Development and PK Challenges of a Murine Model for Acinetobacter baumannii infection Brian Luna, PhD, and Jürgen Bulia, PhD Team Effort

University of Southern California • Brian Luna, PhD • Brad Spellberg, MD

University of Florida • Jürgen Bulitta, PhD • George Drusano, MD • Arnold Louie, MD

Case Western • Robert Bonomo, MD Rationale behind animal chosen

• Mouse model • Murine models have proven useful historically for the study of host/pathogen interactions and preclinical drug efficacy studies.

NIH workshop review paper: Bulitta JB, Hope WW, Eakin AE, Guina T, Tam VH, Louie A, Drusano GL, Hoover JL. Antimicrob Agents Chemother. 2019; 63. pii: e02307-18. PMCID: PMC6496039. Study Design

TWA TWA TWA TWA Infecon iStat iStat (IV or OA) iStat Cytokines Cytokines Cytokines CFUs CFUs hrs ) hrs hrs 18 2 6 Hrs

Infection Infection (0 Mice are euthanized for sample collection Preinfection

TWA = Temperature, Weight, and Activity Scores CFUs = Blood (and lung homogenate for OA model) iStat= blood pH, serum bicarbonate, lactate, and base deficit Cytokines = Measure blood, lung and ELF compartments. proinflammatory IL-1B, IL-6, and TNF and the anti-inflammatory IL-10.

4 Hypotheses • Treatment of mice with FDA-approved antibiotics that are active in vitro against the infecting isolate will result in clinical success • Survival, clinical endpoints, decrease in CFU burden, inflammatory and physiological endpoints

• In contrast, treatment of mice infected with strains that are resistant to the FDA approved antibiotics will result in clinical failure.

5 A. baumannii virulence

Integrated Overview of A. baumannii virulence. A) Early clearance of the microbe by the three primary innate effectors—(1) complement, (2) neutrophils, and (3) macrophages —prevents sustained LPS-TLR4 activation and subsequent cytokine storm. B) If the organism can resist initial innate effector clearance and replicate, it triggers sustained LPS activation of TLR4, resulting in cytokine Wong D, Nielsen TB, Bonomo RA, Pantapalangkoor P, Luna B, Spellberg B. storm and sepsis syndrome. Clinical and Pathophysiological Overview of Acinetobacter Infections: a Century of Challenges. Clin Microbiol Rev. 2017; 30:409–447. PMCID: PMC5217799 6 A. baumannii clinical isolates Table 2. LD100s of A. baumannii Isolates.1 IV bacteremia model OA pneumonia model Table 1. Antibiotic MICs for Clinical Sublethal LD Sublethal LD CFU A. baumannii isolates. 100 100 CFU CFU CFU MIC (µg/ml) *HUMC1 5.4x106 2.0x107 2.5x108 4.7x108 A. baumannii AMK MER PMB LAC-4 1.0x107 2.5x107 1.9x107 2.8x107 Strain R 7 7 7 7 *HUMC1 128 128 0.25 *LAC-4 Col 5.7x10 9.7x10 2.4x10 7.6x10 7 7 8 8 *VA-AB41 8 64 0.50 *VA-AB41 1.0x10 4.3x10 3.2x10 6.2x10 8 8 LAC-4 128 4 0.25 ATCC 17978 5.0x10 9.0x10 N/A N/A 8 *LAC-4 ColR 64 1 64 C-14 5.0x10 N/A N/A N/A ATCC 17978 8 0.25 0.5 C-8 8.6x108 9.6x108 N/A N/A C-14 2 1 8 1The number shown reflect the highest sublethal inocula or lowest lethal inocula identified C-8 8 8 16 Temperature and A) Bacteremia Model B) Oral Aspiration Model Acvity Scores Body Temperature Body Temperature 50 HUMC1 (M) 50 • Blood infection model: The HUMC1 (M) 40 HUMC1 (F) 40 HUMC1 (F) physiology and host LAC-4 ColR (F) 30 LAC-4 ColR (F) LAC-4 ColR (M) 30 response indicates that the LAC-4 ColR (M) 20 VA-AB41 (F) Temp (˚C) Temp 20 VA-AB41 (F) mice die due to sepsis. 10 VA-AB41 (M) (˚C) Temp 10 VA-AB41 (M) • Oral aspiration pneumonia: 0 0 2 18 0 0 2 18 Hours The physiology and host Hours response indicates that the Clinical Activity C) D) Clinical Activity 5 mice die due to respiratory HUMC1 (M) 5 HUMC1 (M) HUMC1 (F) failure. 4 HUMC1 (F) VA-AB41 (M) 4 VA-AB41 (M) VA-AB41 (F) 3 VA-AB41 (F) LAC-4 ColR (M) 3 LAC-4 ColR (M) LAC-4 ColR (F) 2 LAC-4 ColR (F) 2 Clinical Acitivy Score 1 Clinical Acitivy Score 0 2 18 1 0 2 18 Hours Hours Bacteremia Oral Aspiration 400 Glucose 400 Glucose 300 300 Sepsis Biomarkers 200 200 100 100 mg/dl mg/dL 0 0 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20

