BRIEF REPORT
and effective for prevention of HIV-1 mother-to-child trans-
Persistence of K103N-Containing
HIV-1 Variants after Single-Dose
Nevirapine for Prevention of HIV-1
Mother-to-Child Transmission
Tamara S. Flys,1 Deborah Donnell,2 Anthony Mwatha,2
Clemensia Nakabiito,3 Philippa Musoke,3 Francis Mmiro,4
J. Brooks Jackson,1 Laura A. Guay,1 and Susan H. Eshleman1
1
Department of Pathology, Johns Hopkins University School of Medicine,
Baltimore, Maryland; 2Statistical Center for HIV/AIDS Research
and Prevention, Fred Hutchinson Cancer Research Center, Seattle,
Washington; Departments of 3Paediatrics and 4Obstetrics and Gynaecology,
Makerere University, Kampala, Uganda
K103N-containing human immunodeficiency virus (HIV)–1
variants are selected in some women who receive single-dose
(SD) nevirapine (NVP) for prevention of HIV-1 mother-in-
fant transmission. We examined the persistence of K103N in
women who received SD NVP prophylaxis. K103N was de-
tected using the LigAmp assay (assay cutoff, 0.5% K103N).
K103N was detected at 6–8 weeks in 60 (41.7%) of 144
women. Fading (lack of detection) of K103N was documented
in 16 women by 2 years, 43 women by 3 years, and 55 women
by 4 and 5 years. Slower fading was independently associated
with HIV-1 subtype (D1A) and higher pre-NVP viral load.
The HIV Network for Prevention Trials (HIVNET) 012 regimen
of single-dose (SD) nevirapine (NVP) [1, 2] is safe, inexpensive,
Received 24 May 2006; accepted 19 October 2006; electronically published 18 January
2007.
Potential conflicts of interest: S.H.E. is a coinventor of the LigAmp assay, and Johns Hopkins
University has filed a patent application with the US Patent and Trademark Office. The inventors
may receive royalty payments if the patent is awarded and licensed. All other authors report
no potential conflicts.
Presented in part: XV International HIV Drug Resistance Workshop, Sitges, Spain, 13–17
June 2006 (abstract 45).
Financial support: HIV Network for Prevention Trials, National Institute of Allergy and
Infectious Diseases (NIAID), National Institutes of Health (NIH), Department of Health and
Human Services (DHHS) (contract N01-AI-35173 to Family Health International, contract N01-
AI-45200 to Fred Hutchinson Cancer Research Center, and subcontract NOI-AI-35173-417 to
Johns Hopkins University); HIV Prevention Trials Network, NIAID, National Institutes of Child
Health and Human Development (NICHD), National Institute on Drug Abuse, National Institute
of Mental Health, and Office of AIDS Research, NIH, DHHS (grants U01-AI-46745, U01-AI-
48054, U01-AI-068613); NIH, NICHD (grant R01-HD042965-01).
Reprints or correspondence: Dr. Susan Eshleman, Dept. of Pathology, The Johns Hopkins
Medical Institutions, Ross Bldg. 646, 720 Rutland Ave., Baltimore, MD 21205 (seshlem@
jhmi.edu).
The Journal of Infectious Diseases 2007; 195:711–5
 2007 by the Infectious Diseases Society of America. All rights reserved.
0022-1899/2007/19505-0017$15.00
DOI: 10.1086/511433
mission (pMTCT). However, exposure to SD NVP is associated
with selection of NVP-resistant HIV-1 variants. Using a pop-
ulation sequencing–based genotyping assay (ViroSeq), the por-
tion of women with NVP resistance detected 6–8 weeks after
SD NVP varied among HIV-1 subtypes (C1D1A) [3]. The most
common NVP resistance mutation detected in women with all
3 subtypes was K103N. ViroSeq reliably detects the K103N
mutation when it is present in ⭓20% of the viral population
[4]. Using a sensitive and quantitative point mutation assay
(LigAmp), the portion of women with detectable K103N
(10.5% K103N) 6–8 weeks after SD NVP was 41.7% for subtype
A, 55.3% for subtype D, and 69.8% for subtype C. The highest
mean levels of K103N were found in women with subtype C
(2.2% for subtype A, 5.5% for subtype D, and 11.7% for sub-
type C) [5]. Differences in detection of K103N in women with
these 3 subtypes did not appear to reflect differences in the
viral loads of the samples or differences in HIV-1 sequences
at the binding sites of oligonucleotides used in the LigAmp
assay [5].
