Secondary Logo

Journal Logo

Clinical Science

Defining a Cutoff for Atazanavir in Hair Samples Associated With Virological Failure Among Adolescents Failing Second-Line Antiretroviral Treatment

Chawana, Tariro D. MBChB, MSc*; Gandhi, Monica MD, MPH; Nathoo, Kusum MBChB, MRCP, MSc; Ngara, Bernard BSc, MSc§; Louie, Alexander BA; Horng, Howard PhD; Katzenstein, David BA, MD; Metcalfe, John MD, PhD, MPH; Nhachi, Charles F. B. BSc, MSc, MSc, Postgrad DipTox, DPhil*; Adolescent Treatment Failure (ATF) study team

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: September 1, 2017 - Volume 76 - Issue 1 - p 55-59
doi: 10.1097/QAI.0000000000001452



Sustaining optimal antiretroviral (ARV) exposure in HIV-infected adolescents is challenging. Treatment adherence may be compromised by the transition to adulthood while coming to terms with their HIV status for the first time, leading to anger, denial, and rebellion at a time when there is desire for peer conformity.1 Perinatally infected adolescents often have long treatment histories and may be multidrug experienced, leading to treatment fatigue, inadequate adherence, accumulation of drug-resistant strains, and treatment failure.2 Moreover, when to switch from weight-based to age-based dosing remains unclear. Perinatally infected adolescents can demonstrate low body mass index-for-age (BMI-for-age),3 which can lead to toxicity if ARVs are dosed by age or inadequate exposure if dosed by weight.4 Therefore, defining metrics to assess ARV adherence and exposure is critical for monitoring adolescents living with HIV infection.

Accurate measurement of drug exposure is most crucial in the setting of treatment failure, especially when self-reported adherence is claimed to be adequate. Measurement of drug levels in hair integrates, in one assay, both biology (pharmacokinetics) and behavior (adherence) in the preceding weeks to months.5 Hair grows at about 1 cm/mo, incorporating drugs from the circulation as it grows.6 Subtherapeutic ARV drug levels in hair could be a result of suboptimal adherence, reduced absorption, or increased clearance. Supratherapeutic levels can result in toxicity, leading to nonadherence.

The utility of hair concentrations as a metric of adherence and exposure in HIV-infected adults has been explored,5,7 but the study of this measure in adolescents is more limited.8 This study sought to assess the relationship between atazanavir hair concentration and virological outcomes, as well as self-reported adherence, among HIV-infected adolescents with virological failure on atazanavir/ritonavir-based second-line treatment in Harare, Zimbabwe. We also sought to determine whether a home-based adherence intervention increased atazanavir concentrations in hair in the same cohort.


Study procedures were described earlier.9 Briefly, this was a randomized controlled trial conducted among 50 HIV-positive adolescents at Harare Hospital, Zimbabwe, with virological failure (defined by HIV viral load (VL) ≥1000 copies/mL) for ≥6 months on atazanavir/ritonavir-based second-line treatment. The intervention, modified directly administered antiretroviral therapy (mDAART), consisted of structured home visits during the week and SMS text messages during weekends, plus standard-of-care over 90 days.9 The control arm received standard of care alone. At baseline and follow-up, questionnaires were administered using validated instruments, and hair and VL samples were collected.10,11

Hair was collected according to a standard protocol and stored at room temperature until analysis.12 Samples were shipped to the University of California, San Francisco (UCSF) Hair Analytical Laboratory, where hair assays were developed, validated, peer reviewed, and approved by the Division of AIDS Clinical Pharmacology and Quality Assurance.13 Participants without baseline and follow-up hair samples were excluded from this analysis.

Atazanavir hair concentrations were measured using liquid chromatography/mass spectrometry/mass spectrometry, with an assay range of 0.05–20 ng/mg. Of note, the UCSF Hair Analytical Laboratory has previously shown that atazanavir hair levels do not vary significantly by race or ethnicity.14

The study was registered with the Pan African Clinical Trial Registry (PACTR201502001028169) and NIH Clinical (NCT02689895) and approved by the Harare hospital Institutional Review Board, Joint Research Ethics Committee (JREC/51/14), Biomedical and Research Training Institute, Medical Research Council of Zimbabwe (MRCZ/A/1840), Research Council of Zimbabwe (Ref: 02810), and UCSF Committee on Human Research (CHR #11-07442).

