Monitoring ART in resource-limited settings: option or necessity ?

Public Health Approach

Prof Charlie GilksUNAIDS Country Coordinator, India

5th IAS conference on Pathogenesis,

Cape Town, 21 July 2009

Outline

• The Public Health Approach and what it means in terms of lab monitoring

• Evidence of the impact of different ART monitoring strategies

• Options for resource-limited settings

“Three by Five”

Initiative started in 2003

The target: three million people

on treatment by the end of 2005

The goal : universal access to anti-retroviral therapy (ART) as

a human right to health

•

Public Health ART Strategy

Key elements for public sector ART

• First-line then second-line regimens

• Simple recommendations for when to start, toxicity substitutions & switch

• Tiered laboratory support for clinical decision-making

• Standard population-based HIVDR monitoring and surveillance

• Population-based Pharmacovigilance and toxicity monitoring

• Integrated and decentralised care with task shifting

• Chronic disease management

Harmonised ART Policy Guidance

Different guidelines for different populations and ART approaches

• Public sector ART• First then second line regimens • Limited number of ARVs used • Limited human resources• Limited laboratory services

• Physician/specialist-led ART • Initial regimen then multiple options• All ARVs available for use • Sophisticated labs to tailor regimen choice• Any detectable vl triggers regime change• Few cost constraints

Consider guidelines in their context

Core elements of ART monitoring

• ARV toxicity and SAEs• ART efficacy• HIV drug resistance

• Clinical monitoring • Laboratory monitoring

• Individual and/or population level

Evidence base on laboratory monitoring strategies in PHA

Clinical end-point trials including cost-effectiveness

• DART trial: preliminary data, IAS Cape TownART toxicity and efficacyClinically driven monitoring versus clinical + CD4 monitoring

• Modelling study: Phillips et al. Lancet 2008: 371 1443-51

• HBAC trial: presented but as yet unpublished dataART efficacy Clinical monitoring; clinical + CD4; clinical and CD4 and/or +

VL

• No end-point data on ART switch with detectable vl Targeted viral loads: July 2009 CID paper & editorial

IAS July 2009 10

0.0

0.2

0.4

0.6

0.8

1.0

Pro

port

ion

even

t-fr

ee

0 1 2 3 4 5

Years from randomisation (ART initiation)

LCM CDM

Grade 4 AEp=0.18

SAE p=0.20

ART-modifying AEp=0.85

DART routine toxicity monitoring

Grade 3/4 AEp=0.52

DART: first-line ART received

• Median follow-up to 31 December 2008 4.9 years (IQR 4.5-5.3)• 98% and 99% of expected nurse and doctor visits attended

0.0

0.2

0.4

0.6

0.8

1.0

Pro

port

ion o

f p

ati

ents

aliv

e o

n t

rial

0 1 2 3 4 5Years from randomisation (ART initiation)

LCMCDM

Second-line

Originalfirst-line

Substitutedfirst-line

IAS July 2009 12

Switch to second-line

Note: Adjusted for competing risk of death before switch to second-line

Proportion switched

to second-line

0.0

0.1

0.2

0.3

0.4

0.5

0 1 2 3 4 5

Years from randomisation (ART initiation)

CDMLCM

HR(CDM:LCM) = 0.84 (95% CI 0.72-0.98), p=0.03

HR(CDM:LCM) 0.58 0.48 0.77 0.90 1.35 1.10heterogeneity p=0.001

IAS July 2009 13

Survival (secondary endpoint)

0.90

0.87

0.92

0.90

0.95

0.94

1494 1445 1395 749CDM1656 1552 1501 1468 1436 796LCM

0 1 2 3 4 5

0.0

0.2

0.4

0.6

0.8

1.0

Pro

port

ion

aliv

e

1660 1542

HR(CDM:LCM) 1.08 2.05het p=0.004

(95% CI) (0.85-1.39) (1.43-2.93)

Years from randomisation (ART initiation)

HR(CDM:LCM) = 1.35 (1.10-1.65) p=0.004

LCMCDM

Number needed to monitor = 130

IAS July 2009 14

Sensitivity Analysis: Sensitivity Analysis: Minimal MonitoringMinimal Monitoring

LCMN = 1656

CDMN = 1660

Difference(LCM – CDM)