K103N-containing variants can persist at low levels for years
in the absence of antiretroviral drug exposure in patients who
are infected with resistant strains [6] and in patients who dis-
continue treatment with a nonnucleoside reverse-transcriptase
inhibitor (NNRTI) [7]. Long-term persistence of K103N-con-
taining variants in women after SD NVP could potentially re-
duce the efficacy of SD NVP for pMTCT in subsequent preg-
nancies or the efficacy of NNRTI-based regimens for future
treatment of their HIV-1 infection. In the HIVNET 012 trial,
1–2-year follow-up samples were available for 9 women who
had NVP resistance mutations detected at 6–8 weeks (3 of
subtype A and 6 of subtype D). Using the LigAmp assay, we
detected low levels of K103N-containing variants in 3 of those
women 1–2 years after SD NVP, all with subtype D [8]. In this
report, we used the LigAmp assay to analyze the persistence of
K103N-containing variants in women with subtype A and D
HIV-1 who were enrolled in a 5-year follow-up study of the
HIVNET 012 trial.
Subjects, materials, and methods. Two-hundred thirty-
two (75.8%) of 306 antiretroviral drug–naive Ugandan women
who received SD NVP in HIVNET 012 were subsequently en-
rolled in a follow-up study with annual study visits 2–5 years
after their original SD NVP exposure. Eighty-eight of the 232
women were excluded from analysis for the following reasons:
(1) the receipt of SD NVP in a subsequent pregnancy during
the follow-up period (n p 39), (2) the presence of a subtype
BRIEF REPORT • JID 2007:195 (1 March) • 711
other than A or D (3 of subtype C, 23 recombinant, and 5 of
undetermined subtypes), (3) the receipt of treatment with com-
bination antiretroviral therapy during the follow-up period
(n p 2), (4) the lack of a 6–8-week sample available for analysis
(n p 7), (5) the failure of the 6–8-week sample to amplify
(n p 4), (6) the occurrence of the 6–8 week-study visit 1100
days after receiving SD NVP (n p 4 ), and (7) the presence of
an unusual K103Q polymorphism in HIV-1 reverse transcrip-
tase (n p 1). In the woman with the K103Q polymorphism,
3%–6% K103N was detected in samples both before and up
to 5 years after SD NVP exposure. Analysis of HIV-1 clones
from the pre-NVP sample confirmed that detection of K103N
represented a false-positive test result. In contrast, analysis of
other samples with 0.5%–1% K103N with an independent assay
confirmed the presence of low levels of K103N-containing var-
iants [8].
The LigAmp assay involves a mutation-specific oligonucle-
otide ligation step followed by a universal real-time polymerase
chain reaction (PCR) detection step [4]. The LigAmp assay was
performed using PCR products remaining after HIV-1 geno-
typing with ViroSeq (assay cutoff, 0.5% K103N). Samples were
analyzed in duplicate, and the results for the percentage of
K103N were averaged.
HIV-1 subtyping was performed previously by phylogenetic
analysis of HIV-1 pol region sequences [3]. Baseline HIV-1 load
and CD4 cell count were measured before NVP administration
in the HIVNET 012 trial, as described elsewhere [1, 2]. In-
formed consent was obtained from all subjects for participation
in HIVNET 012 and the follow-up study. HIVNET 012 and
the follow-up study were approved by the local institutional
review boards in Uganda and at Johns Hopkins University
School of Medicine. Guidelines of the US Department of Health
and Human Services and the authors’ institutions were followed
in the conduct of this research.
