Patterns of detectable viraemia among children and adults with HIV infection taking antiretroviral therapy in Zimbabwe

(1)

Patterns of detectable viraemia among children and adults with HIV infection taking antiretroviral therapy in Zimbabwe

Evgeniya Sovershaeva

^a,

*, Tinei Shamu

^b

, Tom Wilsgaard

^a

, Tsitsi Bandason

^c

, Trond Flægstad

^d

, David Katzenstein

^e

, Rashida A. Ferrand

^c,f

, Jon Odland

^a,g

aDepartmentofCommunityMedicine,FacultyofHealthSciences,UiT—TheArcticUniversityofNorway,Tromsø,Norway

bNewlandsClinic,Harare,Zimbabwe

cBiomedicalResearchandTrainingInstitute,Harare,Zimbabwe

dDepartmentofPaediatrics,UniversityHospitalofNorthNorway,Tromsø,Norway

eDivisionofInfectiousDiseases,StanfordUniversity,California,USA

fClinicalResearchDepartment,LondonSchoolofHygieneandTropicalMedicine,London,UK

gDepartmentofPublicHealthandNursing,FacultyofMedicineandHealthSciences,NTNUTheNorwegianUniversityofScienceandTechnology,Trondheim, Norway

ARTICLE INFO Articlehistory:

Received19September2018

Receivedinrevisedform24October2018 Accepted25October2018

CorrespondingEditor:EskildPetersen,Aar- hus,Denmark

Keywords:

Antiretroviraltherapy HIV

Sub-SaharanAfrica Viralblip Viralload Viraemia

ABSTRACT

Objective:Toinvestigatetheincidenceandpredictorsofviraemiaamongindividualsonantiretroviral therapy(ART)inHarare,Zimbabwe.

Methods: Children(0–19 years)and adults(>19years) startingARTbetween2013and 2015were followedforamedianof2.8and2.7years,respectively.Theincidenceratesofvirologicalfailure(VF), low-levelviraemia(LLV),andviralblipswereassessedandthepredictorsofviraemiaweredetermined usinglogisticandparametricsurvivalregressionanalyses.

Results:Atotalof630individualsinitiatedART,and19.7%ofchildrenand5.6%ofadultsdidnotachieve viralsuppressionby12months.YoungerageandCD4count200cells/mm³atbaselinewereassociated withnotbeingvirallysuppressedat12monthsinadults.Amongthosewhoachievedviralsuppression duringthefollow-upperiod,theincidenceofVFwashigherinchildren(4.0/100person-yearsvs.0.4/100 person-yearsinadults;p<0.001),aswastheincidenceofLLV(1.9/100person-yearsvs.0.3/100person- yearsinadults;p=0.03).Theincidencerateofblipswas10.9per100person-yearsinchildrenand4.0per 100person-yearsinadults.

Conclusions:Childrenarelesslikelytoreachviralsuppressionandareathigherriskofviraemiawhileon ARTthanadults.ThesigniﬁcanceofLLVandblipsneedsfurtherstudy.

ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by- nc-nd/4.0/).

Introduction

Access to antiretroviral therapy (ART) has substantially increasedsurvivaland improvedqualityof lifefor HIV-infected individualsworldwide(AntiretroviralTherapyCohortC,2017;De LaMataetal.,2016).SustainedviralsuppressionachievedonART reducestherisk of immunodeﬁciency, clinicalprogression, and HIVtransmission(Cohenetal.,2016).AccordingtoWorldHealth Organization(WHO)guidelines,anHIVviralload(VL)of1000 copies/mlshouldpromptenhancedcounsellingandarepeat VL

test within 6 weeks, and two sequential VL measurements of 1000copiesisconsideredvirologicalfailure(VF)andshouldlead toregimenchange(WorldHealthOrganization,2016).

A proportion of individuals on ART experience low-level viraemia(LLV)and/ortransientviraemia;forexample,viralblips havebeenreportedinupto40%ofHIV-positiveindividualsonART (Grennanetal.,2012;Havliretal.,2001;Sorstedtetal.,2016).Viral blips have been shown to be associated with higher HIV pre- treatmentVL andlowerCD4countatthetimeofARTinitiation (Farmeretal.,2016;Havliretal.,2001;Sorstedtetal.,2016).There is someevidence that they may impair CD4cell recovery and maintain ongoing low-grade immune activation (Taiwo et al., 2013;Zoufalyetal.,2014).

DataondetectableviraemiaafterARTinitiationarescarceand particularlytheincidenceandsigniﬁcanceoftransientand/orLLV in low-incomesettings, where– unlike high-incomesettings –

* Correspondingauthorat: Department ofCommunity Medicine,Facultyof HealthSciences,UiT—TheArcticUniversityofNorway,HansineHansensveg18, 9019Tromsø,Norway.

