Vancomycin Guideline 2020:

How to Calculate
AUC/MIC in Practice
Jennifer Le
Pharm.D., MAS, BCPS-ID, FIDSA, FCCP, FCSHP
Professor of Clinical Pharmacy
University of California San Diego
Skaggs School of Pharmacy
 Disclosures
Dr. Le declares no conflicts of interest, real or apparent, and no financial
interests in any company, product, or service mentioned in this program,
including employment, gifts, stock holdings, and honoraria.
Dr. Le has previously and currently receives investigator-initiated research
funding from
• National Institutes of Health — 1K23AI089978 (PI Le, NIAID),
1R01HD095547-01 (PI Bradley, NICHD), 1U54HD090259 (PI Caparelli
& Tremoulet, NICHD), 1T32HD087978-01 (PI Tremoulet & Capparelli,
NICHD), HHSN27500027 (PI Benjamin, NICHD)
University of California San Diego and Rady Children’s Hospital utilize
PrecisePK (formerly T.D.M.S. 2000) for clinical decision-support in the
implementation of AUC-guided vancomycin dosing.
Content
2020 Guideline Recommendations

AUC24 vs Trough Dosing & Monitoring

AUC24 : MIC Calculation

Bayesian AUC Estimation

3
 QUESTION 1

Raise Your Hand if

You Believe Vancomycin is Dead?

Yes
No
Therapeutic Vancomycin Dosing and Monitoring
AUC24:MICBMD of
Endorsement AUC vs Trough
400 – 600
• Trough • Adults &
• ASHP monitoring not Pediatrics
• IDSA recommended
• Outcomes
• AUC-Guided
• SIDP Dosing and data from
• PIDS Monitoring adult studies
Suggested

Am J Health Syst Pharm 2020 May 19;77(11):835-864 

AUC-Guided Dosing and Monitoring
Once-weekly
monitoring Pediatrics and adults
for stable
patients
Aggressive dosing for serious MRSA infections

Doesn’t
require steady- Patient High risk of nephrotoxicity (e.g., critically-ill
state serum Population patients receiving concurrent nephrotoxins)
levels AUC-Guided
Dosing and Unstable (i.e., deteriorating or significantly
Monitoring improving) renal function

Prolonged courses of therapy (more than 3 – 5 days)

Monitor
Peak and a
Two levels with Bayesian estimation. Trough
AUC using
trough only monitoring may be sufficient in some
Bayesian
monitoring patients but more data are needed.
software
preferreda
programs
AUC24/MIC
• Effectiveness
• Nephrotoxicity
Vancomycin PK/PD: Murine Thigh Infection Model

Change in Log10 CFU/Thigh Over 24 Hr

2

0 Stasis
-1

-2

-3

-4
30 100 300 1000 10 100 300 1000 0 25 50 75 100
24-Hr AUC/MIC Peak/MIC T>MIC

Ebert S. In vitro cidal activity and pharmacokinetic parameters for vancomycin against methicillin-susceptible and resistant S.
aureus [abstract 439]. In: Program and abstracts of the 27th Interscience Conference on Antimicrobial Agents and Chemotherapy
(New York). Washington, DC: American Society for Microbiology; 1987.
 Vancomycin Pharmacodynamics
in Patients with S. aureus Pneumonia
100

AUC/MIC <400
80

Percent Culture Positive 60

40
AUC/MIC >400 P = 0.0402

20

0
0 10 20 30
Day of Eradication

Odds ratio for clinical success; VAN AUC/MIC > 350 = 7.19; CI 1.91-27.4: P 0.0036
Moise-Broder PA et al. Clin Pharmacokinet. 2004;43:925-942.
 QUESTION 2

While clinical studies in adults indicate strong

association between vancomycin AUC and treatment
effectiveness, trough alone should be used to assess
the risk of acute kidney injury
per guideline recommendation.
True
False
 QUESTION 2

While clinical studies in adults indicate strong

association between vancomycin AUC and
treatment effectiveness, trough alone should be
used to assess the risk of acute kidney injury
per guideline recommendation.
True
False (CORRECT)
Time-to-Nephrotoxicity Stratified Kaplan-Meier Analysis

