ORIGINAL ARTICLE

Genotypic testing for HIV-1 drug resistance using dried bloodsamples

Rosalia Lira • Hilda Valdez-Salazar • Guillermo Vazquez-Rosales • Othon Rojas-Montes •

Martha Ruiz-Tachiquin • Rocio Torres-Ibarra • Carlos Cano-Dominguez •

Angelica Maldonado-Rodrıguez • Alejandro Gomez • Onofre Munoz • Ma-Teresa Alvarez-Munoz

Received: 19 November 2009 / Accepted: 10 May 2010! Springer-Verlag 2010

Abstract In third-world countries, dried blood samples(DBS) are a convenient alternative to plasma for moni-

toring viral load during HIV-1 therapy. In this study, we

evaluated the feasibility of using DBS to perform HIV-1drug resistance genotyping in a ViroSeq assay in which the

protease and reverse transcriptase regions of the pol geneare analyzed. Fifty-seven antiretroviral genotypes fromplasma samples were tested, and drug resistance genotypes

were determined. Only 38.6% paired DBS samples were

sequenced. Failure to amplify DNA from DBS samplesgenerally correlated with plasma viral loads below log105.1. The majority of the mutations identified in plasma polsequences were also found in their DBS counterpart, with aconcordance in genotype interpretation of 96.4%. Several

factors were identified that could potentially improve both

the sensitivity and the quality of genotype data, such as

sample storage conditions and sequence analysis. There-fore, DBS sampling is useful to determine viral load and

drug resistance genotypes in HIV patients.

AbbreviationsDBS Dried blood sampleP Plasma

HAART Highly active antiretroviral therapy

FDA Federal Drug AdministrationPR Protease

RT Reverse transcriptase

NRTI Nucleoside/nucleotide RT inhibitorNNRTI Non-nucleoside RT inhibitor

PI Protease inhibitor

VL Viral loadIMSS Instituto Mexicano del Seguro Social

This work was supported by Consejo Nacional de Ciencia yTecnologıa SALUD/CONACYT grant 2004-01-008.

R. Lira ! H. Valdez-Salazar ! G. Vazquez-Rosales !O. Rojas-Montes ! M. Ruiz-Tachiquin ! A. Maldonado-Rodrıguez ! A. Gomez ! O. Munoz ! M.-T. Alvarez-Munoz (&)Unidad de Investigacion Medica en Enfermedades Infecciosas yParasitarias, Hospital de Pediatria, Centro Medico NacionalSiglo XXI, Instituto Mexicano del Seguro Social, Cuauhtemoc330 Col. Doctores, Delegacion Cuauhtemoc, Mexico City06720, Mexicoe-mail: mtalvarezm@yahoo.com.mx

R. Lirae-mail: rolica36@yahoo.com

H. Valdez-Salazare-mail: hildaavs@hotmail.com

G. Vazquez-Rosalese-mail: vazguill@aol.com

O. Rojas-Montese-mail: othonrojas@yahoo.com.mx

M. Ruiz-Tachiquine-mail: mertachiquin@yahoo.com.mx

A. Maldonado-Rodrıgueze-mail: mangimr@yahoo.com.mx

A. Gomeze-mail: agomez1992@aol.com

O. Munoze-mail: munoz@himfg.edu.mx

R. Torres-Ibarra ! C. Cano-DominguezCentro Medico Nacional La Raza, IMSS, Clinica de Hepatitis,Hospital de Infectologia, Mexico City, Mexicoe-mail: drarocio@prodigy.net.mx

C. Cano-Domingueze-mail: drcanod@hotmail.com

123

Arch Virol

DOI 10.1007/s00705-010-0696-y

Introduction

Highly active antiretroviral therapy (HAART) has sig-

nificantly reduced the rate of HIV-1/AIDS-related

mortality, but its clinical benefit is compromised andlimited because the exposure to antiretroviral drugs

has resulted in the emergence of drug-resistant viral

populations. The drugs most commonly used for treat-ment are protease inhibitors (PIs), nucleoside/nucleotide

reverse transcriptase inhibitors (NRTIs), and non-

nucleoside inhibitors (NNRTIs). In this context, it isimportant analyze the protease (PR) and reverse trans-

criptase (RT) genes in the patient undergoing treatment

in order to identify mutations associated with resistance.It has been demonstrated that resistance genotyping

results in better treatment guidelines for the patient,

improving treatment and reducing costs [1]. TheViroSeq system (Celera Diagnostics), a commercial kit

available for HIV-1 subtype B drug resistance geno-

typing, is used for the amplification, sequencing andanalysis of a 1.8-kb fragment from the HIV-pol region(containing the entire PR and the region encoding for

the first 335 amino acids of the RT open readingframe). Although plasma is considered the ideal speci-

men for viral load (VL) and genotypic drug resistance

testing, collection and shipment of plasma are notalways possible in settings where resources are limited.