Statistical Considerations

Data were entered into Research Electronic Data Capture and analyzed in Stata 14.15,16 Chi-square (and Fisher exact test where appropriate) and Student t tests were used to determine associations between atazanavir hair concentrations and virological outcomes at follow-up, as well as with study group and self-reported adherence. P values (2 sided) were considered statistically significant if ≤0.05. Treatment outcome assessed was VL suppression to <1000 copies/mL.

Factors demonstrating P values <0.25 in bivariate analysis and possible confounders were included in stepwise multivariate logistic regression models assessing the relationship between the mDAART intervention and virological suppression. These factors included age, sex, WHO clinical stage at ART initiation, latest CD4 cell count, total time on ART, VL at follow-up, self-reported adherence [assessed by visual analog scale and AIDS Clinical Trials Group adherence follow-up questionnaire (QLO702)], atazanavir hair concentrations, and BMI-for-age.17

Sample size determination was carried out post hoc on the basis of comparing study arms and follow-up VL (suppressed and unsuppressed) relative to log-transformed follow-up atazanavir hair concentrations using the “power twomeans” function in STATA. The mean log-transformed hair concentration was 1.5 (SD = 0.6) and 1.1 (SD = 0.4) for the control and intervention groups, respectively. Therefore, a sample size of 42 gives 80% power. The mean log-transformed atazanavir hair concentration was 1.6 (SD = 0.7) and 1.0 (SD = 0.5) for suppressed and unsuppressed groups, respectively; therefore, a sample size of 42 gives 83% power.


Participant Demographics and Acceptability of Hair Collection

All 50 adolescents, median age 16 years, 27 females and 23 males receiving atazanavir/ritonavir had mean VL log10 4.8 copies/mL at baseline.9 Twenty-three were randomized to intervention and 27 were randomized to control arms. Two (4%) participants declined to have hair collected both at baseline and during follow-up because they did not want their hairstyles disrupted. Four (8%) and 6 (12%) cut their hair too short before baseline and follow-up visits, respectively, to allow for collection, leading to an overall collection rate of 86%.

Effect of mDAART Intervention

Study arms were well matched (Supplemental Digital Content, Table 1, At baseline, mean atazanavir hair concentration was 1.4 ng/mg. Forty percent reported average adherence <80% over the previous 1 mo, 30% were underweight, whereas 15% were overweight. After follow-up, the mDAART intervention resulted in significantly lower VL (P = 0.048), greater reduction in VL (P = 0.031), higher average self-reported adherence over the previous month (P = 0.05), higher following of the dosing schedule by self-report in the previous 4 d (P < 0.001), and a trend toward increased virological suppression rates to <1000 copies/mL (P = 0.105) (Supplemental Digital Content, Table 2, Atazanavir hair concentrations did not differ significantly in the two arms. In multivariate analysis, participants in mDAART were more likely to report closely following dosing schedule in the past 4 d (RR = 10, 95% CI: 1.9 to 52, P = 0.007) (Supplemental Digital Content, Table 3,

Factors Associated With Virological Suppression Across Both Study Arms

The median and interquartile range (IQR) of atazanavir hair concentrations among those with virological suppression and failure were n = 18; median hair level 3.21, IQR 2.35–6.61 ng/mg and n = 24; median hair level 0.94, IQR 0.16–2.73 ng/mg, respectively. The value of 2.35 ng/mg (lower IQR for those with virological suppression) defined a cutoff below which most participants experienced virological failure (Fig. 1 and Supplemental Digital Content, Table 4,

Box and whisker plots showing atazanavir concentration in hair by virological outcome after follow-up (P < 0.001).

The following characteristics were associated with adequate (>2.35 ng/mg) atazanavir hair concentrations: male sex (P = 0.03), virological suppression after follow-up (P = 0.013), greater reduction in VL (P = 0.006), and change in average self-reported adherence over the previous 1 mo at baseline and at follow-up (P = 0.031) (Supplemental Digital Content, Table 5,

In multivariate analysis, suboptimal (≤2.35 ng/mg) atazanavir hair concentrations were strongly associated with virological failure (RR = 7.2, 95% CI: 1 to 51, P = 0.049) (Table 1).

Multivariate Logistic Regression to Determine Factors Associated With Virological Failure


In this cohort of adolescents living with HIV infection on second-line ART, atazanavir concentrations in small hair samples were the strongest independent predictor of virological suppression. Although other studies have shown this association in adults, our study is the first to explore this association in adolescents with confirmed virological second-line treatment failure.5 Moreover, we have defined in this study, a threshold of atazanavir hair concentrations (2.35 ng/mg) above which most participants experienced virological suppression and below which most participants experienced virological failure. Defining such a cutoff allows hair levels to serve as a useful tool in the clinical setting.

