Arevir Meeting, Bonn, April 23, 2009 M. Zazzi on behalf of the - - PowerPoint PPT Presentation

▶
arevir meeting bonn april 23 2009 m zazzi on behalf of
SMART_READER_LITE
LIVE PREVIEW

Arevir Meeting, Bonn, April 23, 2009 M. Zazzi on behalf of the - - PowerPoint PPT Presentation

Oct 12, 2022 366 likes •760 views

Arevir Meeting, Bonn, April 23, 2009 M. Zazzi on behalf of the EuResist Network (www.eurestist.org) EuResist status Funded by the EU JAN-06 to JUN-08, then set as a European Network (legal entity) Data collected from ~30,000 patients

slide-1
SLIDE 1

Arevir Meeting, Bonn, April 23, 2009

  • M. Zazzi on behalf of the EuResist Network

(www.eurestist.org)

slide-2
SLIDE 2

EuResist status

Funded by the EU JAN-06 to JUN-08, then

set as a European Network (legal entity)

Data collected from ~30,000 patients (Italy,

Germany, Sweden, Luxembourg, Belgium, Spain)

Data modeling by IBM Israel, Max Planck

Institute for Informatics, Informa & Rome TRE University

Several methods investigated, much effort

  • n feature selection and derivation

DB still expanding, models being updated

and refined

slide-3
SLIDE 3

EuResist – TCE definition

time Genotype Treatment switch Viral load 0 to 12 weeks Short-term model: 4-12 weeks Viral load Pre-therapy HIV RNA CD4 Patient demographics (age, gender, race, route of infection) Past genotypes Past treatments Past AIDS diagnosis

slide-4
SLIDE 4

EuResist – labeling therapies

Baseline data HIV genotype at 0 to 12 weeks before treatment VL at 0 to 12 weeks before treatment Additional variables when available Treatment switch VL at 4 to 12 weeks (8-week outcome)

SUCCESS Undetectable or >2 log decrease VL FAILURE Detectable and not >2 log decrease VL

slide-5
SLIDE 5

The data funnel…

slide-6
SLIDE 6

EuResist – engines

Three prediction engines developed

independently

Generative-Discriminative (by IBM) Evolutionary (by Max-Planck Institute) Mixed effects (by Rome TRE & Informa)

Then, combined into a unique engine

and made freely available on the web

slide-7
SLIDE 7

Data-driven systems –EuResist Generative/Discriminative engine

Model response to treatment in the absence of genotype with a Bayesian network For any defined regimen, compute a probability of success (Generative step) Use the probability as an additional feature for logistic regression together with genotype and

  • ther covariates (Discriminative step)
slide-8
SLIDE 8

Data-driven systems –EuResist Evolutionary engine

Model HIV evolution under therapy from longitudinal and cross-sectional sequence data For any defined genotype, neighbor mutants can be computed in silico and the contribution of the expected mutants to resistance can be calculated Functions weight for probability and expected time for mutants to occur Probability to remain susceptible to a drug (below a defined phenotypic threshold) GENETIC BARRIER

Altmann et al, AVT 2007

slide-9
SLIDE 9

Data-driven systems –EuResist Mixed-Effects engine

  • Focuses on interactions among variables
  • drug x drug
  • drug x drug x drug
  • drug x mutation
  • drug x previous drug class exposure
  • drug x previous drug exposure
  • mutation x mutation
slide-10
SLIDE 10

Data-driven systems –EuResist Combined engine

The combination

(mean) of the engines performs equal to or better than the individual engines

The combined

engine learns faster, i. e. it is more accurate when trained on limited data sets

Altmann et al, PLoS ONE 2008

slide-11
SLIDE 11

The EuResist combined engine

3143 therapies, Short-term outcome (8 weeks)

slide-12
SLIDE 12

EuResist

  • vs. Expert interpretation

(EVE study)

Form the invitation letter: The requested response include a categorical (C) answer and a quantitative (Q) estimate: C) Given this HIV genotype and patient information, will the indicated therapy be successful (i. e. will it make HIV RNA decrease by at least 2 logs or to undetectable levels in 8 weeks) ? Q) Given this HIV genotype and patient information, what probability