It has been suggested that dried blood samples (DBS)carry less biohazard risk than liquid samples, facilitate

shipment to a centralized laboratory for testing, and

require only minimal storage facilities, since the nucleicacids are stable at room temperature [2–4]. Efforts have

been made to validate the use of DBS samples to

minimize cost in VL and drug resistance testing forHIV patients. Masciotra and colleagues [5] have

reported a high concordance between resistance geno-

types from plasma and DBS, using the ViroSeq geno-typing system and one in-house method for drug

resistance testing. Several other in house methods have

also been assayed for drug resistance testing with DBSsamples [6–11]. In third-world countries, the DBS

sampling method is increasingly used for diagnosis and

monitoring of HIV-1 infection [3, 7, 12]. In Mexico,where HIV-1 subtype B is predominant [13, 14], vali-

dation of DBS as a method of specimen collection for

drug resistance testing is urgently needed. In our labo-ratory, which is a reference center for monitoring HIV-1

patients, we have already validated the utility of DBS

samples for VL testing [2]. The aim of this study was toevaluate the sensitivity and reproducibility of DBS

sampling in the characterization of HIV-1 drug resis-

tance mutations in Mexican HIV-1 patients, using theViroSeq genotyping system.

Materials and methods

Patients

For this study, samples of blood and plasma from HIV-1patients at the Instituto Mexicano del Seguro Social

(IMSS) were collected over a period of 6 months. Because

the sensitivity of the genotyping assay has a limit of2,000 copies/mL, only 57 HIV-1 patients with viral loads

above this limit were included in this study. Matched

samples from patients of five States of Mexico, DistritoFederal (25), Veracruz (10), Jalisco (9), Yucatan (9), and

Nuevo Leon (4), were obtained. Informed consent was

obtained from all study participants. The study wasapproved by the local Committee of Research and Ethics

from the IMSS. Clinical and demographic data were

obtained for 51 patients, but no data were obtained for 6 ofthem. The age range was 24–66 years, 78.9% were men,

and 84.5% were undergoing HAART treatment (Table 1).

Samples

Two tubes of EDTA-containing whole blood specimens(5 mL/tube) were taken from each HIV-infected patient.

One tube was used to separate plasma by centrifugation at

4,000 rpm for 20 min. The plasma supernatant was ali-quoted and stored at -70"C until use. The second tube was

used to prepare DBS samples as follows: four cards werespotted with 6 9 50 lL of whole blood applied to filter

paper (Schleicher & Schuell 903, Keene, NH, USA).

Samples from each patient were collected on the same dayand allowed to dry at room temperature for 4 h, placed in a

zip-locked bag with a non-indicating silica gel desiccant

sachet (WA Hammond Drierite Co., Ltd, USA), andshipped from the different regions of the Mexican Republic

at room temperature to Hospital de Pediatria, CMN Siglo

XXI, IMSS in Mexico City and stored at -70"C until use.The range of storage time for DBS samples at room tem-

perature at collection sites before shipment was 5–7 days,

and the average time at -70"C was 12 months.

Table 1 Demographic and virological patient data

Number of patients n = 57

Ratio M:F 45:12

Age range (years) 24–66

VL range in plasma (RNA copies/mL) 2,290–958,000

log10 3.36–5.98

VL range in DBS (RNA copies/mL) 470–950,000

log10 2.67–5.98

CD4? range (cells/mm3) 4–878

R. Lira et al.

123

Plasma and DBS VL determinations

Two different methods were used to measure the HIV-1RNA VL of paired samples. Plasma samples were pro-

cessed by Cobas Amplicor HIV-1 MonitorTM test, version

1.5 (Roche Diagnostics, IN, USA), and DBS by Nucli-Sens# HIV-1 QT assay, NASBA Diagnostics (Organon

Teknika, Bostel, the Netherlands), according to manufac-

turer’s instructions. The Amplicor Monitor assay has asensitivity of 50 copies/mL and uses 500 lL of plasma.

Once the VL from plasma was determined, the DBS

samples were used for VL comparison. For DBS HIV-RNAVL determination, two circles of dried blood spots (100 lLof whole blood) were transferred to a tube containing 9 mL

of lysis buffer NucliSens# (Organon Teknika, Durham,NC, USA) and incubated for 2 h at room temperature with

rotation. Nucleic acids were then extracted, resuspended in

50 lL elution buffer and stored at -70"C until use. Theextracted nucleic acids were used for VL determination,

following the manufacturer’s instructions (NASBA).

HIV-1 genotyping from paired DBS and plasma

samples

HIV-1 genotyping from 57 matched plasma and DBS

samples was performed using the Celera Diagnostics

ViroSeq HIV-1 Genotyping System (version 2.0, AbbottMolecular, Des Plains, Illinois, USA). The ViroSeq assay

amplifies a 1.8-kb fragment of the pol gene (containing the

entire protease region and the region encoding the first335 amino acids of the RT open reading frame) and has

a sensitivity of detection of 2,000 RNA copies/mL of

plasma.