Results. Among 144 women in this study, 60 (41.7%) had
K103N detected at the 6–8-week visit (⭓0.5% K103N). We an-
alyzed persistence of K103N among those 60 women by use of
samples collected 2, 3, 4, and 5 years after SD NVP adminis-
tration. For each woman, after fading of K103N was docu-
mented (!0.5% K103N detected in a sample), samples from
subsequent study visits were not tested. Many of the 60 women
did not have 2-year samples because approval of the amend-

Figure 1. Analysis of K103N-containing variants in women after single-dose nevirapine administration. The present study analyzed women enrolled
in the long-term follow-up study of the HIV Network for Prevention Trials 012 cohort (see Subjects, materials, and methods). Sixty of those women
had K103N detected at the 6–8-week visit by use of the LigAmp assay (assay cutoff, 0.5% K103N) and 84 did not. Follow-up samples were analyzed
with the LigAmp assay. For each subsequent study visit, samples were tested from women who either had no sample available from the prior visit
or who had K103N detected at the prior visit. For each woman, after fading of K103N was documented (!0.5% K103N), samples from subsequent
study visits were not analyzed. Data are the no. of women tested at each study visit who had positive results (K103N was detected at ⭓0.5% with
the LigAmp assay), negative results (K103N was not detected with the LigAmp assay), or no sample available for testing.

712 • JID 2007:195 (1 March) • BRIEF REPORT

 Table 1. Turnbull estimates of the percentage of women with !0.5% K103N in plasma
samples among all women (n p 144) and among the women with K103N detected 6–8
weeks after single-dose (SD) nevirapine (NVP; n p 60).

Group, study visit Subtype A Subtype D Total

All womena
6–8 weeks 65.9 (55.4–75.0) 46.4 (33.9–59.4) 58.3 (50.1–66.1)
2 years 96.4 (86.1–99.1) 70.2 (49.4–85.1) 87.6 (78.9–93.0)
3 years 98.7 (91.6–99.8) 81.1 (67.8–89.7) 91.9 (85.7–95.5)
4 years 100.0 (100.0–100.0) 97.6 (85.2–99.7) 99.1 (93.9–99.9)
5 years 100.0 (100.0–100.0) 97.6 (85.2–99.7) 99.1 (93.9–99.9)
Women with K103N detected
6–8 weeks after SD NVPb
2 years 89.4 (64.4–97.5) 30.1 (10.9–60.2) 63.0 (45.5–77.7)
3 years 96.3 (77.9–99.5) 65.2 (45.5–80.7) 80.7 (67.8–89.2)
4 years 100.0 (100.0–100.0) 95.6 (75.0–99.4) 97.8 (86.3–99.7)
5 years 100.0 (100.0–100.0) 95.6 (75.0–99.4) 97.8 (86.3–99.7)

NOTE. Data are estimated cumulative percentages (95% confidence intervals) of women in whom
fading was documented. Fading was defined as failure to detect K103N using the LigAmp assay, with an
assay cutoff of 0.5%. A survival analysis approach was used to estimate fading of K103N-containing variants.
Estimates of time to fading were obtained using a modified Kaplan-Meier estimator (Turnbull estimator)
with discrete time steps (visits 1–5 corresponding to the 6–8-week, 2-year, 3-year, 4-year, and 5-year visits).
Women were considered right-censored at their last sample collection visit if they were still resistant at
this last visit (5 women). Women were considered interval-censored if fading was observed between visits
(55 women). Finally, women were considered left-censored if no K103N was detected at the 6–8-week
visit (84 women). In those women, the event (fading) was considered to have occurred before the obser-
vation period began. Both univariate and multivariate analysis of baseline factors associated with fading
used a Weibull survival model for time to fading of resistance. To compute P values for the difference
between point estimates at a given visit, a 2-sided Z test was used, with pointwise SEs obtained from the
Turnbull estimator. Analyses were conducted using SAS (version 9.1; SAS Institute).
a
Subtype A, n p 88; subtype D, n p 56; total, n p 144.
b
Subtype A, n p 30; subtype D, n p 30; total, n p 60.