E-mailaddress:evgeniya.sovershaeva@uit.no(E.Sovershaeva).

https://doi.org/10.1016/j.ijid.2018.10.019

1201-9712/©2018TheAuthors.PublishedbyElsevierLtdonbehalfofInternationalSocietyforInfectiousDiseases.ThisisanopenaccessarticleundertheCCBY-NC-ND license(http://creativecommons.org/licenses/by-nc-nd/4.0/).

InternationalJournalofInfectiousDiseases78(2019)65–71

ContentslistsavailableatScienceDirect

International Journal of Infectious Diseases

j o u r n a l h o m ep a g e : w w w . e l s e v i e r . c o m / l o c a te / i j i d

(2)

regularVLmonitoringisnotthenorm(Haasetal.,2015).Thefew studiesthathavebeenconductedinlow-incomecountrieshave focusedmainlyonadults(Kanapathipillaietal.,2014).However, childrenarepotentiallyatahigherriskofdevelopingviraemiadue toweight-baseddosing,whichmayleadtovariabledruglevels, poortolerabilityofdrugs,andsuboptimaladherence(Easterbrook etal., 2002; Youngetal., 2015).Moreover,psychosocialfactors includingdependenceonacaretakeranddelayeddisclosureofHIV statustothechild mayputchildrenathigherriskofremaining viraemicafterARTinitiation(Lall etal.,2015).Adolescenceis a speciﬁc high-risk period for poor adherence in many chronic conditions(Taddeoetal.,2008).

In thisstudy, theincidenceofand riskfactorsfordetectable viraemiaincludingVF,LLV,andblipswereinvestigatedinacohort ofHIV-infected children and adultsinitiated onART in Harare, Zimbabwe.

Methods

A retrospective cohort study was conducted using data collectedfrom patientswho attended Newlands Clinic, Harare, Zimbabwe, a not-for-proﬁt HIV clinic that provides care for childrenandadults.ARTisprovidedfreeofchargeaccordingto national guidelines. Routine 6-monthly VL monitoring is per- formed(duringthestudyperiod,thiswasdoneusingtheRoche COBAS AmpliPrep/COBAS TaqMan48 version 2.0 test; Roche MolecularSystems,Pleasanton,CA,USA).Adherenceandpsycho- socialcounsellingisprovidedateachclinicvisit(atleastevery3 months).Adherenceisassessedbypillcountforeachantiretroviral drug at every clinic visit. In the case of suspected VF (a VL 1000copies/ml), the patient receives intensiﬁed adherence counsellingfor 6 weeksand repeated VL testing.Thosewith a VL1000copies/ml in two consecutive VL measurements are considered to have VF and are switched to second-line ART (proteaseinhibitor(PI)-basedregimenandchangeofatleastone nucleoside reverse transcriptase inhibitor (NRTI)). Those on second-line ART who have a VL 1000 copies/ml despite counsellingundergoHIVdrugresistancetestingandareconsid- eredforthird-lineART.

Individuals who wereART-naïve and initiated ART between August2013andAugust2015andwhohadatleasttwoVLtests afterARTinitiationwereincludedinthestudy.Thefollowingdata were extracted: age, sex, date of ART initiation, ART regimen, adherence,height,weight,clinicalhistory(WHOHIVdiseasestage, historyof tuberculosis,opportunisticinfections, chroniccomor- bidities),andlaboratoryparameters(VL,CD4count,haemoglobin) atthetimeofARTinitiationandduringthefollow-upperioduntil September28,2017.

Dataanalysis

Allpatientdataarestoredinasecureelectronicdatabase.The datawereanonymizedpriortoanalysis.Statisticalanalyseswere performedinStata14(StataCorpLLC,CollegeStation,TX,USA).The outcomesweretheincidenceofVF,LLV,andviralblips.Viralblip wasdefined asa VL measurement50copies/ml precededand followed by a VL below the limit of detection (<50copies/ml) (Havliretal.,2001; Kanapathipillaietal., 2014;Martinezetal., 2005).LLVwasdefinedasaVL50to<1000copies/mlinatleast twoconsecutiveVLtests.TheWHOdefinitionforVFwasused(VL 1000copies/ml in two consecutive VL measurements) (World HealthOrganization,2016).

The proportion of participants who did not achieve viral suppression by 12 months of ART (the 12-month cut-off was chosentoallowforaVLtesttoconﬁrmsuppressionatmonth12)

was estimated and thefactors associated withvirologicalnon- suppressionwerestudiedusinglogisticregression.