Plasma

For HIV RNA extraction, 500 lL of sample was used, and

RNA was eluted in a final volume of 50 lL dH2O, fol-

lowing the manufacturer’s instructions. Ten microliter ofHIV RNA (equivalent to 100 lL of starting material) was

used to perform retrotranscription with Moloney murine

leukemia virus RT, followed by amplification of the 1.8-kbpol fragment using two different sequence-specific primers,

following the manufacturer’s instructions. The amplified

product was purified using spin columns and analyzed by1% agarose gel electrophoresis. The whole PCR product

was sequenced in both the sense and antisense orientations

using the seven different overlapping sequence-specificprimers (A–H) included in the kit, with AmpliTaq Gold

DNA polymerase. All amplifications were performed in aBiometra T3 thermocycler (Biometry GmbH, Goettingen,

Germany).

DBS samples

We decided to extract the RNA from DBS samples withNucliSens# lysis buffer instead of ViroSeq lysis buffer

because RNA extraction with ViroSeq had not been vali-

dated for dried blood specimens. Four DBS circles,equivalent to 200 lL blood (50 lL/circle), were used for

HIV-1 RNA extraction as described above for VL deter-

minations and eluted in 50 ll of water. Twenty to 50 lL ofthe eluted HIV RNA (equivalent to 80–200 lL of the ini-

tial whole blood sample) was used for the amplification of

the 1.8-kb DNA product. Because the kit’s amplificationreaction allows a maximum volume of 10 lL of RNA per

reaction, it was necessary to concentrate the sample to the

required volume, using an Eppendorf Model 5301Vacu-Fuge Concentrator.

Analysis of sequences

Protease and RT genotypes were analyzed and interpreted

using the ViroSeq Celera Diagnostic ViroSeq HIV-1 Geno-typing System version 2.8 software. The reports of drug

resistance were sent to the patients and their physicians.

Sequences from matched samples (22) were also editedmanually with CLC bio 5.02-version software (CLC bio).

The genotypes were interpreted from the sequences using

the Stanford Genotypic Resistance Interpretation Algo-rithm (Version 5.0 beta), available at http://surveillance.

stanford.edu/cpr/servlet/CPR. This bioinformatic software

was also used to detect mixtures of nucleotide (nt)sequences in the consensus HIV pol sequences from mat-

ched sequences.

The nt sequences of the PR and RT regions (1,302 bp)were aligned using the Mega 4.0 program [15] to calculate

the percentages of amino acid and nt similarity. A phylo-

genetic tree was constructed by the neighbor-joiningmethod, using the sequences from paired plasma and DBS

samples, and the reliability of branching orders was

assessed by using the bootstrap approach with CLC bio5.02 version software (CLC bio). All sequences were

analyzedwith theGenotyping program from theNCBI (http://

www.ncbi.nlm.nih.gov/projects/genotyping/formpage.cgi) todetermine the HIV-1 subtype.

Statistical analysis

Statistical analysis was performed using SPSS software,

version 10.0 (SPSS Chicago IL). Pearson correlation wasused to determine the association between VL from DBS

and plasma, and a paired-samples t test (two-tailed) wasused to compare VL between groups. A p value of\0.05

was considered statistically significant.

Testing for HIV-1 drug resistance using dried blood samples

123

Results

Correlation of HIV-1 RNA VL in plasma and DBS

samples

To further support our previous findings, which suggested a

good correlation of VL between plasma and DBS samples

[2], we determined the VL from the plasma and DBSsamples. The correlation data from the 57 paired specimens

tested is shown in Fig. 1. In plasma samples, a range of VL

from 3.36 to 5.98 log10 was obtained, and for DBS, therange was 2.67–5.98. The mean value ± SD of HIV-1

RNA levels in plasma and DBS were 4.64 ± 0.69 and

4.58 ± 0.73 log10 copies/mL, respectively. Linear regres-sion analysis of log-transformed RNA measurements

(copies/mL) for paired specimens showed a very good

correlation between HIV-1 RNA VL in liquid plasma andDBS. The Pearson correlation between paired samples was

0.9; this correlation is significant at the 0.01 level (2-tailed

analysis).

Efficiency of amplification of the 1.8-kb pol fragment

in plasma and DBS samples

All plasma samples were successfully genotyped using the

ViroSeq assay. Of the 57 DBS specimens, only 24 (42.1%)could be amplified. However, there were two samples that

could not be evaluated with the software, and for thisreason, only 22 paired samples (38.6%) were successfully

edited. To determine if VL affected the probability of a

successful amplification, we classified the samples intothree groups according to the estimated plasma VL. For

group I (VL range 3.2–4.1 log10), none of the samples were

amplified (0/17), for group II (4.2–5.0 log10) a 42.8%

(9/21) amplification rate was obtained, and for group III(5.1–6.0 log10) 79% (15/19) of the samples were sequenced

and successfully genotyped.

