles and the expected drug susceptibility profiles of specimens collected onto each type of filter paper. Of the 21 matched specimens that were genotyped, 41 nucleoside reverse transcriptase inhibitor (NRTI) (n = 10) and nonnucleoside reverse transcriptase inhibitor (NNRTI) (n = 12) HIVDR mutations were detected in specimens collected onto each of the filter paper types (Table 3). Among the 21 matched specimens, we found 1 discordant NRTI mutation detected for a specimen on W-903 filter paper but not on the test filter papers and 5 discordant NNRTI mutations detected for only one of the three types of filter paper (Table 3). All of these discordant mutations were the result of base mixtures. Only one NNRTI mutation at position V106MV resulted in a change in the drug susceptibility profile. This mutation was identified on the M-TFN test filter paper and not the control paper (Table 3), indicating that M-TFN filter paper not only is comparable to W-903 filter paper but also appears to be more sensitive for this particular specimen type in identifying a major drug resistance mutation. In two of the three unmatched specimens, major NRTI and NNRTI drug resistance mutations were missed with at least one of the filter paper types (A-226 and/or W-903) due to the lack of amplification (Table 3).

DISCUSSION

In this study, we sought to determine whether A-226 and M-TFN filter papers were comparable to W-903 filter paper for the purpose of HIVDR monitoring surveys of HIV patients on ART. To our knowledge, this is the first study to report HIV-1 load and HIVDR genotyping results from A-226 and M-TFN filter papers. Quantitative VL analysis via the Wilcoxon signed-rank test and Bland-Altman analysis (Fig. 1) demonstrated similar results between A-226, M-TFN, and W-903 filter papers. Qualitative analysis using a VL of ≥3.00 log₁₀ copies/ml as a cutoff for virological failure illustrated “good” agreement between A-226 or M-TFN and W-903 filter papers according to the kappa statistic analyses but lower-than-expected sensitivities of 71.43% for A-226 filter paper and 65.71% for M-TFN filter paper (Table 1). The genotyping efficiency of specimens meeting the criteria of having a detectable VL (≥2.90 log₁₀ copies/ml) and having a VL of ≥3.00 log₁₀ copies/ml with at least one of the three filter papers was excellent for specimens collected onto M-TFN filter paper (100.00%) and was slightly reduced for specimens collected onto W-903 and A-226 filter papers (91.7%) (Table 2). A direct comparison of DR mutation profiles illustrated that despite minor differences in the detection of base mixtures, both A-226 and M-TFN filter papers performed similarly to W-903 filter paper in detecting HIVDR mutations in matched specimens. Overall, these data indicate that there are no major or consistent deficiencies of A-226 or M-TFN filter paper as a dried-whole-blood collection method for performing HIV-1 load analysis and HIVDR genotyping for the purpose of monitoring surveys. Our data do, however, illustrate the variability in qualitative VL analysis using DBS specimens, which, coupled with the small sample size of this study, advocates the need for further studies comparing DBS collected onto different types of filter papers to the gold standard of plasma specimens.

One limitation of this study was the lack of a plasma specimen control to serve as a gold standard for VL and genotyping analyses. Because we utilized leftover patient samples, the blood specimens were stored at an ambient temperature for an average of 1.30 ± 0.62 days (range, <1 to 3 days) before being made available for this study, thus surpassing the length of time for blood storage prior to plasma separation recommended by the manufacturer for VL quantification (23) and inhibiting the collection of plasma specimens. The analyses of these data were therefore compromised by the lack of a true gold-standard specimen, as we had to use a substandard specimen for comparison. To compensate for the absence of a gold-standard specimen, we limited our VL and genotyping analyses to specimens that had a detectable VL with all three filter paper types based on the lower detection limit (2.90 log₁₀ copies/ml) of the VL assay used and thus were likely to be true virological failures.

To date, only one study assessing A-226 and M-TFN filter papers for HIV-1 detection in infants using DNA PCR has been reported (S. Masciotra, S. Khamadi, A. Ramos, and S. Subbarao, presented at the 16th Conference on Retroviruses and Opportunistic Infections, Montreal, Canada, 2009) and demonstrated that A-226 and M-TFN filter papers were comparable to W-903 filter paper for qualitative HIV-1 DNA PCR analysis for HIV-1 detection in infants. In the current study, qualitative VL analysis was performed by using RNA instead of DNA and revealed lower-than-expected sensitivity values (71.43% for A-226 and 65.71% for M-TFN) for the test filter papers compared to W-903 filter paper in their ability to detect virological failure, defined as a VL of ≥3.00 log₁₀ copies/ml. We believe that these low values are the result of false-positive (falsely indicating virological failure) VL results from W-903 filter paper, as 8 out of the 10 specimens with a VL of ≥3.00 log₁₀ copies/ml from W-903 filter paper and with a VL of <3.00 log₁₀ copies/ml from A-226 filter paper did not amplify during the genotyping process. Likewise, 9 out of 12 specimens that were identified as having a VL of ≥3.00 log₁₀ copies/ml from W-903 filter paper and a VL of <3.00 log₁₀ copies/ml from M-TFN filter paper also did not amplify. We have previously observed such false-positive VL results for DBS specimens when we compared virological failure determinations between DBS and the gold standard plasma (19). Similar to our current study, these false-positive VL specimens did not amplify during the HIVDR genotyping procedure (19). Importantly, false-positive virological failure determinations with DBS specimens would result in more specimens being genotyped for an HIVDR monitoring survey but would not affect the outcome of the survey. Interestingly, when M-TFN filter paper was used as the gold standard for 2-by-2 analyses instead of W-903 filter paper, the sensitivity of A-226 filter paper increased substantially, to 83.8% (data not shown), and the kappa value increased from 0.682 ± 0.067 to 0.752 ± 0.061 (mean ± SE). Given that M-TFN filter paper had the highest genotyping efficiency (100.00%) of the three filter papers (Table 2), M-TFN filter paper may have been a better standard than W-903 filter paper for the detection of virological failure of HIV patients on ART in this study.

DBS specimens are emerging as the method of choice for specimen collection in resource-limited settings and may soon replace plasma for HIVDR monitoring surveys. To date, W-903 filter paper is the only filter paper that has been used for VL measurements and HIVDR genotyping (4–15, 20, 21). Having only one type of filter paper available could lead to increased costs and presents a greater potential for product shortages. M-TFN and A-226 filter papers are commercially available for use in newborn screening, and here, we demonstrated that they could be used for HIVDR monitoring surveys as well. Due to the novelty of these data, the small sample size of this pilot study, and the discussed limitations, further study is needed to determine the viability of A-226 and M-TFN filter papers for VL measurement and HIVDR genotyping in HIVDR monitoring surveys recommended by the WHO for resource-limited settings.