[NATAP] Reyataz Predicting Outcome: PK, Muations, Pheno Cutoff

[Guest guest]

NATAP - http://www.natap.org

Reyataz parameters predicting outcomes

Reported by Jules Levin

There were 3 oral presentations back-to-back at the 14th HIV Drug Resistance Wksp (June 7-11, 2005, Quebec City, Canada) which reported study results analyzing predicting outcomes to atazanavir and atazanavir/rtv therapy in treatment-experiened patients (studies 043 and 045) based on Genotypic Inhibitory Quotient, drug levels (PK), and the types of resistance mutations (Mutation Score); as well, the first study below reports on their determination of clinical phenotypic cutoffs to ATV & ATV/rtv. The second study below reports on the Mutation Score they found that predicted response to atazanavir/rtv.

“Determination of phenotypic clinical cutoffs for atazanavir and atazanavir/ritonavir from AI424-043 and AI424-045â€

Study authors: EP Coakley1, C Chappey1, JF Maa2, S Wang2, M Bates2, V Wirtz2 and D Seekins2

1ViroLogic, Inc, South San Francisco, CA, USA

2Bristol-Myers Squibb Virology, Plainsboro, NJ, USA

……. Optimum responses to ATV were observed at FCs <2.2 in the 043 cohort and to ATV/r at FCs <5.2 in the 045 cohort: in study 043 (atazanavir), the proportions with HIV RNA <400 copies/mL for those with baseline ATV FC <2.2 and ≥2.2 were 76% and 45%, respectively. In study 045 (atazanavir/ritonavir), proportions with HIV RNA <400 copies/mL for those with baseline ATV FC <5.2 and ≥5.2 were 77% and 12%, respectively…..

BMS 043--In the Virologic database the updated ATV biologic cutpoint utilizing the 99th percentile for the distribution of 9,764 samples without recognized protease mutations, is at 2.1

--Therefore, the proposed cutoff for unboosted ATV at 2.2 is proximate to, but above, the biologic cutoff.

CAVEATS

These various models did not control for the impact of background therapy. The 045 study included the use of TDF which would be anticipated to reduce boosted ATV exposure by about 25% (area under the curve). Although all subjects in 045 study had a history of failure of >=1 PI only 36% were on a PI at the time of screen. These various factors may have served to lower the clinical cutoff for ATV/r (and to a lesser extent ATV) and these cutoffs may be considered conservative estimates. Differences in clinical cutoffs between drugs do not imply differences in potency.

Atazanavir (ATV) (400 mg daily) and ATV/r (ATV 300 mg daily with ritonavir 100 mg daily) are widely used in combination treatment of HIV-1. The PhenoSense assay currently utilizes a biologic cutoff for ATV at 2.3-fold change (FC). The current study seeks to define a phenotypic clinical cutoff for ATV and ATV/r, above which virological response to therapy begins to decrease.

METHODS: Data from two clinical trials in subjects with prior PI failure, BMS AI424-043 (ATV+2NRTIs) and AI424-045 (ATV/r+TDF+1NRTI), were included for all subjects having baseline phenotype, baseline viral load ≥400 copies/mL, and viral load at weeks 12 and/or 24 (ATV n=131 and ATV/r n=111). ATV phenotyping (PhenoSense) was performed at baseline and correlated with virological outcome within each study.

Efficacy outcomes focused on change in viral load from baseline and the proportion of subjects with HIV RNA levels <400 copies/mL at week 24. Logistic regression and Fischer's exact tests were used to identify the ATV and ATV/r FC cutoffs.

STATISTICAL METHODS

The variables;

--baseline ATV fold-change (log10 FC)

--HIV-RNA at baseline & follow-up to week 12 & week 24

--an HIV-RNA <400 copies/ml or <50 copies/ml at follow-up.

The endpoints:

--the HIV RNA change at follow-up (continuous endpoints)

--achieving an HIV-RNA <400 or <50 copies/ml at follow-up (binary endpoints)

Fisher's exact test was used to derive the cutoffs for ATV & ATV/r

--the categories were FC <cutoff & FC>=cutoff

the Boneferroni adjustment for multiple comparisons was used (p-value <0.0017)

--Exploratory analysis were performed to further investigate the relationship between the virologic outcomes & the ATV phenotype:

linear regression

recurative partitioning

logistic regression

BMS A1424-045

Subjects had virologic failure of >= regimens:

--the median prior PI exposure was 2.6 yrs with 36% of subjects being on a PI at screening

--the median number of PI mutations was 2 with 44% having >=3 mutations.