Comparison of drug resistance mutations in paired

samples

The mutations associated with drug resistance and geno-

type interpretation generated from 22 paired plasma- andDBS-derived PR and RT gene sequences are shown in

Table 2. Of the paired samples, 77% (17/22) contained

mutations associated with drug resistance in the PR and RTgenes. In the first analysis, we realized that 7 out of 22

sequences showed some degree of discordance in drug

resistance mutations between plasma and DBS.Therefore, to review the nature of these discordances,

we decided to reanalyze the sequence data using a different

edition of the software. The CLC bio software allowsmanual alignment and editing of the sequences. Once we

obtained the pol consensus sequences, they were analyzed

with the Stanford University genotyping tool for theidentification of drug resistance mutations. Using this

software, six of the seven discordances were resolved.

Therefore, the concordance in resistance-associated muta-tions as well as interpretation of drug resistance genotypes

was 95.4% (21/22) between plasma and DBS. The

remaining sample (7V) showed two mutations in theplasma sample that were not identified in its DBS coun-

terpart. The plasma PR resistance mutation (G73S) resulted

in resistance to ATV/r, and the RT mutation (M184V)resulted in 3TC and FTC resistance. It is interesting to note

that this patient had been treated with RTV, SQV and 3TC.

Therefore, the presence of a mixed infection could not beruled out.

Detection of genotypic mixtures

Samples containing mixtures of wild type and drug-resis-

tance mutations (Table 2, underlined mutations) were alsodetected using the Stanford program. In four cases, wild

type and mutation mixtures were detected in both samples,

and in another four cases, a mixture was detected in onlyone of the samples.

Analysis of consensus pol sequences

The pol consensus sequences corresponding to 1,302 bp

of the 1.8-kb amplified fragment were analyzed to determinethe similarity of the nucleotide sequences between plasma

and DBS. The mean (±SD) similarity between nucleotide

pol sequences from plasma and DBS was 98.3 ± 0.94%and ranged between 95 and 99%. A phylogenetic tree

Fig. 1 Log10–log10 graph showing a linear regression comparing VLdata obtained from 57 matched samples of DBS versus plasma. Log10measurements of the HIV-1 RNA in DBS were made using NASBANucliSens, and in plasma samples, using COBAS AMPLICOR HIV-1Monitor. The values for the DBSs are plotted against the values forthe matched plasma samples. A linear correlation between the twosamples is shown (R2 = 0.79). The Pearson correlation coefficientwas 0.9

R. Lira et al.

123

Tab

le2

Drugresistance

mutationsandgenotypeinterpretationfrom

pairedsamplesofplasm

aandDBS

Sam

ple

IDPlasm

aVL

(copies/mL)

PR

mutations

RTmutations

Genotyperesistance

interpretation

Plasm

aDBS

Plasm

aDBS

Plasm

aDBS

11DF

168,000

None

None

None

None

None

None

19DF

249,000

I54V,G73S,

V82F,I84V,

L90M

I54V,G73S,

V82F,I84V,

L90M

M41L,D67N,T69D,

Y181C,G190A,L210W,

T215D

M41L,D67N,T69D,

Y181C,G190A,L210W,

T215D

ATV/r,DRV/r,FVP/r,ID

V/r,

LPV/r,NFV,SQV/r,TPV/r.

ABC,AZT,D4T,DDI,TDF,

DLV,EFV,ETR,NVP

ATV/r,DRV/r,FVP/r,ID

V/r,

LPV/r,NFV,SQV/r,TPV/r.

ABC,AZT,D4T,DDI,TDF,

DLV,EFV,ETR,NVP

28DFa

500,000

D30N,M46IM

,F53L,I54VL,

N88D,L90M

D30N,M46I,

F53L,I54VL,

N88D,L90M

M41L,D67N,K70R,

K103N,F116Y,Q151M,

G190A,T215V,K219E

M41L,D67N,K70R,

K103N,F116Y,Q151M,

G190A,T215V,K

219E

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r.ABC,AZT,

D4T,DDI,TDF,DLV,EFV,

ETR,NVP.

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r.ABC,AZT,

D4T,DDI,TDF,DLV,EFV,

ETR,NVP.

29DF

363,000

M46I,F53L,

I54V,V82A,

I84V,L90M

M46I,F53L,

I54V,V82A,

I84V,L90M

M41L,D67N,V75M,

K103M,M184V,L210W,

T215Y

M41L,D67N,V75M,

K103M,M184V,L210W,

T215Y

ATV/r,DRV/r,FVP/r,ID

V/r,

LPV/r,NFV,SQV/r,TPV/r.

3TC,ABC,AZT,D4T,DDI,

FTC,TDF,DLV,EFV,NVP

ATV/r,DRV/r,FVP/r,ID

V/r,

LPV/r,NFV,SQV/r,TPV/r.

3TC,ABC,AZT,D4T,DDI,

FTC,TDF,DLV,EFV,NVP

30DFa

93,000

L24I,G48M,

F53L,I54V,

G73CG,V82S

L24I,G48M,

F53L,I54V,

G73C,V82S

M184V,T215Y

M184V,T215Y

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r,TPV/r.3TC,

ABC,FTC

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r,TPV/r.3TC,

ABC,FTC

32DF

129,000

L90M

L90M

M4IL,M104V,L210W,

T215D

M4IL,M104V,L210W,

T215D

IDV/r,NFV,SQV/r.3TC,

ABC,AZT,D4T,DDI,FTC,

TDF.