1.0
Stratum 0, Linezolid
Probability of Remaining

0.8 Stratum 1, standard

Vancomycin < 4 g/day
Non-Nephrotoxic

0.6

0.4
P < 0.0001
Stratum 2, high dose
0.2
Vancomycin > 4 g/day
0 5 10 15 20
Days After Initiation of Therapy
Lodise T. et al. Antimicrob Agents Chemother. 2008;52:1330-36.
 Vancomycin-Associated Nephrotoxicity
In the “15-20 mg/L” Trough Era: A Systematic Review and Meta-Analysis
Troughs > 15 mg/L Troughs < 15 mg/L
Odds Ratio
M-H, Random, Odds Ratio
Study or Subgroup Events Total Events Total Weight 95% Cl M-H, Random, 95% Cl
High troughs ≥15mg/L Low trough <15mg/L Odds Ratio Odds Ratio
Bosso et al (3) 42 4.30 [2.19, 8.43] Study or Subgroup Events Total Events Total Weight M-H, Random, 95%CI M-H, Random, 95% CI
142 13 146 10.7%
Bosso et al. (3) 42 142 13 146 10.7% 4.30 [2.19, 8.43]
Cano et al (4) 22 7 4.32 [1.74, 10.69]
89 99 8.1 Cano et al. (4) 22 89 7 99 8.1% 4.32 [1.74, 10.69]
Chung et al (7) 12 1.85 [0.69, 4.96]
Chung et al. (7) 12 25 16 48 7.4% 1.85 [0.69, 4.96]
25 16 48 7.4 Hermsen et al. (19) 5 16 4 39 4.3% 3.98 [0.91, 17.46]
Hermsen et al (19) 5 4 3.98 [0.91, 17.46] Hidayat et al. (20) 11 63 0 32 1.4% 14.24 [0.81, 249.87]
16 39 4.3 Jeffres et al. (24) 27 49 13 45 8.6% 3.02 [1.28, 7.11]
Hidayat et al (20) 11 0 14.24 [0.81, Kralovicova et al. (26) 21 60 29 138 10.7% 2.02 [1.04, 3.96]
63 32 1.4 249.87] Kullar et al. (27) 27 139 23 141 11.5% 1.24 [0.67, 2.28]
Jeffres et al (24) 27 3.02 [1.28, 7.11] Kullar et al. (28) 8 116 1 84 2.4% 6.15 [0.75, 50.13]
49 13 45 8.6 Lodise et al. (36) 7 27 14 139 7.1% 3.13 [1.12, 8.69]
Kralovicova et al (26) 21 2.02 [1.04, 3.96] McKamy et al. (38) 16 57 8 110 8.0% 4.98 [1.98, 12.52]
60 29 138 10.7
Minejima et al. (40) 17 72 25 155 10.5% 1.61 [0.80, 3.21]
Kullar et al (27) 27 1.24 [0.67, 2.28] Prabaker et al. (49) 7 54 24 294 8.2% 1.68 [0.68, 4.11]
139 23 141 11.5
Zimmermann et al. (63) 8 12 0 33 1.3% 126.56 [6.19, 2585.90]
Kullar et al (28) 8 1 6.15 [0.75, 50.13]
116 84 2.4
Total (95%CI) 921 1503 100.0% 2.76 [1.94, 3.93]
Lodise et al (36) 7 3.13 [1.12, 8.69]
27 14 139 7.1
Total events 230 177
Heterogeneity: Tau² = 0.18; Chi² = 23.80, df = 13 (P = 0.03); I² = 45%
McKamy et al (38) 16 8 4.98 [1.98, 12.52] 0.01 0.1 1 10 100
57 110 8.0 Test for overal effect: Z = 5.66 (P < 0.00001) Low troughs <15mg/L High troughs ≥15mg/L
Minejima et al (40) 17 1.61 [0.80, 3.21]
72 25 155 10.5 0.01 0.1 1 10 100

Prabaker et al (49) 7 1.68 [0.68, 4.11] Troughs < 15mg/L Troughs > 15mg/L
54 24 294 8.2
Zimmerman et al (63) 8 0 126.56 [6.19, 2585]
12 33 11.3
Limit of Vancomycin Exposure and Nephrotoxicity
18
16
P = 0.020 15.6
Percent of Patients (%)