(ed note from Jules Levin: considering that a significant proportion of patients entering 045 had been on nelfinavir in the past but were not on NFV immediately before switching to ATV/r in this study, PI resistance may not have been readily recognizable or detectable.

Randomization was to:

--ATV/r or LPV/r + (TDF+ 1 NRTI)

To week 48 the proportions of subjects with HIV-1 RNA <400 copies/ml (ITT) were:

64% in the ATV/r arm (n=120)

62% in the LPV/r (n=123)

VARIABLES & OUTCOMES for the BMS A1424-045 DATASET (n=111)

BASELINE N MEAN MEDIAN

HIV RNA (log c/ml) 111 4.41 4.48

ATV fold-change 111 4.14 1.18

FOLLOW-UP N MEAN MEDIAN

Change in HIV RNA 109 -1.88 -2.04

To wk 24

(log c/ml)

N %Responders

HIV RNA <400 c/ml at wk 12 108 63%

HIV RNA <400 at wk 24 109 67%

RESULTS IN PROGRAM ABSTRACT:

For the 043 and 045 studies, the median baseline HIV-1 RNA value was 15,477 and 29,898 copies/mL, respectively, and the median (range) baseline FC to ATV was 1.2 (0.4-90) and 1.2 (0.2-57), respectively. (in program abstract)

In 043, the proportion of subjects with HIV RNA <400 copies/mL at week 24 was relatively constant below 2.2-fold but was reduced at higher FCs (P=0.001). The proportions with HIV RNA <400 copies/mL for those with baseline ATV FC <2.2 and ≥2.2 were 76% and 45%, respectively.

In 045, the proportion of subjects with HIV RNA <400 copies/mL at week 24 was relatively constant below 5.2-fold but was reduced at higher FCs (P<0.0001). The proportions with HIV RNA <400 copies/mL for those with baseline ATV FC <5.2 and ≥5.2 were 77% and 12%, respectively.

ORAL PRESENTATION RESULTS:

FISCHER's EXACT TEST for the BMS A1424-045 DATASET

CUTOFF <50 c/ml <50 c/ml <400 c/ml <400 c/ml

at wk 12 at wk 24 at wk 12 at wk 24

3.1 0.3643 0.0064 0.0015* 0.0001*

3.5 0.4797 0.017 0.0072 0.0008*

4.1 0.3438 0.0076 0.004 0.0004*

4.6 0.0207 0.0005* 0.0002* <0.0001**

5.1 0.0401 0.0002* <0.0001** <0.0001**

5.2 0.0401 0.0002* <0.0001% <0.0001**

6,1 0.1001 0.0013* 0.0004* <0.0001**

**p-values <0.0001; *p-values ranging >0.001 but <0.0017

BMS A1424-045 RECURSIVE PARTITIONING with CLASSIFICATION & REGRESSION TREES (CART)

HIV RNA RESPONSE N RESPONSE CUTOFF MISCLASSIFY MISCLASSIFY

N(%) <cutoff, % >cutoff, %

<50 c/ml at wk 12 108 33 (30.6%) 12.5 33% 0%

<50 c/ml at wk 24 109 50 (45.9% 4.9 46% 6%

<400 c/ml at wk 12 108 68 (63.0%) 5.1 28% 17%

<400 c/ml at wk 24 109 73 (67.0%) 5.1 23% 12%

BMS A1424-045

LOGISTIC REGRESSION

HIV RNA ENDPOINT UPPER 95% CI p-value CORRECT PREDICTION

Log ATV FC (FC) Of The Model

<50 c/ml at wk 12 0.78 (5.8) 0.01 68.8%

<50 c/ml at wk 24 0.57 (3.7) 0.002 64.5%

<400 c/ml at wk 12 0.42 (2.6) 0.0003 75.8%

<400 c/ml at wk 24 0.38 (2.4) 0.0002 75.2%

SUMMARY of RESULTS - ATV/r in BMS 045

The Fischer's test identified a cutoff at 5.2 as the lowest FC with the lowest p-values across the greater number of endpoints tested:

--the CART analysis identified cutoffs in the range 4.9-12.5

--the logistic regression analyses identified cutoffs in the range 2.4-5.8

By baseline ATV FC <5.2 and >=5.2 the proportions with HIV RNA <400 c/ml at week 24 were 77% and 12%, respectively.