  • f success would you estimate for the indicated therapy?
slide-13
SLIDE 13

EuResist

  • vs. Expert interpretation

(EVE study)

slide-14
SLIDE 14

Distribution of the 25 therapies by year and type

5 4 3 2 1 1998 1999 2000 2001 2002 2003 2004 2005 2006 Year of therapy

  • No. of

cases NRTI-only therapy NNRTI-based therapy *PI-based therapy

*17 cases with boosted PI, 2 ATV, 1 NFV

slide-15
SLIDE 15

Patient characteristics

FEATURE MEDIAN (IQR) Baseline log viral load 4.67 (4.38-4.99) Baseline CD4 counts 298 (134-412) Number of previous treatment lines 5 (3-6) Number of NRTI mutations at baseline 3 (3-4) Number of NNRTI mutations at baseline 1 (0-2) Number of PI mutations at baseline 2 (0-3) Number of available previous viral load data 15 (8-25) Number of available previous CD4 counts 14 (10-30) Number of available previous genotypes 1 (0-3)

slide-16
SLIDE 16

EuResist vs. Expert interpretation

(EVE study)

25 HAART cases randomly selected form the EuResist db:

  • Obsolete therapies excluded
  • Wild type genotype excluded
  • All clinical and virological

information available 12 experts enrolled, response

  • btained from 10:
  • On‐line anonymous rating
  • Only European (E) vs. non‐

European (N) setting traceable

  • Use of any interpretation

system allowed (and declared)

slide-17
SLIDE 17

AUC 95% CI Best_expert 0.853 0.655 - 0.961 euresist 0.787 0.578 - 0.923 mean_expert 0.777 0.567 - 0.917 Worst_expert 0.653 0.438 - 0.830

P = 0.011

slide-18
SLIDE 18

Correlation between EuResist and mean(expert) probability of success

EuResist vs. Expert interpretation

(EVE study)

slide-19
SLIDE 19

xxx

20 40 60 80 100 50 40 30 20 10

  • 10
  • 20
  • 30

AVERAGE of EuResist and mean(expert) EuResist - mean(expert) Mean 7.0

  • 1.96 SD
  • 25.0

+1.96 SD 38.9

EuResist

  • vs. Expert interpretation

(EVE study)

slide-20
SLIDE 20
  • No. of errors vs. no. of

interpretation systems used

slide-21
SLIDE 21

How confident are you in your prediction?

Average absolute difference between the predicted probability of success and the cut-off value (50%)

slide-22
SLIDE 22

EuResist

  • vs. Expert interpretation

(EVE study)

slide-23
SLIDE 23

EuResist

  • vs. Expert interpretation

(EVE study)

EXPERT 1 EXPERT 3 EXPERT 4 EXPERT 5 EXPERT 6 EXPERT 7 EXPERT 8 EXPERT 9 EXPERT 10 EXPERT 11

EuResist

ACTUAL OUTCOME

F F S F F S F F F F F F S S S S S S S S S F F S F F F F S S F F F F S

Incorrect prediction Treatment success

S

Treatment failure

F

EuResist & most experts incorrect

slide-24
SLIDE 24

EuResist

  • vs. Expert interpretation

(EVE study)

EXPERT 1 EXPERT 3 EXPERT 4 EXPERT 5 EXPERT 6 EXPERT 7 EXPERT 8 EXPERT 9 EXPERT 10 EXPERT 11

EuResist

ACTUAL OUTCOME

F F S F F S F F F F F F S S S S S S S S S F F S F F F F S S F F F F S

Incorrect prediction Treatment success

S

Treatment failure

F

Unexpected drug efficacy

slide-25
SLIDE 25

Past treatment lines

AZT DDI AZT DDC DDC SQV 3TC D4T IDV D4T EFV RTV

Nadir CD4: unknown Zenith VL: 72,300

Case #12843 (patient 17363) (27/07/2001: ABC, D4T, LPV/r )

1 a

slide-26
SLIDE 26

Case #12843 (patient 17363) (27/07/2001: ABC, D4T, LPV/r )

Genotype

L10I M36I G48V I54V L63P A71V T74S

V77I V82A L90M I93L

D67N T69D K70R K103N V118I G190A

T215C K219Q G333E

Past drug resistance mutations

unknown

Baseline VL: 72,300 Follow-up VL: 314 (-2.36 log)

1 b

slide-27
SLIDE 27

Case #12843 (patient 17363) (27/07/2001: ABC, D4T, LPV/r )

Treatment more effective than expected

T215C revertant?