14
P= 0.004 P = 0.006
12 P =0.003
11.4 11.5
10 10.4

8
6 *Suggests an AUC limit
5.2 of 700 mg*h/L to lower
4 the risk of nephrotoxicity
3.7
3.3
2 2.4
< 1218 > 1218 < 677 > 677 < 683 > 683 < 18 > 18
0
AUC0-48 AUC0-24 AUC24-48 Cmin24

Zasowski et al. Antimicrob Agents Chemother. 2018;62:401684-17

 AUC- vs Trough-Guided Dosing on
Vancomycin-Associated Nephrotoxicity
P<0.001

AUC MONITORING
AUC monitoring
independently
associated TROUGH MONITORING
with lower
nephrotoxicity

Finch, N. et al. Antimicrob Agents Chemother.

2017 doi: 10.1128/AAC.01293-17. e01293-17.
AUC24 Calculation
Trapezoidal Method

Equation Method

• Dose Approach
• Concentration Approach
AUC24 Calculation: Trapezoidal

AUC2-3 = Cp2 + Cp3  (t3 – t2)

2
 AUC24 Calculation: Equation
Dose Approach Concentration Approach
(Pediatrics - Le 2014; Le 2013) (Adults - Pai 2014)

• AUC24 = Dose24 ÷ CL • Overestimation

– (Csoi – Ct) ÷ ke
– CL = ke x Vd
– Csoi = Back-extrapolate to start of infusion
– Calculate Vd using estimated or
observed peak • Underestimation
– Use two post-distribution – (Tinf x [Ceoi + Ct]/2) + (Ceoi – Ct)/ke
concentrations – Ceoi = Back-extrapolate to end of infusion
ke = ln(C1/C2)/Δtime • Multiply by interval/day = AUC24
– Use total daily dose
Consider alternative therapy for MRSA with MIC >1.5 mg/L.

Le J. Therapeutic Drug Monitoring, 2014;36:510-8 Le J. Ped Infect Dis J 2013;32:e155-e162 Pai MP. Adv Drug Deliv Rev 2014;77:50-57
AUC24 Calculation: Two-Level Post Dose

ke and Vd derived from a

prior population PK model from
the literature

Uses linear trapezoidal rule

to calculate AUC
 QUESTION 3
In a patient, peak and trough concentrations were measured
after the second dose. Since a PK software was unavailable,
short-infusion model was used to determine clearance to
estimate non-Bayesian AUC. After analysis of the measured
concentrations, the dosing interval was revised to every 8 hour
from 6 hour (total daily dose unchanged).
Do both the AUC24 and trough values change?
Yes, Both AUC24 and Trough Change Equations:
AUC24 = Dose24 ÷ CL
No, Only Trough Changes Cpss = Dose (1- e-ktin)e-kt
(CL)(tin)(1- e-k)
 QUESTION 3
Do both the AUC24 and trough values change?
• Same Dose
• Changed Dosing

Concentration
Interval from
every 6 (green) to
8 (blue) hr

AUC6H = AUC8H
Yes, Both AUC24 and Trough Change Time
No, Only Trough Changes (CORRECT)
Vancomycin Susceptibility: MIC Creep

Regional Differences
Clonal outbreaks
Vitek-2
Reverse MIC creep
MicroScan Diaz, R. et al. J Glob Antimicrob Resist. 2017;10:281-284
Diaz, R. et al. Clin Microbiol Infect. 2017;1:June

Susceptibility 23.doi:1016/j.cmi.2017.06.017
McNeil, J.et al. Pediatr Infect Dis J. 2016;3:263-268.
Goldman, JL. et al. Pediatr Infect J. 2014; 2:216-8
Testing Methods Sancak B. et al. J Antimicrob Chemother. 2013;11:2589-
91.Rybak, MJ. et al. J Clin Microbiol. 2013;7:2077-81
Phoenix Reynolds, R. et al. J Antimicrob Chemother. 2012;12:2912-8
Edwards B. J Clin Microbiol. 2012;2:318-25.
van Hal SJ, et al. J Antimicrob Chemother. 2011;10:2284-7
Sader, HS. et al. Diagn Microbiol Infect Dis. 2011;2:412-6
Sader, HS. et al. Antimicrob Agents Chemother.
2009;10:4127-32
Holmes, RL & Jorgensen JH. Antimicrob Agents Chemother.
E-test 2008;2:757-6
 Vancomycin MIC Distributions of S.
aureus: Sentry Program 1997-2016
No. (Cumulative %) at Each Vancomycin MIC, mg/L