BMS A1424-043

Subjects had virologic failure of a PI-based regimen (median prior PI use: 2.6 yrs):

--the proportion with >=4 PI mutations ranged from 23%-26%

Randomization was to:

--ATV 400mg QD + 2 NRTIs (n=150)

--LPV/r 400/100mg BID + 2 NRTIs (n=150)

To week 24 the proportions with HIV RNA <400 c/ml (ITT) were:

--59% in the ATV arm (n=150)

--77% in the LPV/r arm (n=150) p<.05)

RESULTS

ATV CUTPOINT in BMS A1424-043

Among the 131 subjects the mean (median) baseline ATV FC was 2.9 (1.2)

--66% had HIV RNA <400 at week 24

The Fishers test identified a cutpoint of 2.2

--the CART analyses identified cutoffs in the range 1.5-5.0

--the logistic regression identified cutoffs in the range 1.4-7.1

“Clinical validation of atazanavir/ritonavir genotypic resistance score in PI-experienced patientsâ€

authors concluded:

….most PI-experienced patients who switched to ATV/rtv containing HAART had a viral response..,.13 mutations were identified to decrease the efficacy of ATV/rtv…7 mutations were retained in an ATV/rtv resistance score (10, 16, 33, 46, 60, 84, 85) that predicts reduced response in patients with >=3 mutations among this set….the number of active drugs was also associated with viral response…

The authors added:

some of these 7 mutations (M46I, I84V) were previously described to be selected in vitro in PI naïve viruses (NL4-3; RF) under selective pressure of ATV (Gong Y et al, AAC 2000).

Some of these mutations were not previously described to reduce the efficacy of PIs (16, 60, 85), but are linked to PI exposure in the Stanford database.

study authors: S Vora1, A-G Marcelin6, H Günthard2, P Flandre7, HH Hirsch3, P Vernazza4, H Furrer5, B Masquelier 8, A Zinkernagel2, G Peytavin6, V Calvez6, B Hirschel1,

L Perrin1, S Yerly1, and the Swiss HIV Cohort Study (SHCS)

1Geneva, University Hospital, Switzerland

2Zurich, University Hospital, Switzerland

3Basel, University Hospital, Switzerland

4St-Gallen, University Hospital, Switzerland

5Bern, University Hospital, Switzerland

6Pitié-Salpêtrière Hospital, Paris, France

7INSERM Unit 472, Paris, France

8CHU Bordeaux, France

BACKGROUND: Virological response to ATV/RTV containing HAART in PI-experienced patients was used to develop a clinically relevant genotypic resistance score.

METHODS: We included 62 PI-experienced patients (>6 months) switched to ATV/RTV (300 mg ATV, 100 mg RTV, plus other ARVs) with HIV-1 RNA >1000 copies/ml at baseline and detectable ATV in plasma at month three (M3). The impact of baseline protease mutations on virological response (>1 log RNA decrease) at M3 was analysed using Fischer's exact test. Mutations with prevalence >8% and P<20% were retained. Cochran- Armitage's test was used to select the combination of mutations most strongly associated with the virological response (lowest P value). Robustness of the score was investigated using bootstrap re-sampling.

RESULTS:

BASELINE CHARACTERISTICS

N=62

77% male

risk factors: 44% MSM; 30% HT; 22% IDU

subtype B: 87%

reasons for switch: 84% viral failure; 16% drug intolerance

Median baseline viraemia and CD4 count were 4.3 log10 copies/ml and 226/mm3, respectively. Patients had been treated with a median of 7.5 (2-15) ARV drugs including two PIs (1-6); 4 NRTIs (1.7); 79% were treated with NNRTIs. At M3, 82% of patients had a virological response and 56% had RNA<50 copies/ml. ATV/r co-treatment- RTIs+TDF: 69%; RTIs+NNRTI: 5%; RTIs+T20: 5%.

MUTATIONS ASSOCIATED with REDUCED VIRAL RESPONSE (<1 log)

82% of patients harbored at least 1 log of HIV RNA decrease at M3. 13 mutations were associated with a decrease of viral response.

The PI mutations associated with a reduced response were:

MUTATION p-value

10F/I/V (<0.0001)

16E (0.035)

33I/F/V (0.004)

46I/L (0.001)

54L/V/M/T (0.004)

60E (0.14)

62V (0.16)

71I/T/V/L (0.034)

82A/T (0.20)

84V (0.035)

85V (0.035)

90M (0.004)

93L (0.15)

The strongest association with the virological response was found with the combination of 10F/I/V, 16E, 33I/F/V, 46I/L, 60E, 84V and 85V (P=8.0x10-9). Using this ATV resistance score, virological response was observed in 100% (n=25), 100% (n=17), 80% (n=5), 42% (n=12), and 0% (n=3) of patients with 0, 1, 2, 3, ≥4 mutations, respectively.