Theincidenceratesofviralblips,LLV,andVFamongthosewho hadachievedviralsuppressionduringthefollow-upperiodand had at least two VL tests after initial VL suppression were estimated.Nelson–Aalencumulativehazardcurveswereplottedto evaluate the incidence of VF, LLV, and viral blips after VL suppression byageat ARTinitiation (childrenaged 0–19years andadultsaged>19years).

Thefactorsassociatedwiththeoccurrenceofviralblipswere investigatedusing survivalanalysis. Forthe model,participants wereincludedintheanalysisattime0(timeoffirstsuppressedVL testafterARTinitiation)andfolloweduntilaviralblipoccurred,or forthosepatientswhoremainedsuppresseduntilthelastVLtest available,bytheendofthefollow-up.Participantswhoreported treatment interruption in ART for more than 2 weeks were excludedfromthisanalysis.Sincetheestimatedcumulativehazard of blips increased exponentially with time, we fitted the parametricsurvivalregressionwithWeibulldistributionstratified byagegroup.AvalueofP(theshapeparameter)>1confirmedthat thehazardoffailure(viralblip)increasedwithtime.

Age,sex,bodymassindex(BMI),stunting(inchildrenonly), pre-treatmentVL,CD4count,anaemia,WHOclinicalstageatART initiation, history of tuberculosis beforeART initiation, chronic comorbidities(adultsonly),ARTregimen,andaverageadherence wereinvestigatedaspredictorsofdetectable viraemia.Stunting was deﬁned asa height-for-age Z-score of< 2 (childrenonly).

Height-for-ageZ-scoresandBMI-for-ageZ-scoresinchildrenwere calculatedusingWHOreferencestandards.Anaemiawasdeﬁned accordingtoWHOcriteria(haemoglobin<11g/dlforchildren<5 years;haemoglobin<11.5g/dlforchildren5–11.99years;haemoglobin<12g/dlforchildren12–14.99years;haemoglobin<12g/dl forfemalesaged15years;haemoglobin<13g/dlformalesaged 15 years). Adherence was calculated as a percentage of the numberoftabletsdispensedatthelastvisitminusthenumberof tabletsreturnedatthecurrentvisitdividedbynumberoftablets that should have been consumed between visits. The average adherence over the study period was calculated for each participant.Age, sex,and CD4countwereadjusted fora priori.

Allstatisticaltests weretwo-tailedandp-values of <0.05 were consideredstatisticallysigniﬁcant.

EthicalapprovalforthestudywasobtainedfromtheNewlands Clinic Research Committee, the Medical Research Council of Zimbabwe, and Regional Committee for Medical and Health ResearchEthics(Norway).

Results

Of the725 participants who initiated ART duringthe study period,17wereexcludedastheyweresuppressedatbaseline(and thusmayhavereceivedARTpreviously)and78wereexcludedfor having <2 VL measurements available following ART initiation (Figure 1). The baseline demographic, clinical, and laboratory characteristicsoftheremaining630participants(127childrenand 503adults)areshowninTable1.Themedianfollow-uptimewas 2.8(interquartilerange(IQR)2.3–3.2)yearsforchildrenand2.7 (IQR1.8–2.8)yearsforadults.Thesexdistributionwasequalinthe paediatricgroup(49.6%femalevs.50.4%male),whiletherewere moreadultfemalethanmaleparticipants(62.2%femalevs.37.8%

male),consistentwithstudiesinSub-SaharanAfrica,whichhave shownhigher HIV treatmentcoverageamongwomen (UNAIDS, 2013). Twenty-one children, all aged below 3 years at ART initiation, were commenced on PI-based regimens, as per the nationalguidelines,whichrecommendPIsaspartoftheﬁrst-line regimeninchildrenbelow3years.Tenadultsinitiatedtreatment with a PI-based regimen based on clinician judgement, with

66 E.Sovershaevaetal./InternationalJournalofInfectiousDiseases78(2019)65–71

(3)

reasonsincludingsevereanaemia,Kaposisarcoma,WHOstage4, thirdtrimesterofpregnancy,andbreastfeeding.

Ofthe630participants,signiﬁcantlymorechildrenthanadults didnotachieveviralsuppressionby12monthspostARTinitiation (19.7%childrenvs.5.6%adults;p<0.001).Younger ageand CD4

count 200cells/mm³ at baseline were associated with not achievingviralsuppressioninadults(Table2).Overthefollow- upperiod,106(83.5%)childrenand482(95.8%)adultsreachedVL suppression,withthemediantimetoVLsuppressionbeing0.5(IQR 0.4–1.3)yearsforchildrenand0.5(IQR0.2–0.8)yearsforadults.