150 BUN 150 BUN • Blood infection model: The 100 100 physiology and host 50 50 mg/dl mg/dL 0 0 response indicates that the 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20

mice die due to sepsis. 7.4 pH 7.3 pH 7.2 7.1 • Oral aspiration pneumonia: 7 pH pH 6.9 6.8 The physiology and host 6.6 6.7 -2 0 2 4 6 8 10 12 14 16 18 20 -2 0 2 4 6 8 10 12 14 16 18 20 response indicates that the 29 29 HCO3 HCO3 mice die due to respiratory 24 24 19 19

failure. mmol/L 14 mmol/L 14 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20

20 Base Deficit 20 Base Deficit 15 15 10 10 5 5

mmol/L 0

0 mmol/L 0 2 4 6 8 10 12 14 16 18 20 0 2 4 6 8 10 12 14 16 18 20 Time (hours) Time (hours) HUMC1 (M) HUMC1 (F) HUMC1 (M) HUMC1 (F) VA-AB41 (M) VA-AB41 (F) VA-AB41 (M) VA-AB41 (F) LAC-4 Col.R (M) LAC-4 Col. R (F) LAC-4 Col.R (M) LAC-4 Col. R (F) Bacteremia Cytokines 10000 TNF Il-6 1000000 1000 10000 100 100 pg/ml 10 pg/mL • Blood infection model: The 1 1 0 2 4 6 8 101214161820 0 2 4 6 8 101214161820 physiology and host Oral Aspiration (Blood)

response indicates that the 10000 TNF 1000000 Il-6 1000 mice die due to sepsis. 10000 100

• Oral aspiration pneumonia: pg/mL 10 100 pg/mL The physiology and host 1 1 0 2 4 6 8 101214161820 0 2 4 6 8 101214161820 response indicates that the Oral Aspiration (Lungs) TNF 10000 1000000 100000 Il-6 mice die due to respiratory 1000 10000 100 1000 failure. 100 10 10 1 1 pg/mL 0 2 4 6 8 101214161820 pg/mL 0 2 4 6 8 101214161820 Time (hours) Time (hours) • Mice in the bacteremia HUMC1 (M) HUMC1 (F) HUMC1 (M) model tend to be sicker VA-AB41 (M) HUMC1 (F) than those in the OA model. Mouse model summary

• A. baumannii Isolates • Available upon request. • Isolates will be deposited to CDC & FDA Antibiotic Resistance Isolate Bank • Trigger to treat • Treatment is initiated 2 hrs post infection. • Decrease in activity score. • Increase in proinflammatory cytokines Mouse model summary

To validate the mouse models by testing the efficacy of FDA- approved antibiotics, which have been confirmed by FDA to be effective for treating clinical infections based on adequate and well-controlled clinical trials.

Our goal is to further characterize their validity for predicting efficacy of proven antibiotics that will or will not be effective at treating A. baumannii infections. Projected use of animal model in future drug development • Useful for testing novel therapeutic agents or novel therapeutic combinations • Academia and Industry • No other validated mouse model for the study of A. baumannii infection • Immune competent infection models • Characterized host/pathogen interaction • Humanized PK regimens • PK modeling in sick mice • Validated clinical endpoints Projected use of animal model in future drug development (cont.) • Mouse ≠ Human • Limitations • Comparatively higher inoculum to cause disease • Not all human clinical isolates are virulent in the mice (for A. baumannii, hypervirulent strains differ from non-hypervirulent strains; this is important to consider for PK/PD studies) • Differences in mouse vs human PK A. baumannii IV HUMC1 + A. baumannii IV HUMC1 + Amikacin (117 mg/kg) Blood CFUs Amikacin (117 mg/kg) Blood CFUs Efficacy 1010 1010 8 10 108