IDV/r,NFV,SQV/r.3TC,

ABC,AZT,D4T,DDI,FTC,

TDF.

34DFa

155,000

L24I,M46I,

I54L,G73C,

V82A

L24I,M46I,

I54L,G73C,

V82A

M41L,D67N,V75M,

M184MV,L210W,

T215Y,G190S

M41L,D67N,V75M,

M184MV,L210W,

T215Y,G190S

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r.3TC,ABC,

AZT,D4T,DDI,FTC,TDF,

EFV,NVP

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r.3TC,ABC,

AZT,D4T,DDI,FTC,TDF,

EFV,NVP

38DFb

207,000

M46IM

,I54M,

I84V,L90M

M46IM

,I54M,

I84V,L90M

D67G,K70R,V75T,

T215F,K219E

D67G,K70R,V75T,

T215F,K219E

ATV/r,DRV/r,FVP/r,ID

V/r,

LPV/r,NFV,SQV/r,TPV/r.

ABC,AZT,D4T,DDI,TDF.

ATV/r,DRV/r,FVP/r,ID

V/r,

LPV/r,NFV,SQV/r,TPV/r.

ABC,AZT,D4T,DDI,TDF.

40DF

229,000

None

None

None

None

None

None

1J

435,000

None

None

Y188L,V118I

Y188L,V118I

DLV,EFV,NVP

DLV,EFV,NVP

10J

958,000

M46I

M46I

None

None

NFV

NFV

16Jb

83,200

None

None

K103N,Y181C,T215Y

K103N,Y181C,T215Y

AZT,D4T,DLV,EFV,ETR,

NVP.

AZT,D4T,DLV,EFV,ETR,

NVP.

19J

310,000

None

None

None

None

None

None

21J

70,600

None

None

None

None

None

None

22Ja

24,000

M46I,F53L,

V82T,I84V,

L90M

M46I,F53L,

V82T,I84V,

L90M

M41L,D67N,T69D,L74I,

M184V,L210W,T215D,

G190GRS,K103N

M41L,D67N,T69D,L74I,

M184V,L210W,T215D,

G190S,K103N

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r,TPV/r.3TC,

ABC,AZT,D4T,DDI,FTC,

TDF,DLV,EFV,NVP.

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r,TPV/r.3TC,

ABC,AZT,D4T,DDI,FTC,

TDF,DLV,EFV,NVP.

23J

84,000

None

None

M184V

M184V

3TC,FTC.

3TC,FTC.

Testing for HIV-1 drug resistance using dried blood samples

123

Tab

le2

continued

Sam

ple

IDPlasm

aVL

(copies/mL)

PR

mutations

RTmutations

Genotyperesistance

interpretation

Plasm

aDBS

Plasm

aDBS

Plasm

aDBS

15Yb

432,000

D30N,V82A,

L90M,N88D

D30N,V82A,

L90M,N88D

None

None

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r.

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV,SQV/r.

20Y

50,000

None

None

None

None

None

None

?7Va,b

28,000

G73S,L90M

L90M

M41L,M184V,K219R

M41L,K219KR

ATV/r,ID

V/r,NFV,SQV/r.

3TC,ABC,FTC.

IDV/r,N

FV,S

QV/r.A

ZT,D

4T

10V

275,000

I84V,N88D

I84V,N88D

None

None

ATV/r,FVP/r,ID

V/r,NFV,

SQV/r.

ATV/r,FVP/r,ID

V/r,NFV,

SQV/r.

12Vb

635,000

V32I,M46I,

V82A

V32I,M46I,

V82A

M184V

M184V

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV.3TC,FTC

ATV/r,FVP/r,ID

V/r,LPV/r,

NFV.3TC,FTC

1NLa

499,000

M46I,I47V,

I54AS,V82F,

L90M

M46I,I47V,

I54AS,V82F,

L90M

M41L,D67N,L74I,

V106A,G190A,T215D

M41L,D67N,L74I,

V106A,G190A,T215D

ATV/r,DRV/r,FVP/r,ID

V/r,

LPV/r,NFV,SQV/r,TPV/r.

ABC,AZT,D4T,DDI,FTC,

TDF,DLV,EFV,ETR,NVP.

ATV/r,DRV/r,FVP/r,ID

V/r,

LPV/r,NFV,SQV/r,TPV/r.

ABC,AZT,D4T,DDI,FTC,

TDF,DLV,EFV,ETR,NVP.