ment for the follow-up study and their consent for the follow-
up study were obtained 12 years after they received SD NVP
in HIVNET 012. The level of K103N detected at 6–8 weeks
was similar among women with 2-year samples (median, 2.2%)
to that of those without 2-year samples (mean, 2.3%; P p
.44); however, there is some imbalance in subtype. Women with
subtype A compose a higher proportion of those with samples
at 2 years (14/20 of those who were tested at 2 years were
subtype A vs. 16/40 of those who were not tested; P p .055).
Among all 144 women, we documented undetectable K103N
in a total of 100 women by 2 years, 127 by 3 years, and 139
by 4 and 5 years (figure 1). The remaining women either had
K103N detected or had no sample available for testing. Figure
1 shows the number of women tested at each study visit. Among
the 60 women who had K103N detected at 6–8 weeks, we
documented undetectable K103N in 16 women by 2 years, 43
by 3 years, and 55 by 4 and 5 years. Whenever K103N was
detected in samples after the 6–8-week visit, the level of K103N
in the viral population was low. The mean percentage of K103N
among women with detectable K103N was 1.2% at 2 years
(n p 4 women; range, 0.9%–1.8%) and 0.8% at 3 years (n p
9 women; range, 0.5%–1.2% K103N).
A statistical model was used to estimate the cumulative rate
of fading of K103N. In this study, in which many women did
not have 2-year results but did have 3-year results (interval
censored), standard Kaplan-Meier estimates could not be used
to estimate the cumulative probability of fading at each study
visit. The Turnbull estimate (a Kaplan-Meier–type estimate
modified for interval censored data) uses the intuitive as-
sumption that the distribution of missing test data at each study
visit can be estimated from available test data from the same
visit. This approach provides estimates and corresponding con-
fidence intervals (CIs) of the cumulative percentage of women
with undetectable K103N at each study visit (cumulative rate
of fading). The estimated cumulative rates of fading for all 144
women at 2, 3, 4, and 5 years by use of this model were 87.6%,
91.9%, 99.1%, and 99.1%, respectively (table 1). When the
analysis was limited to the 60 women with K103N detected at
6–8 weeks, the estimated cumulative rates of fading at 2, 3, 4,
and 5 years were 63.0%, 80.7%, 97.8%, and 97.8%, respectively
(table 1). The estimated rates of fading were consistently lower
in subtype D than A, both among all 144 women and among
the subset of 60 women who had K103N detected at 6–8 weeks
(30 women with subtype A and 30 women with subtype D)
(table 1). Among those 60 women, the mean number of years
to the first available follow-up sample was similar for women
with subtype A, compared with those with D (2.9 years for
both subtypes; P p .94, t test).

BRIEF REPORT • JID 2007:195 (1 March) • 713

 In univariate models examining predictors of fading among
all 144 women, subtype (D1A), higher baseline viral load, and
lower baseline CD4 cell count were significant predictors for a
longer time to fading of K103N. In a multivariate model, sub-
type (hazard ratio [HR] for D vs. A, 0.50 [95% CI, 0.33–0.77];
P p .0007) and baseline viral load (HR per log increase in
baseline viral load, 0.63 [95% CI, 0.48–0.83]; P p .0006) were
independently associated with time to fading, but baseline CD4
cell count was not (HR per decrease of 100 cells, 0.97 [95%
CI, 0.90–1.05]; P p .43). The time to fading was approximate-
ly twice as long for women with subtype D, compared with
women with subtype A, and a log increase in baseline viral
RNA increased time to fading by a factor of 1.57. Among the
subset of 60 women who had K103N-containing variants de-
tected at 6–8 weeks, time to fading was predicted by the level
of K103N at the 6–8-week visit in a univariate model but not
in a multivariate model. In a multivariate model, results for
this subset of 60 women were similar to those obtained for all
144 women; only subtype and viral load predicted time to
fading (HR for subtype, 0.34 [95% CI, 0.17–0.68] [P p .005];
HR for viral load, 0.45 [95% CI, 0.22–0.90] [P p .003]).