Resistant 6 (IV Model) 10 106

(CFUs/mL) 4 10 (CFUs/mL) 104 10 10 2 10 102 Log

R Log • HUMC1 (AMK ) 0 10 100

• VA-AB41 susceptibility:

S Male, PBS Male, AMK Male, PBS Male, AMK (AMK ) Female, PBSFemale, AMK Female, PBSFemale, AMK

In vitro 4 Hrs 18 Hrs A. baumannii IV VA-AB41 + A. baumannii IV VA-AB41 + Amikacin (117 mg/kg) Blood CFUs Amikacin (117 mg/kg) Blood CFUs • Response to 1010 1010 108 108 therapy Sensitive 106 106

(CFUs/mL) 4 (humanized PK) (CFUs/mL) 104 10 10 10 102 102

reflects Log Log predicted clinical 100 100 outcomes susceptibility: Male, PBS Male, AMK Male, PBS Male, AMK Female, PBSFemale, AMK Female, PBSFemale, AMK In vitro 4 Hrs 18 Hrs In vitro susceptibility: Resistant In vitro susceptibility: Sensitive Efficacy A. baumannii OA HUMC1 + A. baumannii OA HUMC1 + A. baumannii OA VA-AB41 + A. baumannii OA VA-AB41 + Amikacin (96.72 mg/kg) Blood CFUs Amikacin (96.72 mg/kg) Blood CFUs Amikacin (96.72 mg/kg) Blood CFUs Amikacin (96.72 mg/kg) Blood CFUs 10 10 10 1010 10 1010 8 8 (Lung 10 108 10 108 6 6 10 106 10 106 (CFUs/mL) 4 (CFUs/mL) 4 (CFUs/mL) 4 10 10 10 (CFUs/mL) 4

10 10 10 10 10 Model) 2 2 2 10 10 10 2

Log 10 Log Log Log 0 0 10 100 10 100

Male, PBS Male, PBS Male, AMK Male, PBS Male, AMK Male, AMK Male, PBS Male, AMK • HUMC1 Female, PBSFemale, AMK Female, PBS R Female, PBSFemale, AMK Female, AMK Female, PBSFemale, AMK (AMK ) 4 Hrs 18 Hrs 4 Hrs 18 Hrs • VA-AB41 (AMKS)

A. baumannii OA HUMC1 + A. baumannii OA HUMC1 + A. baumannii OA VA-AB41 + A. baumannii OA VA-AB41 + Amikacin (96.72 mg/kg) BAL CFUs Amikacin (96.72 mg/kg) BAL CFUs Amikacin (96.72 mg/kg) BAL CFUs Amikacin (96.72 mg/kg) BAL CFUs 1010 1010 1010 1010 • Response to 108 108 108 108 therapy 106 106 106 106 (CFUs/mL) (CFUs/mL) 4 4 (CFUs/mL) 10 104 10 (CFUs/mL) 104 10 10 10 (humanized 10 102 102 102 102 Log Log Log PK) reflects Log predicted 100 100 100 100

clinical Male, PBS Male, PBS Male, AMK Male, AMK Male, PBS Male, AMK Male, PBS Male, AMK Female, PBSFemale, AMK Female, PBSFemale, AMK Female, PBSFemale, AMK Female, PBSFemale, AMK outcomes 4 Hrs 18 Hrs 4 Hrs 18 Hrs PK- Amikacin Amikacin dose range study: visual predictive check – IV challenge 1.37 mg/kg 13.7 mg/kg 137 mg/kg Plasma conc. (mg/L) Plasmaconc.

Time (h)

18 Amikacin dose range study: visual predictive check – OA challenge 1.37 mg/kg 13.7 mg/kg 137 mg/kg Plasma conc. (mg/L) Plasmaconc.