Sam

plesweream

plified

andsequencedusingtheCeleraDiagnosticsViroSeq

HIV

-1GenotypingSystem

(version2.0).Sequence

analysiswas

perform

edwithCLCbio

5.1software.Residues

inplain

textindicatecodonsin

thePRandtheRTgenes.Residues

inbold

arecodonsthat

differin

thepairedplasm

aandDBSsamples.ARV

drugsin

bold

arethose

that

differedin

thepaired

plasm

aandDBSsamples

DFDistritoFederal,JJalisco,YYucatan,VVeracruz,

NLNuevoLeon

aMixture,wildtypeandmutation

bForthesesamples,discordanceswereobserved

withViroSeq

software.

?A

proteaseresistance

mutationwas

detectedin

aDBSspecim

enbutnotin

itsplasm

acounterpart,oraprotease

resistance

mutationwas

detectedin

theplasm

aspecim

en,butnotin

itsDBScounterpart,partially

changingthegenotypeinterpretationin

theRTI

R. Lira et al.

123

derived with the 22 paired pol sequences showed a closeclustering of sequences from the same paired samples in all

cases (Fig. 2). For comparison, five subtype B reference

sequences were included. Bootstrap values obtained forplasma/DBS sequence pairs were higher than 99% for

almost all of the specimens (data not shown). All of the

sequences analyzed with the NCBI Genotyping tool, weresubtyped as B (data not shown).

Discussion

In this work, we were interested in validating the use ofDBS for drug resistance testing in HIV-1 patients. We have

been working in our laboratory, which is a reference center

in Mexico City, on validation of DBS samples, with thepurpose of using them for molecular diagnosis and

monitoring of HIV-1 patients. We first evaluated the VLfrom paired samples of DBS and plasma with two different

commercial kits, Cobas Amplicor HIV-1 Monitor v.1.5,

Roche for plasma and NucliSens HIV-1 for DBS, becausethe last of these had been recommended for DBS. The

comparison confirmed previous evidence of high correla-

tion between the VL from DBS versus plasma samples [2].The data allow us to strongly recommend the use of DBS

as a simpler sampling method for HIV-1 VL determination

in our country.Genotyping from DBS and dried serum spots using

laboratory-developed assays has been reported [6, 9–11,

16, 17], suggesting that DBSs are a feasible alternativespecimen type for genotypic drug-resistance testing. We

were therefore interested in the DBS sampling method as

an alternative not only for VL determination but also fordrug resistance genotyping. The ViroSeq kit is a

Fig. 2 Phylogenetic treeconstructed using sequence datagenerated from the pol region of22 paired plasma (P) and DBS(D) samples. The tree wasconstructed using sequence datafrom paired plasma and DBSsamples with 1,000 non-parametric bootstraps. Thephylogenetic tree was producedby the neighbor-joining method,and the reliability of thebranching orders was assessedby using the bootstrap approachwith CLC bio (CLC bio 5.02software). Five referencesequences (circled) representingHIV subtype B were included(K03455.1, M1745.1,DQ853463.1, U21135.1,U63632.1). Patient sampleswere from V (Veracruz), DF(Distrito Federal), J (Jalisco),Y (Yucatan), and NL (NuevoLeon)

Testing for HIV-1 drug resistance using dried blood samples

123

commercial assay that amplifies a 1.8-kb pol region frag-

ment and has a sensitivity of detection of 2,000 HIV-1RNA copies/mL of plasma [9]. From the 57 paired samples

tested with the ViroSeq assay, the amplification rate of the

1.8-kb pol-fragment falls within the range of 4.2–5.0 log10and increases for viral loads greater than 5.1 log10 copies/

mL, resulting in amplification of 79% (15/19) of the

samples. We therefore conclude from this study that sam-ples with VL above 5.1 log10 were successfully amplified

with the ViroSeq kit. The rates of successful amplificationusing laboratory-developed genotyping assays with com-

mercial and in-house methods have a reported range of

7.82–93.1%, indicating that DBSs are a feasible alternativespecimen type for genotypic drug-resistance testing [6, 11,

18]. In our study, 38.6% of the DBS samples were

amplified, suggesting that improved assay sensitivity is stillneeded for the routine use of DBS for monitoring of ART-

treated individuals for drug resistance using the ViroSeq

system. Technical solutions offering better elution andextraction efficiencies with DBS samples should improve

the sensitivity of the method. Elution from the filter paper

is likely to be one of the most sensitive steps, but this doesnot rule out some other problems, such as correct storage of

the samples, errors in sequencing, and reagent or kit

variability.The use of equivalent volumes of sample could be an

important factor involved in amplification failure. In this

study, the volume of plasma present in four DBS spots(50 lL/spot), considering an initial volume of 200 lL of

whole blood with a hematocrit value of 40% was approxi-

mately equivalent to 100 lL of plasma required for oneViroSeq assay. In the case where all of the volume eluted

from four spots was needed, equivalent volumes of both

samples were used. However, several unsuccessful attemptswere made to assay samples that were not amplified,

including increasing the number of spots to eight. We

therefore suggest that for those sample, the storage condi-tions, mainly the humidity, could be a determining factor.