Table1

Characteristicsofstudyparticipants.Resultsarepresentedasthenumber(percentage),orasthemedian(interquartilerange),unlessindicatedotherwise.

Characteristics Children(n=127) Adults(n=503)

Demographic

Male 64(50.4%) 190(37.8%)

AgeatARTinitiation,years 10(3–15) 37(31–44)

StuntedatARTinitiation(height-for-ageZ-score< 2) 24(18.9%)

UnderweightatARTinitiation(BMIZ-score< 2forchildrenorBMI<18.5kg/m²foradults) 13(10.2%) 40(8.0%) Clinical

Historyoftuberculosis 10(7.9%) 59(11.7%)

Prevalenceofchroniccomorbidities^a 2(1.6%) 90(17.9%)

Opportunisticinfections^b 13(10.2%) 75(14.9%)

WHOclinicalstage3or4atARTinitiation^c 31(24.8%) 144(28.9%)

ARTregimenattreatmentinitiation

2NRTI+NNRTI 106(83.5%) 493(98.0%)

PI-based 21(16.5%) 10(2.0%)

AverageadherencetoARTbypill-count 98.4% 99.6%

YearsonART 2.8(2.3–3.2) 2.8(1.8–2.8)

Laboratory

CD4countatARTinitiation,cells/mm³ 341(137–733) 220(104–334)

ViralloadatARTinitiation,log10copies/ml^d 4.8(4.4–5.3) 4.8(4.3–5.2)

AnaemiaatARTinitiation 82(64.6%) 248(49.4%)

ART,antiretroviraltherapy;BMI,bodymassindex;WHO,WorldHealthOrganization;NRTI,nucleosidereversetranscriptaseinhibitor;NNRTI,non-nucleosidereverse transcriptaseinhibitor;PI,proteaseinhibitor.

aChroniccomorbiditiesincludedhypertension,diabetes,orrenalfailurediagnosedbeforeinclusioninthestudy.

b AfterARTinitiation,hadatleastoneepisodeofoneofthefollowing:oralcandidiasis,necrotizinggingivitis,herpeszoster,histoplasmosis,cryptococcalmeningitis, molluscumcontagiosum,genitalwarts,tonsillitis.

cDatamissingfortwochildrenandﬁveadults.

d Datamissingfor42childrenand159adults.

Figure1.Flowchartofparticipantrecruitment.

E.Sovershaevaetal./InternationalJournalofInfectiousDiseases78(2019)65–71 67

(4)

Ofthe588participantswhoachievedviralsuppressionafterARTinitiation,516hadsufﬁcientVLdatatostudytheincidenceofdetectableviraemia.Overthefollow-upperiod,57(11%)partic-ipantsexperiencedaviralblip.Blipsoccurredmorecommonlyinchildrenthaninadults(22.9%vs.8.3%;p<0.001),withtheincidenceratebeing10.9(95%conﬁdenceinterval(CI)7.2–16.6)per100person-yearsinchildrenand4.0(95%CI2.8–5.5)per100person-yearsinadults(Figure2A).Fiftypercentofblipsinchildrenand71.4%ofblipsinadultswereoflowmagnitude(50–199copies/ml).Themediantimefromviralsuppressiontoablipwas1.9(IQR1.4–2.3)yearsinchildrenand1.8(IQR1.0–2.1)yearsinadults.Seven(1.4%)participants(fourchildrenandthreeadults)experiencedLLVand12(2.3%)participants(eightchildrenandfouradults)developedVFafterinitialsuppressionduringfollow-up.TheincidenceofLLVwas1.9(95%CI0.7–5.1)per100person-yearsinchildrenand0.3(95%CI0.1–1.0)per100person-yearsinadults Table 2

Logistic regression analysis of the risk factors for not achieving viral suppression by 12 months of ART.

Characteristic Children (19 years of age) (n= 127) Adults (>19 years of age) (n= 503)

Unadjusted analysis Adjusted analysis^a Unadjusted analysis Adjusted analysis^a

OR (95% CI) p-Value OR (95% CI) p-Value OR (95% CI) p-Value OR (95% CI) p-Value

Female 1.69 (0.69–4.11) 0.25 1.71 (0.69–4.19) 0.24 0.59 (0.27–1.26) 0.17 0.72 (0.33–1.60) 0.43

Age at ART initiation (per 1 year older) 1.05 (0.98–1.13) 0.18 1.06 (0.96–1.18) 0.23 0.97 (0.93–1.01) 0.15 0.95 (0.91–1.00) 0.04

Stunted at ART initiation (height-for-ageZ-score< 2), yes vs. no 0.53 (0.14–1.93) 0.33 – – – – – –