Discussion. This report demonstrates that HIV-1 variants
with K103N can persist at very low levels for years in some
women following SD NVP exposure and that HIV-1 subtype
(D1A) and higher baseline viral load are both associated with
slower fading of those variants. Since we first described selection
of K103N-containing variants after SD NVP [9], concerns have
been raised about whether persistence of NVP-resistant HIV-
1 would lower the efficacy of SD NVP for pMTCT in subse-
quent pregnancies. Two preliminary studies found no signifi-
cant difference in the efficacy of the HIVNET 012 regimen for
pMTCT with first versus second use [10, 11]. Early reports of
NVP resistance after SD NVP also raised concerns that prior
exposure to SD NVP would adversely affect the response of
women to future treatment with NNRTI-containing regimens.
Available studies suggest that NVP-resistant variants that emerge
after SD NVP exposure are unlikely to compromise subsequent
treatment of women with NNRTI-containing regimens, pro-
vided that there is sufficient time between the initial NVP ex-
posure and treatment initiation to allow fading of NVP-resis-
tant strains [12–15]. Data in this report suggest that the rate
of fading of NVP-resistant strains may vary from one geo-
graphic region to another, depending on which subtypes are
prevalent. Further studies are needed to evaluate whether sub-
type-based differences in the fading of NVP-resistant HIV-1
variants influence clinical or virologic outcome in NVP-ex-
posed women who are subsequently treated with NNRTIs.
Many countries now offer pregnant women potent antiret-
roviral regimens for pMTCT and for treatment of their own
HIV-1 infection. Women with higher HIV-1 loads are more
likely to transmit HIV-1 to their infants and are also more
714 • JID 2007:195 (1 March) • BRIEF REPORT
likely to have NVP-resistant strains detected after SD NVP
exposure. We now show that K103N-containing variants are
also more likely to persist in the years following delivery among
women with higher HIV-1 loads. These findings emphasize the
importance of evaluating HIV-1 disease status in pregnant HIV-
1–positive women and the importance of providing eligible
women with highly active antiretroviral therapy (HAART) dur-
ing pregnancy, if it is available.
Unfortunately, HAART is still not available to pregnant
women in many countries that shoulder the greatest burden
of the HIV-1 epidemic, because of cost, lack of needed infra-
structure, and other factors. Additionally, many women in re-
source-limited settings do not know their HIV-1 infection
status and do not present for medical care until close to or at
the time of delivery; in such situations, initiation of HAART
during pregnancy, even if available, would not be possible. Our
findings demonstrate fading of NVP resistance to very low levels
after SD NVP exposure, and recent clinical studies suggest that
the response to NNRTI-based HAART is not compromised by
prior SD NVP exposure if treatment initiation occurs at a time
distant from the exposure. These observations support contin-
ued use of SD NVP as an important option for pMTCT in
settings where more complex regimens are not available.
Acknowledgments
We thank the staff in Uganda and Johns Hopkins for assistance with
sample processing. We also thank Clinical Trials Specialist Melissa Allen
(Family Health International) for assistance and Lynne M. Mofenson (Na-
tional Institutes of Child Health and Human Development) for critical
review of the manuscript.
References
1. Guay LA, Musoke P, Fleming T, et al. Intrapartum and neonatal single-
dose nevirapine compared with zidovudine for prevention of mother-
to-infant transmission of HIV-1 in Kampala, Uganda. HIVNET-012
randomised trial. Lancet 1999; 354:795–802.
2. Jackson JB, Musoke P, Fleming T, et al. Intrapartum and neonatal
single-dose nevirapine compared with zidovudine for prevention of
mother-to-child transmission of HIV-1 in Kampala, Uganda: 18
months follow-up of the HIVNET 012 randomised trial. Lancet 2003;
362:859–68.
