Evaluation of Dried Blood Spots for Human Immunodeficiency Virus Type 1 Drug Resistance Testing

Samples.We evaluated samples collected from two different sources. The first group of specimens included matched plasma, DBS, and DPS obtained from the Virology Quality Assessment program (VQA). Samples were generated from six subtype B HIV-infected donors and were part of a stability study of HIV-1 RNA on DBS (5). Samples were classified in two panels (A and B), with each panel consisting of three samples with a low, medium, or high virus load. Panel A was stored at −30°C for 6 years and contained samples with resistance-associated mutations. Panel B was stored at either room temperature (panel B1) or −70°C (panel B2) for 5 years and contained samples with wild-type viruses.

The second group of specimens included matched plasma and DBS samples from HIV-1-infected donors collected at Cameroonian blood banks (n = 40). These samples were collected as part of a surveillance study of divergent HIV variants and were prepared from units of blood that were found to be positive for HIV (13). Of these samples, 9 were classified as having subtype A, 1 as having subtype F, 1 as having circulating recombinant form 11 (CRF11), and 26 as having CRF02 based on the analysis of gp41 sequences. gp41 was not amplifiable in the remaining three samples. These three samples were classified as having subtypes CRF11 (two samples) and CRF06 (one sample) based on the analysis of p17 sequences. The DBS specimens had been stored at −20°C for a period of 2 to 3 years.

All blood and plasma spots were prepared by pipetting 50 μl of EDTA-anticoagulated whole blood or plasma onto premarked circles on 903 filter paper cards (Schleicher & Schuell, Keene, NH). Each card was dried overnight and was then individually packaged into Bitran zipper-lock bags (Fisher Scientific Company, Pittsburgh, PA) containing a silica gel pack (Mini Pax Sorbent; Multisorb Technologies, Buffalo, NY). Matched plasma samples were aliquoted and stored at −80°C or liquid nitrogen. Plasma virus loads in the Cameroonian specimens were quantified using the Amplicor HIV-1 Monitor assay (version 1.5; Roche Diagnostics, Branchburg, NJ). Plasma virus loads in the VQA panels were also quantified using the Amplicor assay (version 1.5 for panel A and version 1.0 for panel B) (5).

Nucleic acid extraction.HIV-1 RNA from plasma (100 μl) was extracted by using the Nuclisens silica-based extraction method according to the manufacturer's instructions (bioMerieux, Inc., Durham, NC). Total HIV-1 nucleic acids from DBS and DPS were also extracted using the Nuclisens method with some modifications for DBS and DPS processing. Briefly, a whole spot containing 50 μl of blood or plasma was cut with scissors and added into 9 ml of Nuclisens lysis buffer. After a 2-hour incubation at room temperature under gentle rotation, the supernatant was clarified by centrifugation at 250 × g for 5 min and then transferred to a clean 15-ml conical tube. Total nucleic acids were then extracted following the manufacturer's instructions, resuspended in 30 μl of elution buffer, and stored at −80°C until use.

Amplification by RT-nested PCR.A 1,023-base pair fragment of HIV-1 pol corresponding to amino acids 15 to 99 of the protease and 1 to 256 of the RT was amplified from plasma, DBS, and DPS using an in-house reverse transcriptase (RT)-nested PCR method. This assay has been previously validated with plasma samples from patients infected with HIV subtype B and has a sensitivity of detection of 1,000 RNA copies/ml (11). Briefly, a 5- to 10-μl volume of extracted nucleic acids was added to an RT cocktail containing murine leukemia virus (MuLV) RT enzyme (Applied Biosystems) and reverse primer RT-gen.4R (5′-ATC CCT GCA TAA ATC TGA CTT GC-3′). The RT reaction was done for 1 h at 39°C. After a first round of PCR amplification using primers RT-gen4R and Pro-Out.3F (5′-CCT CAG ATC ACT CTT TGG CAA CG-3′), 4 μl was subjected to a second round of PCR amplification using internal primers PR/RT.2F (5′-GAT CAC TCT TTG GCA ACG ACC CAT-3′) and 215/219.3R (5′-CTT CTG TAT GTC ATT GAC AGT CC-3′). Cycling conditions for the first and second amplifications were 4 min at 95°C followed by 40 cycles of 45 s at 95°C, 30 s at 50°C, and 2 min at 72°C. Parallel RT-nested PCRs were also done in the absence of MuLV RT to selectively amplify DNA. Amplifications were done with a T3 thermocycler (Biometra GmbH, Goettingen, Germany).

Sequence analysis.Sequence analysis of HIV-1 pol was done in an ABI 3100 capillary sequencer using primers A35, AV36, AV44, 215/219.3R, HIV 90 (5′-AAT GCT TTT ATT TTT TCT TCT GTC AAT GGC-3′), and PR/RT.2F (19). The Vector NTI program (suite 8) was used to analyze the data and calculate amino acid and nucleotide similarities. Protease and RT genotypes were interpreted using the Stanford Genotypic Resistance Interpretation Algorithm (http://hivdb.stanford.edu/pages/algs/HIVdb.html ) and the list of resistance mutations included in the International AIDS Society USA mutation figures (12).

Nucleotide sequence accession numbers.The nucleotide sequences from 48 matched plasma and DBS specimens have been deposited in the GenBank database (accession numbers DQ869190 to DQ869237).