Time (h) Lung ELF conc. (mg/L) ELFconc. Lung Time (h) 19 Population PK estimates for amikacin dose range studies

IV Model - Amikacin OA Model - Amikacin

Parameter Symbol Unit Mean (SE) BSV (SE) Symbol Unit Mean (SE) BSV (SE) Clearance CL mL/h 9.93 (7.9%) 0.185 (54.8%) CL mL/h 9.93 (7.9%) 0.185 (54.8%) Absorption half-life T1/2abs min 1.75 (12.1%) 0.1 (fixed) T1/2abs min 1.01 (34.9%) 0.1 (fixed) Volume of central compartment V1 mL 11.4 (14.1%) 0.1 (fixed) V1 mL 8.56 (10.6%) 0.1 (fixed) Ratio of AUC(elf) by AUC(plasma) Felf - 0.727 (16.9%) 0.1 (fixed) ELF to plasma equil. half-life T1/2elf min 21.9 (8.2%) 0.1 (fixed) Volume of ELF Velf mL 0.1 (fixed)

• Same clearance after IV and OA challenge. • Slightly larger volume of distribution for IV vs. OA. • Terminal half-life: 48 min for IV and 36 min for OA. • AUC(ELF) 72.7% of the AUC(plasma).

20 Clinical datasets used as input for the humanization

Avg. dose in humans: 20 to 25 mg/kg amikacin

21 Amikacin humanized dosing IV method (4 doses every 6 h) – Plasma Dailydig Doseitiz e–d117_d mg/kg/dayata_linear_scale Fractionsdigiti zofe dailyd_d dosesata_ 61.8%linea atr_ 0s h,c a18.6%le at 6 h, 11.3% at 12 h, and 8.2% at 18 h. digitized_data_log_scale Log Scale digitized#_Linearbduartdae_ Scaleltitn_e1a0rt_hs_cpaelercentile #burdett_10th_percentile 150 AUC (plasma) 150 #burdett_10th_percentile 0-24h #burdett_10th_percentile digitized_data_linear_scale #thburdett_90th_percentile 100 150 90#bpercentileurdett_ (clinical)90th_percen851tile mg·h/L #burdett_90th_percentile #burdett_90th_percentile SimulatedDC1:1 humanized conc. 296 mg·h/L DC1:1 DC1:1 DC1:1 100 10th percentile (clinical) 179 mg·h/L 100 10 100 #burdett_10th_percentile 150 #burdett_90th_percentile 1 50 50 DC1:1 50 100 0.1 0

Plasma concentration (mg/L) Plasmaconcentration 0 6 12 18 24 0 6 12 18 24 0 0 TimeTIM E(h) TimeTIM E(h) 50 0 6 12 18 24 0 6 12 18 24 TIME 22 TIME 0 0 6 12 18 24 TIME Amikacin humanized dosing OA method (4 doses every 6 h) – Plasma Daily Dose – 96.7 mg/kg/day digitized_data_linear_scale digitized_data_linear_scale Fractionsdi gofi tdailyize ddoses_da 65.0%ta_li nate 0a h,r _16.9%sca late 6 h, 10.5% at 12 h, and 7.4% at 18 h.

Linear Scale - PLASMA OLinear#Ab_uErL dScaleFe_tlti_n 1-eELF0arth_percentile OA#_pbluarsdmeatt__o1n0lyth_l_inpeearrcentile #burdett_10th_percentile 150 #burdett_10th_percentile AUC 150 150 200 AUC0-24h(plasma) 125 0-24h 150 digitiz#ebdur_ddetat_t9a0_tlhi_npeearcre_nsticleale #bthurdett_90th_percentile digitized_data_linear_scale 90#thbpercentileurdett_ (clinical)90th_pe851rce mg·hntile/L #burdett_10th_p90#ebrcupercentileerndtilee tt_ 9(clinical)0th_per851ce nmg·htile/L #burdett_10th_percentile 10#0burdett_90th_pSim.erc ehumanizedntile ELF conc. 244 mg·h/L #burdett_90th_percentile 150 Sim.DC Human.1:1 plasms conc. 244 mg·h/L DC1:1 DC1:1 DthC1:1 100 100 10th percentile (clinical) 179 mg·h/L DC1:1 10#bupercentilerdett_ 1(clinical)0th_per179ce nmg·htile/L Celf:1 100 100 #burdett_10th_percentile 75 150 150 100 #burdett_90th_percentile #burdett_90th_percentile 50

50 50 50 DC1:1 DC1:1 50 50 25 100 100 0 0 0(mg/L) Plasmaconcentration 0 6 12 18 24 0 6 12 18 24 0 0 0 TimeTIM E(h) TimeTIM E(h) 50 0 6 12 18 50240 6 12 18 24 0 6 12 18 240 6 12 18 24 TIME TIME 23 TIME TIME 0 0 0 6 12 18 240 6 12 18 24 TIME TIME Amikacin PK validation: OA model visual predictive checks IV model Plasma conc. (mg/L) Plasmaconc. Plasma conc. (mg/L) Plasmaconc. Time (h)

à Large variability

from 18 to 24 h (mg/L) ELFconc. Lung

Time (h) 24 Amikacin IV model Amikacin OA model Linear scale Linear scale - Plasma Plasma conc. (mg/L) Plasmaconc.

