PRESENTED AT THE 9TH IAS CONFERENCE ON HIV SCIENCE - PARIS, FRANCE

IsTenofovirUseinPregnancyAssociatedwithPretermDelivery?ACanadianPerinatalHIVSurveillanceProgramAnalysis

J. Brophy1,2, T. Lee3, A. Bitnun4,5, F. Kakkar6,7, J. Singer3, A. Alimenti8,9, D. Money8,9,10, W. Vaudry11,12, L. Samson1,2, L. Sauve8,9

for the Canadian Pediatric and Perinatal AIDS Research Group 1. Children's Hospital of Eastern Ontario, Ottawa; 2. University of Ottawa, Ottawa; 3. CIHR - Canadian HIV Trials Network, Vancouver;

4. The Hospital for Sick Children, Toronto; 5. University of Toronto, Toronto; 6. CHU Ste-Justine, Montréal; 7. Universite de Montréal, Montréal; 8. BC Women's Hospital and Health Centre, Vancouver; 9. University of British Columbia, Vancouver;

10. Women's Health Research Institute, Vancouver; 11. Stollery Children's Hospital, Edmonton; 12. University of Alberta, Edmonton

Background

■ Preferred cART regimen components for prevention of vertical HIV transmission have changed dramatically over time:

In 2008 DHHS:• 1st line: AZT, 3TC, LPV/r and NVP (CD4<250)• ABC, nelfinavir, saquinavir/r, or indinavir/r listed as alternative agents• TDF, FTC, ATV/r, DRV/r, RAL listed as “insufficient data to recommend”; EFV not

recommended

In 2017 DHHS:• 1st line: ABC/3TC or TDF/FTC are preferred backbones; ATV/r, DRV/r, or RAL are

preferred 3rd agents• ZDV/3TC, LPV/r, EFV, RPV listed as alternative agents• DTG, EVG, TAF, COBI listed as “insufficient data”

■ Perinatal ARV recommendations are generally not informed by clinical trials, but most often from cumulative clinical experience and passive reporting of suspected ARV-related toxicities to mom or infant via:• ARV pregnancy registries• Pharmaceutical post-marketing surveillance

■ The recently published PROMISE trial (Fowler et al, NEJM 2016), a RCT of different ART regimens in pregnancy found that tenofovir-based cART was associated with higher rates of preterm delivery and infant death.

Research Question

n What is the rate of preterm delivery among women treated with tenofovir-basedART versus other ART combinations in Canada in the CPHSP?

Methods

n Data from the CPHSP were reviewed for the period 1997–2015.

n Data evaluated included:• Maternal race/ethnicity and HIV acquisition risk category• Province of birth• Antiretroviral choice• Preterm (<37weeks) delivery

nMother-infant pairs were excluded in cases of:• multiparous births• women not on cART (eg. AZT monotherapy)• women starting cART >37wks

Canadian Perinatal HIV Surveillance Program (CPHSP)

• Active since 1990• Data collected annually from 22 pediatric & HIV care sites across Canada• Funded by Public Health Agency of Canada, and supported by Canadian HIV Trials

Network

n MV analysis was restricted to 2006-2015 when VL data was more complete. This found predictors of PTD to be:• tenofovir use (OR 1.92, p=0.0209)• zidovudine use (OR 2.03, p=0.0187)• IDU (OR 1.80, p=0.0086)• detectable VL (OR 1.62, p=0.0145)

• maternal race/ethnicity, province of birth, use of PI/NNRTI/INSTI, and ART start time were not predictive

• 94% of women received just 1 of the 3 main NRTIs (+ 3TC). The remainder received a combination of 2 or 3 (70/1614, 4%) or none (25/1614, 2%)

• By excluding those who received a combination of 2 or 3 NRTIs, TDF (OR 1.18, p=0.80) and AZT (OR 1.18, p=0.87) no longer predicted preterm delivery. Likewise, preterm delivery was more likely in women receiving combo (OR 1.87-2.54, p<0.05). IDU (OR 1.78, p=0.012) and detectable VL (OR 1.55, p=0.029) remained significant predictors

Vancouver

Edmonton

CalgarySaskatoon

Winnipeg

Ottawa

Toronto

HamiltonLondon

Windsor

SudburyMontréal (2 sites)

Québec city

WhitehorseIqaluit

St John’sHalifax

FrederictonCharlottetown

Yellowknife

Kingston

Preterm Delivery Chi- Squarep-value

TDF 19.4% NoTDF 15.2% 0.024ABC 15.9% NoABC 16.0% 0.950AZT 16.7% NoAZT 15.5% 0.439PI 15.9% NoPI 16.0% 0.931NNRTI 16.2% NoNNRTI 15.0% 0.578INSTI 16.0% No INSTI13.8% 0.573TDF+PI 19.3% TDF/noPI 19.5% 0.963TDF+NNRTI 18.0% TDF/noNNRTI 19.6% 0.733TDF+INSTI 17.8% TDF/noINSTI 19.3% 0.802

• Overall preterm delivery rate was 16%, with a higher rate in tenofovir-treated mothers (19.4% vs 15.2%, p=0.022).

• No differences were found between mothers treated vs not treated with: abacavir, zidovudine, protease inhibitors (PIs), non-nucleoside reverse transcriptase inhibitors (NNRTIs) or integrase inhibitors (INSTIs).

• No difference in preterm delivery rate among women exposed to tenofovirwith versus without PI, NNRTI, or INSTI

Univariate analysis

Results

n Among 2816 cART-treated mother-infant pairs (MIPs) from 1997-2015• 1732 used zidovudine (61.5%)• 575 used abacavir (20.4%)• 501 used tenofovir (15.1%)

nTenofovir use in pregnancy increased from 0.77% in 2004 to 54.1% in 2015;meanwhile, zidovudine use decreased from 100% in 1997 to 14.7% in 2015

Multivariate analysis

0

10

20

30

40

50

60

70

80

90

100

Tenofovir

Zidovudine

Abacavir

% ofpregnancies

Birth year

54.1%

14.7%

30.3%

Discussionn We found an increased rate of preterm delivery (16%) amongst HIV+ women inCanada relative to general population:• Preterm delivery rate in singleton pregnancies in Canada 2000-2013 = 6.2%n IDU and elevated VL were consistently associated with preterm delivery in MVanalysis, but TDF (and AZT) was no longer associated when combinations of multipleNRTIs excluded from analysisnARV components may play a role or may be confounded by other unmeasuredpredictors• Maternal age & comorbidities (DM, infection)• Smoking, nutrition, quality of carenCombinations of the 3 main NRTIs are suggestive of:• 3TC allergy/intolerance (unlikely) • ARV resistance (more likely)

nThe limitations of this study are similar to those of most observational/surveillancestudies:- Unclear role of missing data- Unmeasured covariates & confounders- Non-uniformity of care across geography and over time

Conclusions• CPHSP data re-confirms an increased prevalence of preterm delivery in women

living with HIV (2-3x general population in Canada).• while higher rates of preterm delivery were seen in TDF-treated women, TDF itself

does not seem likely to be causally linked.