With regard to DBS storage conditions, some authors

have reported that HIV-1 RNA from DBS samples wasstable at room temperature and -70"C for at least 1 year,

and possibly longer [18–20]. It has also been reported the

HIV-1 RNA degradation may occur under suboptimalstorage conditions [19, 20]. We were interested in testing if

the failure to amplify some samples could be due to

inappropriate storage conditions. For this purpose, weamplified a smaller fragment, a 700-bp fragment from the

PR region, using an in-house PCR method. Two different

DBS samples that had not been amplified with ViroSeqwere used, and amplification of the 700-bp fragment was

observed with both samples (data not shown), suggesting

that maybe some partial degradation of HIV RNA couldhave occurred, resulting in the inability of the ViroSeq

system to amplify a longer fragment (1.8 kb). Unfortu-

nately, it was not possible to test this with the rest of thesamples because they had been used up during multiple

attempts at amplification with the ViroSeq kit. Youngpairoj

et al. [20] evaluated humidity in DBS after 6 months ofstorage and found that half of the samples had evidence of

humidity. They recommend changing the desiccant bag,

during long-term storage. It is also important to mentionthat our samples were stored in a non-frost freezer, and the

desiccant sachet did not contain a humidity indicator, norwas it changed during the time of storage (more than

6 months). Therefore, we suggest that partial degradation

of the RNA may have occurred during long-term storage at-70"C, possibly due to suboptimal storage temperature,

humidity or both. Therefore, the presence of a mixed

infection cannot be ruled out.We identified discordances in some of the samples

analyzed using ViroSeq software (version 2.8) that could

be due to the low level of resolution of the eight sequences.We demonstrated that edition problems were resolved

when a consensus pol sequence was obtained using CLC

bio software and the Stanford Genotypic ResistanceInterpretation Algorithm. The majority of samples with

previous discordances in the genotype were resolved,

except for one of them. We suggest that mixed populationswere present in the same sample. We were able to identify

mixtures of wild-type and resistant viruses using the

Stanford University HIV drug resistance database. Analysiswith this bioinformatic tool gives a report that shows

mutations when a mixture is present in the amplified

fragment. Evidence of wild-type and resistant populationsbeing present in the same sample was also reported in this

study. Finally, the 22 pol nucleotide and deduced amino

acid sequences that were obtained showed a high degree ofsimilarity, indicating a high sequence concordance (95–

99%). A good concordance was also obtained in resistance-

associated mutations as well as in interpretation of drugresistance genotypes (94.5%). The validation of DBS as a

sampling method for drug resistance genotyping is still

very important, especially in highly immunosuppressedpatients and children from whom fluid blood specimens are

scarce. It has been reported that children have higher rates

of virological failure than adults [21] and a broad range ofviremia (VL 193 to 3 million HIV-1 RNA copies/mL).

Higher rates of drug resistance and failure have been

reported in resource-limited settings than in well-resourcedsettings because, often, only one or few treatment regimens

are available for prophylaxis and patient management. This

is a serious problem and a reason why the validation of lessexpensive sampling methods for drug resistance testing is

needed in our country.

In conclusion, our results demonstrate that DBS samplesare a potential alternative to plasma as a good specimen

R. Lira et al.

123

source for HIV-1 resistance genotyping. Factors were

identified that could potentially improve both the sensi-tivity and the quality of genotype data, such as storage of

sample and sequence analysis. Therefore, DBS sampling is

useful for determining viral load and drug resistancegenotypes in Mexican HIV-1 patients.

Acknowledgments This work was supported by Consejo Nacionalde Ciencia y Tecnologia (CONACYT-SALUD 2004-01-008). Wethank Dr. Ana Ma. Cevallos for valuable comments and reviewing themanuscript, and Dr. Margarita Dehesa for inclusion of patients.

Conflict of interest statement The authors declare that they haveno conflict of interest.

References

1. Zolopa AR (2006) Incorporating drug-resistance measurementsinto the clinical management of HIV-1 infection. J Infect Dis194(Suppl 1):S59–S64

2. Alvarez-Munoz MT, Zaragoza-Rodriguez S, Rojas-Montes O,Palacios-Saucedo G, Vazquez-Rosales G, Gomez-Delgado A,Torres J, Munoz O (2005) High correlation of human immuno-deficiency virus type-1 viral load measured in dried-blood spotsamples and in plasma under different storage conditions. ArchMed Res 36:382–386

3. Cassol SA, Read S, Weniger BG, Gomez P, Lapointe N, Ou CY,Babu PG (1996) Dried blood spots collected on filter paper: aninternational resource for the diagnosis and genetic character-ization of human immunodeficiency virus type-1. Mem InstOswaldo Cruz 91:351–358

4. Uttayamakul S, Likanonsakul S, Sunthornkachit R, KuntiranontK, Louisirirotchanakul S, Chaovavanich A, Thiamchai V,Tanprasertsuk S, Sutthent R (2005) Usage of dried blood spots formolecular diagnosis and monitoring HIV-1 infection. J VirolMethods 128:128–134