Adolescents are vulnerable to inadequate adherence. Early identification of suboptimal atazanavir hair concentrations and timely interventions in HIV-infected adolescents may ultimately reduce HIV-related morbidity and mortality and curb the spread of HIV infection and drug-resistant HIV strains in this vulnerable group. Suboptimal drug levels could trigger intensive adherence interventions before advanced treatment failure sets in and before evolution of drug resistance occurs. Measurement of antiretroviral drug levels may reduce the need for expensive ARV drug resistance testing, allowing this to be reserved for patients who fail treatment with adequate drug levels in resource-limited settings.7 Where VL measurement is prohibitively expensive, drug levels in hair, if performed economically, can serve as a surrogate for VLs.

The acceptance rate for hair collection in our Harare-based adolescent cohort was high and similar to rates reported in Kenya.6 This finding is encouraging, especially for resource-limited settings where collection of biological samples for research can be challenging because of participant distrust or superstition.18 Hair collection is simple and noninvasive compared with venipuncture, which may increase acceptability and feasibility in children and adolescents who are afraid of needles.

The mDAART intervention was not correlated with atazanavir hair concentrations, but there was a trend toward higher virological suppression rates in those receiving the intervention. Moreover, mDAART participants had increased self-reported adherence after follow-up. Hair concentrations, which reflect accumulation of drug levels from the systemic circulation, may have lagged behind initial increases in plasma drug levels. Studies with longer follow-up in the setting of an ongoing intervention are indicated.

After follow-up, some participants had suboptimal atazanavir hair concentrations while being virologically suppressed, although some virologically failing participants had adequate atazanavir hair concentrations. These findings could reflect a lag in accumulation of atazanavir in hair compared with plasma concentrations and a lag in VL suppression as antiretroviral exposure increases, respectively. Failure despite adequate hair levels could also occur in the setting of virological resistance, necessitating genotype. Studies of longer duration could provide clarity. Finally, despite statistical association, there was poor correlation between self-reported adherence and atazanavir hair concentrations. The most likely explanation for this discrepancy is that self-reported adherence is often overestimated because of social desirability.19

In our study, males were more likely to have adequate atazanavir hair concentrations (defined by hair level of 2.35 ng/mg), consistent with other studies that have found that adherence in adolescent females is generally lower compared with males.4 Girls enter adolescence earlier than boys and hence may be more susceptible to peer conformity/pressure and earlier lapses in adherence. Moreover, girls have earlier sexual debut than boys, with contraception and early child-bearing potentially reducing ART exposure.20

BMI-for-age was not associated with atazanavir hair concentrations. Previous studies report contradictory findings.4 However, these studies were conducted in adult patients, and plasma atazanavir concentrations were measured. This finding needs to be evaluated further in adolescents and with atazanavir hair concentrations.

Our study had a few limitations. Frequent home visits, higher at the beginning of the intervention, decreased as the intervention progressed to reduce intrusiveness and maintain acceptability. Participants could have reduced their adherence as the visits decreased, resulting in suboptimal atazanavir hair levels and virological failure. Plasma drug levels were not measured. However, plasma concentrations only distinguish participants who adhered to treatment as their review days approached (“white coat adherence”) while hair concentrations assess adherence over the past weeks to months. In addition, recruitment was done at one center, limiting generalizability.


Measurement of antiretroviral drug levels in hair allows for measurement of both drug exposure and adherence and predicts virological treatment outcome in multiple cohorts, including our sample of HIV-infected adolescents in Harare. We were able to define a threshold of atazanavir concentrations in hair (2.35 ng/mg), above which there were high rates of virological suppression among adolescents on second-line atazanavir/ritonavir-based ART. As in previous studies in adults, atazanavir hair levels were the strongest independent predictor of virological suppression in our cohort. The mDAART intervention did not increase atazanavir hair levels in our small study with a limited follow-up period but was associated with a trend toward increased virological suppression rates. Hair levels may be a useful tool in resource-constrained settings with limited future treatment options to measure ARV adherence and should be further explored among HIV-infected adolescents.


The authors thank Dr. Leslie Z. Benet, Karen Kuncze, and Nhi Phung from the UCSF HAL for work on hair assays; Dr Rashida Ferrand and Dr Hildah Mujuru for providing guidance; Child Protection Society for providing trained community workers; and Noleen Chifamba, the research nurse.