Transient success?

Patient lost to follow-up

Definition of success

1 c

slide-28
SLIDE 28

Past treatment lines

AZT DDC 3TC AZT 3TC D4T IDV DDI NVP SQV/rtv D4T DDI LPV/rtv

Nadir CD4: 8 Zenith VL: 794,328

Case #14503 (patient 19816) (05/10/2001: D4T, EFV, LPV/r )

2 a

slide-29
SLIDE 29

Genotype

L10I G48V I54V Q58E L63P A71V V77I

V82C I84V

M41L D67N L74V K101N V118I Y181C

L210W T215C K219E

Past drug resistance mutations

unknown

Baseline VL: 794,328 Follow-up VL: 1,000 (-2.90 log)

Case #14503 (patient 19816) (05/10/2001: D4T, EFV, LPV/r )

2 b

slide-30
SLIDE 30

Treatment more effective than expected

T215C revertant? V82C not resistant to LPV?

Transient success?

EFV with Y181C? Later VL rebound to 15,900

Definition of success

Case #14503 (patient 19816) (05/10/2001: D4T, EFV, LPV/r )

2 c

slide-31
SLIDE 31

EuResist

  • vs. Expert interpretation

(EVE study)

EXPERT 1 EXPERT 3 EXPERT 4 EXPERT 5 EXPERT 6 EXPERT 7 EXPERT 8 EXPERT 9 EXPERT 10 EXPERT 11

EuResist

ACTUAL OUTCOME

F F S F F S F F F F F F S S S S S S S S S F F S F F F F S S F F F F S

Incorrect prediction Treatment success

S

Treatment failure

F

Adherence issues?

slide-32
SLIDE 32

Past treatment lines

AZT AZT DDC 3TC AZT IDV 3TC AZT NVP

Nadir CD4: 289 Zenith VL: 18,000

Case #25745 (patient 9492) (14/04/2005: 3TC, TDF, ATV/r )

3 a

slide-33
SLIDE 33

Genotype

L63P I93L M41L E44A D67G L74I V118I V179I Y181I

M184V L210W T215Y K219D

Past drug resistance mutations

L10I A71V I84V Y181C

Baseline VL: 18,000 Follow-up VL: 21,000

Case #25745 (patient 9492) (14/04/2005: 3TC, TDF, ATV/r )

3 b

slide-34
SLIDE 34

Poor adherence?

But 3 years with undetectable VL in the past

Underestimated resistance?

Impact of past I84V on ATV L74V as a proxy of hidden K65R impacting

TDF

Case #25745 (patient 9492) (14/04/2005: 3TC, TDF, ATV/r )

3 c

slide-35
SLIDE 35

Past treatment lines

ABC DDI EFV 3TC TDF EFV LPV/r TDF EFV NFV AZT EFV NFV 3TC TDF EFV

Nadir CD4: 11 Zenith VL: 500,000

Case #43708 (patient 8477) (22/04/2005: AZT, EFV, ATV/r )

4 a

slide-36
SLIDE 36

Genotype

M36I T74S I93L K65R L74V V90I Y115F M184V G190Q

K219N

Past drug resistance mutations

Same as at last time point

Baseline VL: 114,370 Follow-up VL: 3,816 (-1.48 log)

But earlier 97 (-3.07 log)

Case #43708 (patient 8477) (22/04/2005: AZT, EFV, ATV/r )

4 b

slide-37
SLIDE 37

Limited adherence?

Three AZT hypersusceptibility mutations

(K65R L74V M184V)

Transient response

Underestimated resistance?

G190Q impact on EFV?

Case #43708 (patient 8477) (22/04/2005: AZT, EFV, ATV/r )

4 c

slide-38
SLIDE 38

Why do we err?

Limitations in the definition of success

“create” prediction errors

Expected adherence cannot be accounted

for based on presently available training data set

Genotype shortcomings Impact of past mutations Short-term drug activity on partially resistant

strains (e. g. 215 revertants, Y181C with EFV)

Unweighted factors (host genetics)