Log-scale Linear scale - ELF Plasma concentration (mg/L) Plasmaconcentration Lung ELF conc. (mg/L) ELFconc. Lung

Time (h) Time (h) 25 Population PK estimates for amikacin from humanized PK validation

IV Model - Amikacin OA Model - Amikacin

Parameter Unit Mean (SE) BSV (SE) Parameter Unit Mean (SE) BSV (SE) CL_0-6h mL/h 7.35 (6.8%) 0.1 (fixed) CL_0-6h mL/h 7.44 (5.5%) 0.1 (fixed) CL_6-12h mL/h 3.81 (6.5%) 0.155 (68.5%) CL_6-12h mL/h 7.69 (4.9%) 0.1 (fixed) CL_12-18h mL/h 1.39 (25.1%) 1.07 (29.9%) CL_12-18h mL/h 5.92 (13.1%) 0.500 (42.3%) CL_18-24h mL/h 1.60 (23.5%) 0.779 (47.4%) CL_18-24h mL/h 4.39 (17.6%) 0.656 (39.7%) T1/2abs min 1.00 (21.2%) 0.115 (146%) T1/2abs min 1.12 (24.7%) 0.1 (fixed) V1_0-6h mL 9.75 (14.7%) 0.1 (fixed) V1 mL 7.69 (6.4%) 0.1 (fixed) V1_6-12h mL 5.24 (10.4%) 0.250 (72.4%) Felf - 0.604 (4.9%) 0.1 (fixed) V1_12-18h mL 3.50 (28.3%) 0.430 (80.9%) T1/2elf min 36.4 (4.4%) 0.1 (fixed) V1_18-24h mL 1.51 (18.6%) 0.195 (147%) Velf mL 0.10 (fixed)

• Clearance decreased over time in both models (more pronounced for IV model). • Volume of distribution also decrease for IV model. • Large between subject variability (BSV) for clearance after 12 h. • ELF penetration results comparable to those in the dose range study.

26 Conclusions of dosing simulations – Amikacin - PK model in mice was used to simulate human-like drug exposures. - We targeted the 90th percentile of concentrations in VABP patients. - The AUCs for amikacin associated with the 10th and 90th percentiles were approximately 179 and 851 mg·h/L - The AUC in patients for a dose of 30 mg/kg is ~308 mg·h/L. - Peaks in critically-ill patient are typically 60 to 100 mg/L (medians). - Humanized concentrations were similar to the targeted conc. range. - Clearances decreased and became more variable after 12 h, potentially due to bacterial damage to the kidneys.

27 PK- Polymyxin B Polymyxin B

concentrations Linear scale (same y-axis range) in mice for IV challenge model

DVID 1 = polymyxin B1 DVID 2 = polymyxin B1-Ile DVID 3 = polymyxin B2 DVID 4 = polymyxin B3 DVID 5 = All 4 components

Polymyxin B1 was the

predominant component. Plasma concentration (mg/L) Polymyxin B3 was the lowest abundancy component. Relave bioavailability (F) esmated during modeling.

29 Structural model for all 5 dependent variables in IV challenge mice Structure: One-compartment model for each component, sum of four F equal to 1

SC dose SC dose SC dose SC dose Total polymyxin B F PB1 FPB1-ILe FPB2 FPB3 concentration modeled as the sum of all 4 components. ASC ASC ASC ASC PB1 PB1-ILe PB2 PB3 V1 and CL were shared across all four components ka ka ka ka of polymyxin B.

Other modeling choices Xplasma Xplasma Xplasma Xplasma and estimation algorithms were thoroughly tested at PB1 (V1) PB1-ILe (V1) PB2 (V1) PB3 (V1) this stage.