Overall Mean Total Cost US$ 2008Adjusted for censoring, discounted at 3%[95% Confidence Interval]*

$2599 $2382 $217[$95 , $334]

Overall survival days** Discounted at 3%[95% Confidence Interval]*

1863 1826 +37[-10 , 83]

Incremental Cost Effectiveness RatioAdjusted for censoring, discounted at 3%[95% Confidence Interval]*

$2146[$721 , Dominated]

Modifications from Adjusted and Discounted Costs and Benefits: • 12-weekly CD4 cell count routinely performed after the 1st year on ART• No routine (12-weekly) Haematology and Biochemistry tests

* 95% CI estimated with bootstrapping percentile method.** Estimated through the area under the Kaplan-Meier survival curve, with censoring applied at the longest observed time of the arm whose maximum observed time occurs first.

Although survival was slightly longer with viral load monitoring, this strategy was not the most cost-effective.

The benefits of Vl or CD4 over clinical monitoring are modest. Development of cheap and robust assays is important; meanwhile widening access to ARVs is the highest priority

Interim HBAC Conclusions

• Adding CD4 to clinical monitoring ($831 - $838 per DALY averted) is about as cost-effective as putting another person on ART in Tororo ($600 per DALY).

• Adding viral load to CD4/clinical monitoring has a cost per DALY averted ($3,600 - $11,900) that is 4 to 20 times higher.

• HBAC analysis suggests that CD4 monitoring or starting a patient on ART are economicallypreferable to viral load monitoring …

IAS July 2009 17

Sensitivity Analysis: Sensitivity Analysis: CD4 count costsCD4 count costs

• At current costs ($7.06 - $8.82), CD4 testing is not cost effective

• We sought to establish the cost per test at which CD4 monitoring would be cost effective

(ICER of $1200 ~3 times GDP per capita; WHO Commission on Macroeconomics and Health)

CD4 count would have to cost $3.8 or less for ART management with 12-weekly CD4 monitoring after 1st year to be cost effective

Targeted viral loads at failure• 3 failure domains which are not the same:

– clinical; immunological and virological • 20% of clinical failures had high CD4 in DART• 15 – 40% switches with clinical/immunological failure unnecessary:

viral suppression or low-level replication

• Targeted viral load testing as “tie-break” to conserve use of second-line and reduce costs: policy in India

• Caveat: are “failing” patients not benefiting from early switch?

• Does not mean that routine viral load monitoring is a necessary or will be cost-effective in resource-limited settings in public sector:– What threshold for viral failure to trigger switch?

– Maximal suppression likely to be far too early

IAS July 2009 19

DART Survival using PHA

0.90

0.87

0.08

0.92

0.90

0.18

0.95

0.94

0.55

0 1 2 3 4 5

0.0

0.2

0.4

0.6

0.8

1.0

Pro

port

ion a

live

Years from randomisation (ART initiation)

LCMCDM

Entebbe Cohort:pre-ART, median CD4 75 at start

What more could VL monitoring add, and at what cost?

CTXp – 50% reduction first 72 weeks of ART (MOPEB020)

HIV drug resistance: population-based monitoring and surveillance

No scope in PHA for different first-line ART according to baseline resistance pattern

Cohort DR monitoring for programme effectiveness

Population DR monitoring for extent of transmitted HIV DR

Articles reporting results from HIVDR transmission surveys in 7 countries; all had <5% DR in incident cases

No need to change ARVs provided in public sector for first-line ART

Summary and conclusions

• PHA is an extremely effective tool for ART scale-up and delivery of effective ART in resource-constrained settings

• (Quality) clinically-driven monitoring can deliver excellent outcomes for the individual

• Small outcome benefit from routine CD4 monitoring

• Likely only small additional outcome benefit from routine VL in addition to CD4 monitoring

• Neither laboratory-based strategies are cost-effective; getting people in need on to ART remains the priority

• Drug resistance monitoring and SAE/toxicity monitoring are best done at a population level to inform PHA ARV choices

Future directions

• Drive for Quality clinical monitoring• CD4 testing for eligibility to start/thresholds• Targeted CD4 ART monitoring with much

cheaper and ideally POC tests • VL testing for Early Infant Diagnosis• Targeted vl as cost-saving tie-break for patients

with clinical or immunological failure• Drug resistance and pharmacovigilance at

population level or in cohort studies

THANK YOU