5. Masciotra S, Garrido C, Youngpairoj AS, McNulty A, ZahoneroN, Corral A, Heneine W, De MC, Garcia-Lerma JG (2007) Highconcordance between HIV-1 drug resistance genotypes generatedfrom plasma and dried blood spots in antiretroviral-experiencedpatients. AIDS 21:2503–2511

6. Bertagnolio S, Soto-Ramirez L, Pilon R, Rodriguez R, ViverosM, Fuentes L, Harrigan PR, Mo T, Sutherland D, Sandstrom P(2007) HIV-1 drug resistance surveillance using dried wholeblood spots. Antivir Ther 12:107–113

7. Buckton AJ (2008) New methods for the surveillance of HIVdrug resistance in the resource poor world. Curr Opin Infect Dis21:653–658

8. Buckton AJ, Bissett SL, Myers RE, Beddows S, Edwards S, CanePA, Pillay D (2008) Development and optimization of an inter-nally controlled dried blood spot assay for surveillance of humanimmunodeficiency virus type-1 drug resistance. J AntimicrobChemother 62:1191–1198

9. McNulty A, Jennings C, Bennett D, Fitzgibbon J, Bremer JW,Ussery M, Kalish ML, Heneine W, Garcia-Lerma JG (2007)Evaluation of dried blood spots for human immunodeficiencyvirus type 1 drug resistance testing. J Clin Microbiol 45:517–521

10. Plantier JC, Dachraoui R, Lemee V, Gueudin M, Borsa-Lebas F,Caron F, Simon F (2005) HIV-1 resistance genotyping on driedserum spots. AIDS 19:391–397

11. Ziemniak C, George-Agwu A, Moss WJ, Ray SC, Persaud D(2006) A sensitive genotyping assay for detection of drug resis-tance mutations in reverse transcriptase of HIV-1 subtypes B andC in samples stored as dried blood spots or frozen RNA extracts.J Virol Methods 136:238–247

12. Beck IA, Drennan KD, Melvin AJ, Mohan KM, Herz AM,Alarcon J, Piscoya J, Velazquez C, Frenkel LM (2001) Simple,sensitive, and specific detection of human immunodeficiencyvirus type 1 subtype B DNA in dried blood samples for diagnosisin infants in the field. J Clin Microbiol 39:29–33

13. Rivera-Morales LG, Novitsky VA, Trujillo JR, Lavalle-MontalvoC, Cano-Dominguez C, Ramos-Jimenez J, Jimenez-Rios E,Flores-Flores L, Lopez-Guillen P, Gilbert P, Vannberg F, Tamez-Guerra R, Rodriguez-Padilla C, Essex M (2001) The molecularepidemiology of HIV type 1 of men in Mexico. AIDS Res HumRetroviruses 17:87–92

14. Gomez-Carrillo M, Pampuro S, Duran A, Losso M, Harris DR,Read JS, Duarte G, De SR, Soto-Ramirez L, Salomon H (2006)Analysis of HIV type 1 diversity in pregnant women from fourLatin American and Caribbean countries. AIDS Res Hum Ret-roviruses 22:1186–1191

15. Tamura K, Dudley J, Nei M, Kumar S (2007) MEGA4: Molec-ular evolutionary genetics analysis (MEGA) software version 4.0.Mol Biol Evol 24:1596–1599

16. Nyambi PN, Fransen K, De BH, Chomba EN, Temmerman M,Ndinya-Achola JO, Piot P, Van der GG (1994) Detection ofhuman immunodeficiency virus type 1 (HIV-1) in heel prickblood on filter paper from children born to HIV-1-seropositivemothers. J Clin Microbiol 32:2858–2860

17. Steegen K, Luchters S, Demecheleer E, Dauwe K, Mandaliya K,Jaoko W, Plum J, Temmerman M, Verhofstede C (2007) Feasi-bility of detecting human immunodeficiency virus type 1 drugresistance in DNA extracted from whole blood or dried bloodspots. J Clin Microbiol 45:3342–3351

18. Hallack R, Doherty LE, Wethers JA, Parker MM (2008) Evalu-ation of dried blood spot specimens for HIV-1 drug-resistancetesting using the Trugene HIV-1 genotyping assay. J Clin Virol41:283–287

19. O’Shea S, Mullen J, Corbett K, Chrystie I, Newell ML, BanatvalaJE (1999) Use of dried whole blood spots for quantification ofHIV-1 RNA. AIDS 13:630–631

20. Youngpairoj AS, Masciotra S, Garrido C, Zahonero N, De MC,Garcia-Lerma JG (2008) HIV-1 drug resistance genotyping fromdried blood spots stored for 1 year at 4"C. J Antimicrob Che-mother 61:1217–1220

21. Shearer WT, Quinn TC, LaRussa P, Lew JF, Mofenson L, AlmyS, Rich K, Handelsman E, Diaz C, Pagano M, Smeriglio V,Kalish LA (1997) Viral load and disease progression in infantsinfected with human immunodeficiency virus type 1. Women andInfants Transmission Study Group. N Engl J Med 336:1337–1342

Testing for HIV-1 drug resistance using dried blood samples

123