ATV/RTV RESISTANCE SCORE; Colonno et al

(10F/I/V, 20M/R, 24I, 36I/L/V, 46I/L, 48V, 54L/V, 63P, 71I/T/V, 73A/C/S/T, 82A/F/T/S, 84V, 90M)

100% response with 0 (n=6), 1 (n=18), 2 (n=9), & 3 (n=4) mutations; 80% response with 4 mutations (n=5). With 5 mutations 20% response (n=4).

FACTORS ASSOCIATED WITH VIRAL RESPONSE

FACTORS P

Baseline HIV RNA

(log copies/ml) 0.42

baseline CD4 0.37

# of active drugs 0.0012

# of previous PIs 0.04

PI resistant (n) <0.001

RESPONSE ACCORDING to ACTIVE DRUGS in PATIENTS with ATV/rtv RESISTANCE SCORE >=3 (nonresponders)

Patients with:

0 active drugs (0 response, n=3)

1 active drug (29% response, n=7)

2 or 3 active drugs (60% response, n=5)

p=0.038

Addition of the 90M to the previous score provides also a strong association with the virological response (P=8.7x10-9). The bootstrap analysis showed the

robustness of both scores.

“Virological response to atazanavir/ritonavir-based regimens: resistance mutations score and pharmacokinetic parameters (Cmin, Cmax, AUC)

(Reyaphar study)â€

author conclusions:

--switch to ATV/r regimen led to viral success at month 6 in 64% (<50 copies/ml) and 76% (<400 c/ml) of patients

--ATV MUTATION SCORE: <6 vs >6 mutations predicted viral failure at month 6 with 35% sensitivity and 95% specificity. In this study: L10F/I/V, K20I/M/R, L24I, L33F/I/V, F53L, I54L/V, G73A/C/S/T, V77I, I84V, L90M were associated with viral failure (>50 c/ml)

--Cmin was significantly associated with viral failure only when the baseline ATV/r mutation score was <6 (Cmin <0.2ug/ml: 47% vs Cmin >0.7ug/ml: 80% VS)

--GIQ for the whole population, predicted VF better than PK alone (<0.33 vs >0.33: VF at month 6 with 91% sensitivity and 36% specificity)

study authors: I Pellegrin1, M Vray2, D Neau3, HJA Fleury1, MH Schrive1, M-C Saux4, J-L Pellegrin3, E Lazaro3, J-M Ragnaud3, D Breilh4

1Department of Virology

2Institut Pasteur Emerging Diseases Epidemiology Unit

3Internal Medicine and Infectious Diseases,

4Pharmacy, Bordeaux University Hospital, Bordeaux, and Paris, France

The aim of this study was to assess the impact of HIV-1 mutations, individual PK parameters (Cmin, Cmax, AUC) and genotypic inhibitory quotient

(GIQ) on virological responses to atazanavir/ritonavir based regimens after 24 weeks in experienced HIV patients with multiple viral failures.

METHODS:

Antiretroviral-experienced subjects with virological failure (HIV-RNA>1.7 log10 copies/ml) were included in an observational cohort study and received an atazanavir/ritonavir (300 mg/100 mg), once daily- based HAART. Sequencing of HIV_reverse-transcriptase and protease was performed at baseline. Virological success (VR) was defined by HIV_RNA <50 (1.7 log) at month 6, <400 (<2.6 log) at month 6, & delta HIV RNA >1 log between M0 & M6.

Atazanavir Cmin, Cmax and AUC were determined at steady-state

using a population PK analysis.

GIQ was calculated=Cmin/number of atazanavir/ritonavir resistance related

mutations among:

L10F/I/V,K20I/M/ R,L24I,L33F/I/V,M36I/L/V,M46I/L,G48V,I54L/V,L63

P,A71I/V/T,G73A/C/S/T,V82A/F/T/S,I84V,L90M

(French ANRS genotype-interpretation guidelines).

Resistance was defined as >6 mutations.

Monitoring was Mo, M1, M3, M6 and then at 3 month intervals.