BMIZ-score (children) or BMI kg/m²(adults) at ART initiation 0.78 (0.57–1.07) 0.12 0.79 (0.56–1.13) 0.21 0.91 (0.84–0.98) 0.01 0.94 (0.87–1.03) 0.22

History of TB, yes vs. no – – – – 1.70 (0.62–4.64) 0.30 0.92 (0.31–2.74) 0.89

Chronic comorbidities^b, yes vs. no – – – – 0.53 (0.16–1.81) 0.32 0.86 (0.24–3.16) 0.83

WHO clinical stage 3 or 4 at ART initiation, yes vs. no 1.23 (0.46–3.30) 0.68 1.44 (0.51–4.06) 0.49 2.24 (1.04–4.84) 0.04 1.18 (0.50–2.83) 0.70

ART regimen at treatment initiation (Ref. 2NRTI + NNRTI)^c 0.64 (0.17–2.36) 0.50 1.06 (0.17–6.72) 0.95 – – – –

Average adherence to ART by pill-count, % 0.88 (0.72–1.07) 0.21 0.85 (0.69–1.05) 0.14 0.97 (0.89–1.06) 0.54 1.00 (0.91–1.09) 0.92

CD4 count at ART initiation, cells/mm³

200 cells/mm³ 2.55 (1.02–6.36) 0.04 2.21 (0.78–6.29) 0.14 5.66 (2.11–15.1) 0.001 5.92 (2.18–16.1) <0.001

>200 cells/mm³ Ref. Ref. Ref. Ref.

Viral load at ART initiation, log10copies/ml 1.15 (0.63–2.11) 0.65 1.48 (0.73–3.00) 0.27 1.78 (1.00–3.18) 0.05 1.26 (0.68–2.33) 0.46

Anaemia at ART initiation, yes vs. no 1.53 (0.58–4.00) 0.38 1.46 (0.55–3.88) 0.45 2.26 (1.00–5.10) 0.05 1.73 (0.72–4.19) 0.22

ART, antiretroviral therapy; OR, odds ratio; CI, conﬁdence interval; BMI, body mass index; TB, tuberculosis; WHO, World Health Organization; NRTI, nucleoside reverse transcriptase inhibitor; NNRTI, non-nucleoside reverse transcriptase inhibitor; PI, protease inhibitor.

aAdjusted for CD4 count at ART initiation, age, and sex.

bChronic comorbidities included hypertension, diabetes, or renal failure diagnosed before inclusion in the study.

cPI-based regimen as the exposure.

Figure2.Nelson–Aalencumulativehazardcurvesforthedevelopmentof(A)viralblips,(B)low-levelviraemia(LLV),and(C)virologicalfailure(VF)afterviralloadsuppression.Curvesforchildrenarepresentedinredandcurvesforadultsinblue. 68E.Sovershaevaetal./InternationalJournalofInfectiousDiseases78(2019)65–71

(5)

(Figure2B).VFwasmorecommoninchildrencomparedtoadults (8.3%vs.0.9%;p<0.001).TheincidenceofVFwas4.0(95%CI2.0– 7.9)per100person-yearsinchildrenand0.4(95%CI0.2–1.2)per 100person-yearsinadults(Figure2C).Notably,2.2%ofchildren

<10yearsand13.7%ofthose10yearsdevelopedVF(p=0.05).All fouradultswithVFhadre-suppression(threefollowingaswitch from a non-nucleoside reverse transcriptase inhibitor (NNRTI)- basedtoaPI-basedregimen,oneremainingonthesameNNRTI- basedregimen),whileonlythreechildrenhadre-suppression(two followingaswitchfromanNNRTI-basedtoaPI-basedregimen, one remaining on the same NNRTI-based regimen) during the follow-upperiod.

Fouroutof22children(18.2%)experiencedbothviralblipsand VF during the follow-up. Blips were followed by VF in two participants,whileblipsoccurredafterVFandre-suppressionin theothertwocases.InthoseparticipantswithblipsfollowedbyVF, themagnitudeofblipswaslow,whileinthosewhohadblipsafter VF, blips were of medium (200–499copies/ml) and high (>500copies/ml)magnitude.Amongtwochildrenwithblipsafter VF, one had re-suppression on the same ART regimen with adherencecounselling,whileanotherhadre-suppressionaftera switchtoaPI-basedregimen.Twochildrenwithblipsfollowedby VFremainedunsuppressedbytheendoffollow-up.

DuetolowratesofLLVinthestudyparticipantsafterinitialVL suppression,riskfactorsforonlyviralblipswereinvestigatedin thesurvivalanalysis. The survivalanalysisincluded507partic- ipants(94inthepaediatricgroupand413intheadultgroup).No baseline characteristics were found to be associated with an increasedriskofviralblipsinchildrenoradults(Table3).