CL CL CL CL

30 Polymyxin B dose range study: VPC for 16 mg/kg – IV challenge Polymyxin B1 Polymyxin B2 Polymyxin B (sum)

Polymyxin B1-ILe Polymyxin B3 Time (h) Plasma concentration (mg/L) Plasmaconcentration

Time (h) 31 Parameter estimates for polymyxin B for IV model (importance sampling in S-ADAPT; Luna et al. dataset)

Symbol Parameter Unit Population Mean Between Subject (relative standard error, Variability, SE%) CV (SE%) T1/2abs(L) Absorption half-life at 8 and 12 mg/kg min 7.20 (9.2%) 0.109 (77.5%) T1/2abs(H) Absorption half-life at 16 mg/kg min 11.7 (54.7%) 0.125 (105%) CL Clearance mL/h 5.60 (6.0%) 0.471 (26.2%) V1 Volume of distribution for central compartment mL 26.3 (6.2%) 0.308 (26.1%) F(PB1) Bioavailability of PB1 - 72.4% (2.7%) small F(PB1-ILe) Bioavailability of PB1-ILe - 12.0% (3.3%) small F(PB2) Bioavailability of PB2 - 12.8% (3.2%) small F(PB3) Bioavailability of PB3 - 2.83% small

Estimating the same volume of distribution for all components stabilized the model and allowed us to estimate the BSV for both CL and V1.

All additive residual error standard deviations were fixed to 0.025 mg/L. All proportional residual errors were fixed to a 10% coefficient of variation. 32 Humanized dosing for polymyxin B for IV model Polymyxin B in critically ill patients 2 mg/kg load, 1.5 mg/kg Q12h maintenance dose; 20 IV: Mouse: 11 mg/kg at 0 h and 10 mg/kg at 12h Mouse_b: 9.5 @ 0h, 1.5 @ 6h, 8.5 @ 12h, 1.5 @ 18h AUC (mg*h/L)

15 90% 116 Percentiles 50% 80 in Patients 10 10% 52 Mouse 83 Mean predictions 5 Mouse_b 81 in mice

Polymyxin B in plasma (mg/L)plasma in PolymyxinB We recommend 11 mg/kg at 0 h 0 and 10 mg/kg at 12 h option 0 4 8 12 16 20 24 Time (h) Total concentrations were matched. This assumes that the protein

binding in plasma was similar between infected mice and patients. 33 Polymyxin B dose range study: VPC for validation – IV challenge Polymyxin B1 Polymyxin B2 Polymyxin B (sum)

Polymyxin B1-ILe Polymyxin B3 Sandri AM, Landersdorfer CB et al. CID 2013; 57:524-31 Plasma concentration (mg/L) Plasmaconcentration

Patients

Time (h) 34 Conclusions for humanizing PMB • For the IV model: we recommend 11 mg/kg (0 h) & 10 mg/kg (12 h) • Match the total drug AUC in plasma from 0 to 24 h between patients and mice • Mouse values stay between the 10th and 90th percentile in humans

• For the OA model, we recommend 7 mg/kg (0 h) & 6 mg/kg (12 h) • This matches the total drug AUC from 0 to 24 h in patient plasma with the unbound AUC in mouse ELF. The mouse plasma AUC are slightly higher.

• The IV vs. OA challenge considerable affected total clearance.

• It seems important to validate the PK in the mouse model using the same isolates and pathogens that are used in efficacy studies. PK- Meropenem Meropenem peak and trough concentrations for bolus and short-term infusion regimens in critically ill patients

• We sought to mirror the PK in critically ill patients with normal renal function by humanizing the mouse dosage regimens. • Targeted dosage regimen in humans: 2g short-term (30min) infusions every 8h. For this dosage regimen: • The average peak concentrations ~60 to 100 mg/L. • The average trough concentrations ~ 1 to 10 mg/L. • Meropenem is minimally protein bound in human plasma. Monte Carlo simulated meropenem concentrations in humans for 2g q8h as 30 min infusion in critically ill patients

Linear scale Log-scale

Meropenem concentrations in critically ill patients Meropenem concentrations in critically ill patients 100 100 90 80 70 10 60 P10% 50 P10% 40 P50% P50% 30 P90% 1 P90% 20 Meropenem conc (mg/L) conc Meropenem Meropenemconc (mg/L) 10 0 0.1 0 4 8 12 16 20 24 0 4 8 12 16 20 24 Time (h) Time (h)