1. Bakanda C, Birungi J, Mwesigwa R, et al. Survival of HIV-infected adolescents on antiretroviral therapy in Uganda: findings from a nationally representative cohort in Uganda. PLoS One. 2011;6:e19261.
2. Claborn KR, Meier E, Miller MB, et al. A systematic review of treatment fatigue among HIV-infected patients prescribed antiretroviral therapy. Psychol Health Med. 2015;20:255–265.
3. Lowenthal ED, Bakeera-Kitaka S, Marukutira T, et al. Perinatally acquired HIV infection in adolescents from sub-Saharan Africa: a review of emerging challenges. Lancet Infect Dis. 2014;14:627–639.
4. Agwu AL, Fairlie L. Antiretroviral treatment, management challenges and outcomes in perinatally HIV-infected adolescents. J Int AIDS Soc. 2013;16:18579.
5. Gandhi M, Ameli N, Bacchetti P, et al. Atazanavir concentration in hair is the strongest predictor of outcomes on antiretroviral therapy. Clin Infect Dis. 2011;52:1267–1275.
6. Hickey MD, Salmen CR, Tessler RA, et al. Antiretroviral concentrations in small hair samples as a feasible marker of adherence in rural Kenya. J Acquir Immune Defic Syndr. 2014;66:311–315.
7. van Zyl GU, van Mens TE, McIlleron H, et al. Low lopinavir plasma or hair concentrations explain second-line protease inhibitor failures in a resource-limited setting. J Acquir Immune Defic Syndr. 2011;56:333–339.
8. Prasitsuebsai W, Kerr SJ, Truong KH, et al. Using lopinavir concentrations in hair samples to assess treatment outcomes on second-line regimens among Asian children. AIDS Res Hum Retroviruses. 2015;31:1009–1014.
9. Chawana TD, Katzenstein D, Nathoo K, et al. Evaluating an enhanced adherence intervention among HIV positive adolescents failing atazanavir/ritonavir-based second line antiretroviral treatment at a public health clinic. J AIDS HIV Res. 2017;9:17–30.
10. Chesney MA, Ickovics JR, Chambers DB, et al. Self-reported adherence to antiretroviral medications among participants in HIV clinical trials: the AACTG adherence instruments. Patient Care Committee & Adherence Working Group of the Outcomes Committee of the Adult AIDS Clinical Trials Group (AACTG). AIDS Care. 2000;12:255–266.
11. Walsh JC, Mandalia S, Gazzard BG. Responses to a 1 month self-report on adherence to antiretroviral therapy are consistent with electronic data and virological treatment outcome. AIDS. 2002;16:269–277.
12. The Women's Interagency HIV study (WIHS) Manual of Operations. Hair Collection Protocol. Version 10.01.2013. Section 24. Baltimore, MD: WIHS Data Management and Analysis Center at Johns Hopkins Bloomberg School of Public Health; 2013.
13. DiFrancesco R, Tooley K, Rosenkranz S, et al. Clinical pharmacology quality assurance for HIV and related infectious diseases research. Clin Pharmacol Ther. 2013;93:479–482.
14. Huang Y, Gandhi M, Greenblatt RM, et al. Sensitive analysis of anti-HIV drugs, efavirenz, lopinavir and ritonavir, in human hair by liquid chromatography coupled with tandem mass spectrometry. Rapid Commun Mass Spectrom. 2008;22:3401–3409.
15. Harris PA, Taylor R, Thielke R, et al. Research electronic data capture—a metadata-driven methodology and workflow process for providing translational research informatics support. J Biomed Inform. 2009;42:377–381.
16. Statistical Software: Release 14.0. College Station, TX: Stata Corporation; 2001.
17. World Health Organisation Multicentre Growth Reference Study Group. WHO child growth standards based on length/height, weight and age. Acta Paediatr Suppl. 2007;95:76–85.
18. Coetzee B, Kagee A, Tomlinson M, et al. Reactions, beliefs and concerns associated with providing hair specimens for medical research among a South African sample: a qualitative approach. Future Virol. 2012;7:1135–1142.
19. Berg KM, Arnsten JH. Practical and conceptual challenges in measuring antiretroviral adherence. J Acquir Immune Defic Syndr. 2006;43:S79–S87.
20. Nyamukapa CA, Gregson S, Lopman B, et al. HIV-associated orphanhood and children's psychosocial distress: theoretical framework tested with data from Zimbabwe. Am J Public Health. 2008;98:133–141.

adolescents; virological failure; atazanavir hair concentrations

Supplemental Digital Content

Copyright © 2017 Wolters Kluwer Health, Inc. All rights reserved.