Discussion

AmongindividualsstartingART,asigniﬁcantlyhigherpropor- tion of children compared to adults did not achieve viral suppressionby12months.Likelyreasonsincludepoortolerability to antiretroviral drugs, inadequate ART dosing, or suboptimal adherence(Boermaetal.,2016).Thisisinlinewiththeﬁndingsof Jobanputraetal.,whoshowedthatbeingachildoradolescentis associatedwithdetectableviraemia(Jobanputraetal.,2015).Some individualsmayrequirelongerperiodstoachievesuppression,as shown in the present study, where over the follow-up period, 83.5%ofchildrenand95.8%ofadultsdidachieveviralsuppression.

OnepossibilityisthatperinatallyHIV-infectedyoungchildrentend to have sustained high VL in the ﬁrst years of life with slow reductionofpeakVLwithincreasingage(McIntoshetal.,1996).

This potentially could contribute to incomplete initial VL suppressionandahigherriskofviraemia.Yetthepresentstudy resultsdonotsupportthis,astheexclusionofchildrenaged3 years didnot change therates of viral non-suppression in the paediatric study group. Lower CD4 count at baseline was associatedwithnotbeingsuppressedat12monthsin adults,a ﬁnding consistent with other studies (Anude et al., 2013;

Collaboration of Observational et al., 2008; Mujugira et al., 2016;Samueletal.,2014).

Thisstudyfoundasignificantlyhigherincidenceofviralblips among childrencompared toadults. The reportedincidence of blipsvariesindifferentpopulations(Farmeretal.,2016;Grennan etal.,2012;Havliretal.,2001;Kanapathipillaietal.,2014;Sorstedt etal.,2016),largelyduetodifferentdefinitionsandvariabilityof theassaysforVL.Weusedacommondefinitionofviralblip,asa singleVLmeasurement50copies/mlprecededandfollowedby VL <50copies/ml (Fung et al., 2012; Ryscavage et al., 2014).

Comparedtostudiesthathaveused asimilardeﬁnition,it was found that a lower proportion of adults in the present study experiencedblips:8.3%during2.1yearsoffollow-upcomparedto

18.6–40%followedupfrom1.08to1.58years(Havliretal.,2001; Table3 Riskfactorsforthedevelopmentofviralblipsbyage. CharacteristicChildren(19years)Adults(>19years) aaUnadjustedanalysisAdjustedanalysisUnadjustedanalysisAdjustedanalysis HR(95%CI)p-ValueHR(95%CI)p-ValueHR(95%CI)p-ValueHR(95%CI)p-Value Female1.41(0.60–3.32)0.431.08(0.42–2.76)0.870.79(0.39–1.58)0.500.76(0.37–1.56)0.45 AgeatARTinitiation,years0.99(0.91–1.07)0.741.06(0.92–1.22)0.431.00(0.96–1.04)0.871.00(0.96–1.04)0.91 StuntedatARTinitiation(height-for-ageZ-score<2)0.46(0.13–1.67)0.240.39(0.08–2.01)0.26–––– 2BMIZ-score(children)orBMIkg/m(adults)atARTinitiation1.12(0.84–1.49)0.440.87(0.55–1.37)0.540.97(0.92–1.03)0.310.97(0.91–1.03)0.33 HistoryofTB2.17(0.60–7.94)0.242.99(0.87–10.3)0.080.87(0.27–2.81)0.820.85(0.24–3.00)0.80 bChroniccomorbidities––––1.17(0.50–2.76)0.711.19(0.50–2.86)0.69 WHOclinicalstage3or4atARTinitiation0.62(0.19–1.98)0.420.75(0.21–2.74)0.660.78(0.33–1.83)0.570.78(0.31–1.94)0.60 cARTregimenattreatmentinitiation(Ref.2NRTI+NNRTI)1.92(0.72–5.11)0.191.65(0.44–6.26)0.460.94(0.13–6.97)0.950.99(0.13–7.42)0.99 AverageadherencetoARTbypill-count,%0.91(0.67–1.24)0.560.90(0.66–1.22)0.501.16(0.86–1.56)0.341.17(0.86–1.58)0.31 Monthstoviralloadsuppression1.06(0.99–1.14)0.091.06(0.98–1.16)0.141.03(0.96–1.12)0.381.03(0.96–1.11)0.35 3CD4countatARTinitiation,logcells/mm2.21(0.84–5.82)0.113.83(0.53–27.5)0.181.19(0.57–2.50)0.641.29(0.61–2.73)0.5010 ViralloadatARTinitiation,logcopies/ml1.01(0.43–2.34)0.991.06(0.41–2.73)0.911.58(0.95–2.62)0.081.56(0.98–2.50)0.0610 AnaemiaatARTinitiation2.33(0.78–6.97)0.131.99(0.65–6.10)0.230.91(0.46–1.81)0.790.97(0.44–2.13)0.94 HR,hazardratio;CI,conﬁdenceinterval;ART,antiretroviraltherapy;BMI,bodymassindex;TB,tuberculosis;WHO,WorldHealthOrganization;NRTI,nucleosidereversetranscriptaseinhibitor;NNRTI,non-nucleosidereverse transcriptaseinhibitor;PI,proteaseinhibitor. aAdjustedforCD4countatARTinitiation,age,andsex. bChroniccomorbiditiesincludedhypertension,diabetes,orrenalfailurediagnosedbeforeinclusioninthestudy. cPI-basedregimenastheexposure.