Amplification and sequencing of large HIV-1 pol fragments from DBS and DPS stored at different temperatures.To investigate the impact of storage temperature on the ability to amplify from DBS, we evaluated the two VQA panels generated and stored under defined conditions. Table 1 shows the efficient amplification of pol in plasma and the corresponding DBS samples from panel A, indicating that long-term storage at −30°C was sufficient to preserve the integrity of nucleic acids on these DBS. Interestingly, all three DBS specimens from panel A were also amplifiable in the absence of reverse transcription, demonstrating the presence of proviral DNA (Table 1). In contrast to the case for plasma and DBS, amplification from panel A DPS samples was possible only for the specimen with the highest plasma virus load (A3; 338,112 RNA copies/ml). As expected, none of the DPS specimens were amplifiable in the absence of reverse transcription, consistent with the absence of proviral DNA in DPS.

Table 1 also shows the efficiencies of amplification of pol sequences from panel B samples that were stored at either room temperature (B1) or −70°C (B2). When the DBS were stored at room temperature for 5 years, none of the specimens were amplified either in the absence or the presence of reverse transcription. However, when the same specimens were stored at −70°C for the same period of time, pol sequences were recovered in the two DBS that had the highest plasma virus load. As we noted for panel A samples, only the DPS specimen obtained from the sample with the highest plasma virus load (B2.3; 57,375 RNA copies/ml) was successfully amplified in the presence of reverse transcription. Overall, these findings suggest that storage temperature is an important determinant for the efficient amplification of pol sequences from DBS stored for prolonged periods of time.

We next sequenced all five positive DBS samples from panels A and B and compared the sequences with those seen in viruses from plasma. The mean (± standard deviation) similarity between nucleotide pol sequences from plasma and DBS was 98.6% ± 1.14% and ranged between 97% and 100%. A high similarity between amino acid sequences was also noted (mean ± standard deviation, 98.8% ± 1.6%; range, 96% to 100%), further indicating a high concordance between plasma and DBS sequences. Samples from panel A had drug resistance mutations which allowed us to compare the genotypic profiles between plasma and DBS. When this analysis was done, we noticed an overall agreement between the resistance profiles from plasma and DBS (Table 2). Of the 17 protease and RT mutations seen in plasma viruses, 14 were also found in viruses from DBS. Of the three mutations that were absent in DBS sequences, one (Y181C) was a major nonnucleoside RT mutation and two were minor protease mutations (M36L and M36I) that are weakly associated with protease resistance only when present with other mutations. Conversely, all 14 mutations seen in DBS sequences were also found in plasma viruses. Taken together, these findings indicate an overall concordance between protease and RT sequences from plasma and DBS. We also analyzed the patterns of mutations seen in HIV DNA from DBS. Table 2 shows that the T69N mutation from plasma A.1 and the Y181C and M36L mutations from plasma A.2 were undetectable in DNA sequences from DBS.

Amplification of HIV-1 pol from DBS collected in Cameroon.We next evaluated the efficiency of amplification of protease and RT sequences from DBS prepared under field conditions and stored at −20°C. We selected 40 matched plasma/DBS samples from Cameroon representing a wide range of plasma virus load. The median plasma virus load in these samples was 23,715 RNA copies/ml and ranged from 665 to 645,256 RNA copies/ml. Of these samples, 11 had virus loads between 100 and 10,000 RNA copies/ml, 13 had virus loads between 10,000 and 50,000 RNA copies/ml, and 16 had virus loads above 50,000 RNA copies/ml. Of the 40 plasma specimens, HIV-1 pol sequences were successfully amplified in 37 samples (92.5%) and 3 were consistently negative after repeated testing (Table 3). Of the negative plasma samples, one (sample 19) had a virus load of 16,515 RNA copies/ml and the remaining two (samples 1 and 2) had virus loads of 665 and 2,271 RNA copies/ml.

We also investigated the amplification of HIV pol sequences from these DBS in the absence of reverse transcription. Similar to that seen in the VQA panels, a significant proportion of the RT-PCR positive samples (24 of 37; 64.9%) were amplifiable in the absence of reverse transcription, demonstrating the presence of proviral DNA. Interestingly, amplification of HIV DNA was more frequent among samples with high plasma virus loads. Table 3 shows that only 11 of the 21 (52.4%) DBS samples with plasma virus loads between 1,000 and 50,000 RNA copies/ml were DNA positive, compared to 13 of 16 (81%) samples with virus loads greater than 50,000 RNA copies/ml. These findings suggest that the DNA contribution to DBS sequences may vary with plasma virus load levels.

We next compared pol sequences from plasma and DBS in a subset of 24 matched samples that had complete RT and protease sequences (data not shown). Similar to that seen in the VQA specimens, the similarity between nucleotide sequences from plasma and DBS was high (mean ± standard deviation, 98.5% ± 1.4%; range, 95% to 100%). A high amino acid similarity was also noted (mean ± standard deviation, 98.0% ± 1.8%) with only one plasma/DBS pair showing a similarity of 94% (patient 8). This particular pair had 10-amino-acid differences between positions 96 to 109 of the RT.

The analysis of protease and RT genotypes in all the 24 plasma and DBS samples showed absence of any major resistance mutations (Table 4). Table 4 also shows a high concordance between minor protease mutations, and only one sample (sample 10) had a K20I mutation in plasma but not in DBS. Two additional samples had unusual amino acids changes at RT positions 210 (sample 4) or 184 and 118 (sample 20) in DBS but not in plasma sequences (Table 4).