RESULTS:

Atazanavir/ritonavir-containing regimen was initiated in 90 patients (M/F=63/27, median age=45, in association with tenofovir+ lamivudine or emtricitabine in 60% of them; or, co-prescribed with ATVr: + 2 NRTIs, 23%; + 1 NNRTI, 2%; 1 patient only (LPV, SQV or FAPV, 15%)

TREATMENT HISTORY

--9 yrs duration

--PI experienced, 91%

--NNRTI=62% (80% of whom with NNRTI-related mutations)

--5 [2-7] previous treatment

--At baseline, median CD4+ and HIV-1_RNA were 273 [196-473]/µl and 3.9 [2.7-4.7] log10 copies/ml

--CD4 nadir: 152 (71-250)

--switch from LPV/r to ATV/r: 47%

CLINICAL:

--CDC A/B/C: 45/11/25

VIRAL EFFICACY at 24 WEEKS

Viral load decline of -1.2 log (-2.8, -0.3)

CD4 increase: +51 (-8, +174)

64% <50 copies/ml

76% <400 c/ml

delta RNA >1 log: 56%

Median number of atazanavir/ritonavir resistance-related mutations

was 2 [1-4] and three [0-4] NRTI-related mutations.

Wnen patients had 6 or more mutations from Mutation Score, the percent <50, <400, or VR>1 log was much reduced, and the the best responses were when there were 0-2 mutations at baseline (almost 80% <50 & 90% <400).

When there were 0 to 2 or less PI mutations at baseline delta HIV RNA decline was -1.8 log. When there were 3 to 6 or less mutations the delta HIV RNA response was -1 log. When there were 6 or more mutations, the viral load response was +0.04 log.

SCORE (<6 vs >=6)

Sensitivity 35%

Specificity 95%

PPV 80%

NPV 71%

RESPONSE to ATV/r REGIMEN by BASELINE GENOTYPE

In Univariate analysis PR mutations (among IAS list) associated with viral failure (VF) at Wk 24:

VF (VL >50 c/ml)

P<0.05:

L10F/I/V

L24I

G73A/C/S/T

I84V

L90M

P (I think it said) between 0.05 & 0.2

K20I/M/R

L33F/I/V

M46I/L

F53L

I54L/V

V77I

VF (VL>400 c/ml):

P<0.05:

L10F/I/V

L24I

L33F/I/V

M46I/L

I62V

L90M

P (I think it said) between 0.05 and 0.2

K20I/M/R

G73A/C/S/T

V77I

V82A/F/T/S

I84V

BMS/ANRS ATV/r mutations score: 10, 20, 24, 33, 36, 46, (48), 54, 63, 71, 73, 82, 84, 90

62 & 77 were significantly associated with VF in the model.

36, 63, & 71 was no longer selected in this populationlike 48 (but its frequency was <5%).

RESPONSE to ATV/rtv BASED REGIMEN ASSESSED by PK

Cmin (ug/ml) (median [Q1; Q3]): 0.3 [0.1; 0.6]

Cmax (ug/ml) (median [Q1; Q3]): 6 [4.3; 7.2]

AUC (mg.h/l) (median [Q1; Q3]): 80 [64; 95]

Response to ATV/r at wk 24 assessed by Cmin:

P=0.16 (NS)

47% response when <0.2 mg/l (n=15, 24%)

63% response when between 0.2 to 0.7 mg/l (n=32, 52%)

80% when >0.7 mg/l (n=15, 24%)

For the whole population, Cmin, Cmax or AUC alone was not associated with VR at wk 24.

However, for patients with mutations score of <6 (n=52), Cmin tended towards a significant association with VR (p=0.07).

RESPONSE to ATV/rtv by PK/PD: GIQ

GIQ predicted VF better than PK parameters alone

GIQ (median [Q1; Q3]: 0.13 [0.06; 0.4]

Delta RNA M6 M0

---GIQ <0.33 (median [Q1; Q3])): -1 [-2.7; -0.2]

---GIQ >0.33 (median [Q1; Q3]): -1.8 [-2.7; -0.4]

Response to ATV/r at Wk 24 Assessed by GIQ

P=0.03

50% response when <0.13 (n=30, 46%)

63% response between 0.13 and 0.33 (n=16, 26%)

88% response when >0.33 (n=16, 26%)

GIQ (<0.33 vs >0.33)

Sensitivity: 91%

Specificity: 36%

PPV: 46%

NPV: 87%

LOGISTIC REGRESSION ANALYSES of VR at Wk 24

Univariate analysis showed virological outcome (VF>50 c/ml) at M6 to be associated with:

VARIABLE P

Plasma HIV RNA zenith: 0.04

No of previous treatment lines 0.07

No of baseline resistance mutations

--NRTI mutations 8 x 10 to the 4th

PI mutations:

Total 4 x 10-4

Major same

Minor 8 x 10-4

BMS/ANRS ATV/r mutations score 8 x 10-4

GIQ 0.01

Nature of PI mutations:

L10F/I/V, L24I, G73A/G/S/T, I84V, L90M <0.05

K20I/M/R, L33F/I/V, M46I/L, F53L, I54L/V, V77I <0.2