E.Sovershaevaetal./InternationalJournalofInfectiousDiseases78(2019)65–71 69

(6)

Martinez et al., 2005; Sklar et al., 2002). Several studies have shownthatpatientsonPIregimensaremorepronetodeveloping viralblips(Grennanetal.,2012;Sorstedtetal.,2016).Havliretal.

(2001) reporteda highincidence of blips (40%) in a cohort of patientswhoreceivedanunboostedPI-basedtreatmentregimen inahigh-incomecountry.Inaddition,themediannumberofVL measurementswas17perparticipant,comparedtoanaverageof sixmeasurementsinthepresentstudyinwhichonly2.1%ofadults werereceivingaritonavir-boostedPI-basedregimen.

It is likely that more frequent VL testing will increase the probabilityofdetectingblips,whichmayexplainthediscrepancies betweenour ﬁndings and those of studies conducted in high- incomesettings.However,inastudycomparingtheincidenceof blips in high-, middle-, and low-income countries, 21% of the individualsintheﬁrstcategoryexperiencedblips,whileonly11%

ofparticipantsdidsointhemiddle/lowincomegroup,despitea similarratioofblipstothenumberofVLmeasurementsacrossthe settings(Kanapathipillaietal.,2014).HIVplasmaRNAmeasure- mentsaresubjecttopre-analyticandlaboratoryerrors,variations incut-offpoint,andvariationsinlowerlimitofdetectionamong thedifferentVLassays,whichmayhaveanimpactontheincidence ofblips(Grennanetal.,2012;Youngetal.,2015).Eventhoughgood agreementbetweenthe differentassaysis observedat highVL levels,thereissubstantialvariabilityatlowVLlevels(Ruelleetal., 2012;Swensonetal.,2014a).

Thereappeartobelimiteddataontheincidenceofviralblipsin childreninthelow-incomesetting(Jobanputraetal.,2015;Szubert etal.,2017).InaretrospectiveanalysisofVLtestsamongchildren who initiated ART in Uganda and Zimbabwe, 46% of children experiencedviralblips(Szubertetal.,2017).Thisisconsiderably higherthantheresultsofthepresentstudy,butmaybeexplained bytheuseofWHO2006criteriaforARTinitiation,whichreliedon clinical and/or immunological assessment and not just the diagnosisofHIVinfection.Themajorityofstudiesinlow-income countries have investigated the incidence of any detectable viraemiainoneortwoVLmeasurements.Forexample,across- sectionalstudyamongadolescentswithHIVconductedinanurban settinginCameroonutilizedtwoconsecutiveVLmeasurementsto detectsustainedVLsuppression(VL<50copies/mlintwoVLtests) (Fokametal.,2017).Inthatstudy,18.6%ofparticipantshadaVL measurementabove50copies/mlinasingletest.Anotherstudy conductedin Ethiopiafounddetectable viraemia(defined asan HIV-1RNAof41–1000copies/ml)inasingleVLmeasurementin 13%ofHIV-positivechildrenwhohadreceivedfirst-lineARTfora medianof24months(Muluetal.,2014).InastudybyJobanputra etal.,38%ofchildrenaged<10yearsand34%ofchildrenaged10– 19yearshadsingledetectableVL(>100copies/ml)followedbyVL re-suppression(Jobanputraetal.,2015).Althoughtheresultsfrom thesestudies arenot directly comparable tothepresent study findings, they show that detectable viraemia is common in paediatricpatients.Amulticentrecohortstudyofdataonchildren withHIVintheUKandIrelandshowedthat22%ofparticipants withsustainedviralsuppressionexperiencedtransientviraemia (definedasasingleVL>50copies/mlbetweentwoVLtestsbelow 50copies/ml), with an incidence of 12 per 100 person-years, similartothefindingsinourstudy(Leeetal.,2007).