3. Eshleman SH, Hoover DR, Chen S, et al. Nevirapine (NVP) resistance
in women with subtype C, compared to subtypes A and D, after ad-
ministration of single dose NVP. J Infect Dis 2005; 192:30–6.
4. Church JD, Jones D, Flys T, et al. Sensitivity of the ViroSeq HIV-1
Genotyping System for detection of the K103N resistance mutation in
HIV-1 subtypes A, C, and D. J Mol Diagn 2006; 8:420–2.
5. Flys TS, Chen S, Jones DC, et al. Quantitative analysis of HIV-1 variants
with the K103N resistance mutation after single dose nevirapine in
women with HIV-1 subtypes A, C and D. J Acquir Immune Defic
Syndr 2006; 42:61–3.
6. Little SJ, Dawson K, Hellmann NS, Richman DD, Frost SDW. Persis-
tence of transmitted drug-resistant virus among subjects with primary
HIV infection deferring antiretroviral therapy. Antivir Ther (Suppl)
2003; 8:S129.
7. Palmer S, Boltz V, Maldarelli F, et al. Selection and persistence of non-
 nucleoside reverse transcriptase inhibitor-resistant HIV-1 in patients
starting and stopping non-nucleoside therapy. AIDS 2006; 20:701–10.
8. Flys T, Nissley DV, Claasen CW, et al. Sensitive drug resistance assays
reveal long-term persistence of HIV-1 variants with the K103N nevi-
rapine (NVP) resistance mutation in some women and infants after
single dose NVP: HIVNET 012. J Infect Dis 2005; 192:24–9.
9. Jackson JB, Becker-Pergola G, Guay L, et al. Identification of the K103N
resistance mutation in Ugandan women receiving nevirapine to prevent
HIV-1 vertical transmission. AIDS 2000; 14:F111–5.
10. Eure C, Bakaki P, McConnell M, et al. Effectiveness of repeat single-
dose nevirapine in subsequent pregnancies among Ugandan women
[abstract 125]. In: Program and abstracts of the 13th Conference on
Retroviruses and Opportunistic Infections (Denver). Alexandria, VA:
Foundation for Retrovirology and Human Health, 2006:93.
11. Martinson N, Ekouevi D, Gray G, et al. Effectiveness of single-dose
nevirapine in consecutive pregnancies in Soweto and Abidjan [abstract
722]. In: Program and abstracts of the 13th Conference on Retroviruses
and Opportunistic Infections (Denver). Alexandria, VA: Foundation
for Retrovirology and Human Health, 2006:311.
12. Jourdain G, Ngo-Giang-Huong N, Le Coeur S, et al. Intrapartum ex-
posure to nevirapine and subsequent maternal responses to nevirapine-
based antiretroviral therapy. N Engl J Med 2004; 351:229–40.
13. Lockman S, Shapiro RL, Smeaton LM, et al. Response to antiretroviral
therapy after a single, peripartum dose of nevirapine. N Engl J Med
2007; 356:135–47.
14. Coovadia H, Marais B, Abrams E, et al. Virologic responses to NNRTI
treatment among women who took single-dose nevirapine 18 to 36
months earlier [abstract 641]. In: Program and abstracts of the 13th
Conference on Retroviruses and Opportunistic Infections (Denver).
Alexandria, VA: Foundation for Retrovirology and Human Health,
2006:278.
15. Zijenah L, Kadzirange G, Rusakaniko S, et al. Community-based ge-
neric ART following single-dose nevirapine or short-course zidovudine
in Zimbabwe [abstract 544]. In: Program and abstracts of the 13th
Conference on Retroviruses and Opportunistic Infections (Denver).
Alexandria, VA: Foundation for Retrovirology and Human Health,
2006:236.
BRIEF REPORT • JID 2007:195 (1 March) • 715