38 Monte Carlo simulated meropenem concentrations in humans for 2g q8h as 30 min infusion in critically ill patients

Linear scale Humanized mouse regimen Log-scale

Meropenem concentrations in critically ill patients Meropenem concentrations in critically ill patients 120 50 mg/kg 100

100 P10% P50% 10 80 P90% 30 mg/kg P10% Mouse 4 doses 60 1 P50% 18 mg/kg 40 P90% 11 mg/kg 0.1 Mouse 4 doses 20 Meropenem conc (mg/L) conc Meropenem Meropenemconc (mg/L)

0 0.01 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 Time (h) Time (h)

Best capture of the human concentration time profile. However, this would result in 12 (=4 x 3) doses per day. 39 Monte Carlo simulated meropenem concentrations in humans for 2g q8h as 30 min infusion in critically ill patients

Linear scale Humanized mouse regimen Log-scale

Meropenem concentrations in critically ill patients Meropenem concentrations in critically ill patients 120 100 50 mg/kg 100 P10% P50% 10 80 P90% P10% Mouse 2 doses 60 1 P50%

40 17 mg/kg P90% 0.1 Mouse 2 doses 20 Meropenem conc (mg/L) conc Meropenem Meropenemconc (mg/L)

0 0.01 0 1 2 3 4 5 6 7 8 0 1 2 3 4 5 6 7 8 Time (h) Time (h)

If we use the criterion not to over-shoot the 90% percentile in patients, 2 doses every 8 h (i.e. 6 doses per day) do likely not well capture the time-course of plasma concentrations in humans.

40 fT>MIC Free time above MIC (fT>MIC) for different humanized regimens in mouse (OA model) using 4, 3 or 2 dosesMIC every (mg/L) 8 h Regimen 0.5 1 2 4 8 16 4Carbapenem doses Q8h 50 targets mg/kg @ in 0h mice:PlasmaStasis:fT>MIC 87% 76% ~20%65% fT>MIC52% 37% 22% 30 mg/kg @ 1.5h ELFNear-max75% killing:63% ~40%47% fT>MIC25% 8% 0% 18 mg/kg @ 3.5h MIC (mg/L) Regimen 11 mg/kg @ 5.5h 0.5 1 2 4 8 16 4 doses Q8h 50 mg/kg @ 0h Plasma 87% 76% 65% 52% 37% 22% 3 doses Q8h 5030 mg/kgmg/kg @@ 0h1.5h PlasmaELF 71%75% 63%63% 51%47% 40%25% 29%8% 18%0% 2518 mg/kgmg/kg @@ 2h3.5h ELF 61% 49% 36% 20% 7% 0% 12.511 mg/kg mg/kg @ @ 5.5h 5h

23 dosesdoses Q8hQ8h 5050 mg/kgmg/kg @@ 0h0h PlasmaPlasma 51%71% 43%63% 36%51% 28%40% 21%29% 13%18% 1725 mg/kgmg/kg @@ 3.5h2h ELFELF 43%61% 35%49% 26%36% 14%20% 7%7% 0%0% 12.5 mg/kg @ 5h 2 doses Q8h would only achieve fT>MIC >35% for MICs up to 2 mg/L in plasma. The covered MICs in ELF were approximately 2-fold lower than those in plasma. 2 doses Q8h 50 mg/kg @ 0h Plasma 51% 43% 36% 28% 21% 13%41 17 mg/kg @ 3.5h ELF 43% 35% 26% 14% 7% 0% Meropenem in mice

• Very short half-life relative to that in humans. • Limited to 6 daily SC injections

• The addition of cilastatin or probenecid may reduce meropenem clearance and improve murine model.

• Uranyl nitrate likely no option, since this compound is toxic and is becoming increasingly hard to purchase. Summary

• Immune competent mouse model • Humanized PK • AMK and PMB • MERO in progress • Therapeutic outcomes • Amikacin- outcomes reflect in vitro antimicrobial sensitivity data. • Polymyxin B & Meropenem- In progress • Each drug to be studied has its own unique problems for developing a humanized dosing strategy Acknowledgements

USC University of Case Western • Brad Spellberg Florida • Robert Bonomo • Jun Yan • George Drusano • Kristine Hujer • Peggy Lu • Arnold Louie • Zeferino Reyna • Dhruvit Sutaria • Funding: FDA • Juan Ruiz- • Yuanyuan Jiao Contract Delgado • Nirav Shah HHSF2230110199C • Rosemary She • David Brown