TheincidenceofLLVandVFafterreachingVLsuppressionwas alsosigniﬁcantlyhigherinchildrenthaninadults.Furthermore,in thepresentstudy,adolescents(thoseaged10years)weremore likelytoexperienceVFcomparedtoyoungerchildren,aﬁnding thathasalsobeenreportedbyotherstudiesconductedinAfrica (Makadzangeetal.,2015).Adolescenceisaperiodofparticularly highriskforpooradherence(Mukuietal.,2016;Nachegaetal., 2009).Takentogether,theseemphasizetheneedforinvestigating and addressing factors that impede viral suppression among adolescents.

TherateofVFinchildrenwhoreachedVLsuppressionislower than thosereportedbyotherstudies (Makadzange etal., 2015;

Salou et al., 2016). A possible reason for the higher VF rates reportedinotherstudiesisthecross-sectionaldesignofstudies andlackofinformationregardinginitialVLsuppressionpriorto inclusioninthestudy(Fokametal.,2017;Salouetal.,2016).Dueto regularVLmonitoringandthelongitudinaldesignofourstudy,it waspossibletodistinguishbetweenthosewhodevelopedVFand thosewhoneverreachedsuppression.Anotherexplanationforlow VFinourcohortisthehighstandardofcareprovidedatNewlands Clinic,withroutineVLtesting,continuousadherencemonitoring support, and a rapid response with intensiﬁed adherence counselling and/or ART switch among those who have a non- suppressedVL(Haasetal.,2015).RoutineVLtesting,whichisnot available asstandard of carein manylow-incomesettings, can detect VF earlier than targeted VL monitoring, i.e., VL testing promptedbyclinicalorimmunologicaldeterioration.

AnumberofstudieshaveshownthatepisodesofLLVandviral blipsareassociated withanincreasedriskof subsequentVFin adults(AntiretroviralTherapyCohortCetal.,2015;Grennanetal., 2012;Lapriseetal.,2013;Leiereretal.,2016).Whilesomestudies havesuggestedthatepisodesoftransientviraemiamayresultin theselectionofdrug-resistantHIVstrains,leadingtoanincreased riskofVF(Clutteretal.,2016;Gonzalez-Sernaetal.,2014;Swenson etal.,2014b),othershavefoundnolinkbetweenblipsandHIV progression(Havliretal.,2001;Kanapathipillaietal.,2014;Nettles etal.,2005).Youngetal.foundagradualincreaseintheriskofVF withincreasingblip magnitude(Younget al.,2015).In another study,onlyblipsof500–999copies/mlmagnitudewereassociated withtheincreasedriskofviralrebound(Grennanetal.,2012).In contrast,otherstudieshaveshownnoevidenceofanassociation between blips and VF in adults (Kanapathipillai et al., 2014;

Martinezetal.,2005;Sklaretal.,2002).However,thesestudies havebeenconductedlargelyinadults.Inthepresentstudy,not onlydidchildrenhaveahigherriskofVF,buttheyalsohadahigher incidenceofviralblipsaswellasLLV,andthesigniﬁcanceofthese in predicting VF needs further study. Furthermore, careful virologicalmonitoringiswarrantedinchildren.

The strengths of this study are the use of a standardized deﬁnition for viral blips, regular VL monitoring, availability of detailedclinicaldataincludingadherencedata,andgoodfollow- uprates.Thelimitationsofthestudyincludetherelativelysmall samplesizeandlownumberofblipsineachagegroup,andthere may therefore be inadequate power to detect an association betweenthecovariatesandtheincidenceofviralblips.Further- more,thefollow-up period wasnot longenoughtoinvestigate whetherblipsorLLVincreasetheriskofsubsequentVF.Adherence mayhavebeenover-estimated,asthepillcountisnottheoptimal measureofadherence.

Inconclusion,thisstudydemonstratedthatdetectableviraemia iscommonamongchildren,althoughitsrolewithregardtolong- termoutcomesnecessitatesfurtherstudyinlow-incomesettings, especiallywhereHIVprevalenceishighandtheavailabilityofVL monitoringremainslimited.

Acknowledgements

WearegratefultothestaffandpatientsofNewlandsClinicfor theirassistance.

Funding

ThisstudywassupportedbygrantfromHelseNord(HNF1387- 17).RAFisfundedthroughaWellcomeTrustSeniorFellowshipin Clinical Science (206316/Z/17/Z).The fundingsources were not

70 E.Sovershaevaetal./InternationalJournalofInfectiousDiseases78(2019)65–71