No. <0.12 0.25 0.5 1.0 2.0 4.0 8.0

MSSA 114,297 49(0.1) 243(0.4) 28,8632 (25.5) 83,549 1469 (>99.9) 25 (100.0)
(98.6)
MRSA 77,145 18(<0.1) 220(0.3) 15,807 57,319 3745 (>99.9) 35 (> 99.9) 1 (100.0)
(20.8) (95.1)
Total 191,442 67(< 0.1) 463(0.3) 44, 669 (23.6) 140, 868 5314 60 (>99.9) 1 (100.0)
(97.2) (> 99.9)

Diekema DJ et al. Open Forum Infect Dis. 2019;6(S1):S47-53.

Bayesian Method
A Priori Data
Population PK model
Inter- and intra-subject
variability
“EMPIRIC”
Bayesian is Superior to
non-Bayesian Estimation
(which may under-predict AUC)
Patient-Specific AUC and Concentrations
A Posteriori Data
Dosing history
Measured serum
concentrations
Bayesian
Analysis
Individual PK estimates
Dosing
“DEFINITIVE” Schumacher 1984
Spiegelhalter 2000
Bayesian Method
“Population-based”
“EMPIRIC” pharmacokinetic
parameters
A Priori

A Posteriori

“DEFINITIVE”
“Patient-specific” Bayesian-Derived
pharmacokinetic
parameters
 Bayesian Method
• One-Compartment Model
• Maximize the posterior probability

• Non-linear regression to minimize error function

 QUESTION 4
To conduct therapeutic drug monitoring at a
community hospital, a Bayesian software tool
must be implemented to accurately assess
and predict AUC exposure.
True
False
 QUESTION 4
To conduct therapeutic drug monitoring at a
community hospital, a Bayesian software tool
must be implemented to accurately assess
and predict AUC exposure.

True
False (CORRECT)
 Bayesian Method
Advantages Disadvantages
• Potentially requires only one serum • Learning new (but old) concept
concentration • Capital expenditure
Not limited to trough only
• Samples do not need to be measured
at steady-state
• Modifiable to include select
pharmacokinetic models
• Adaptive program
• Prior data that can integrate different
patient populations (pediatric,
obesity, critically-ill)

Pai M et al. Advan Drug Del Rev 2014

 Therapeutic Drug Monitoring Optimization Software
Steady-state & non- Adjust according to
Evaluate ICU versus
Bayesian analysis steady state AUC calculation renal- or liver
stable patient
evaluations function

Use different models

Uses or allows for Easy to learn and
for obesity and Safeguard for data
different equations Adaptive Function use
different patient entry
for all ages
populations

Integration into
Storage capability to
electronic health Desktop versus PDA Compatible with
patient data the Technical service
records (HIPAA application different printers
patient
compliant)

Cost: institutional vs enterprise use

(Restriction on number of users and patients)
 Therapeutic Drug Monitoring Optimization Software
Software
Patient
records/database Yes No Yes Yes Yes Yes No
-EPIC -EPIC -EPIC
Electronic health -Cerner -Cerner -Cerner
records integration Yes No -Allscripts -Premier -Mirth Connect -Allscripts No
-Data
anonymization or
-HIPAA compliant -HIPAA compliant -HIPAA compliant informed consent -HIPAA compliant
-Encrypted or -Encrypted or -Encrypted or -Encrypted or -Encrypted or
Privacy and data password-protected -Personal login and password-protected password-protected password-protected password-protected
security database secure password database database database database None
User interface Desktop, Desktop Web-based, Desktop, Desktop,
platform PDA Web-Based Mobile device Web-based Mobile device Web-based Web-based
Required training Minimal Training required Minimal Minimal Training required Training required Minimal
Call support
User manual
User manual Live chat Live chat Tutorial Video
Online discussion User manual User manual User manual 8-5 PST live chat
User support User Manual forum Web support service Web support service Instructional videos (Olark) None
Number of user & No
patient restriction No NR NR Yes Yes No
1-compartment Yes Yes Yes Yes Yes Yes Yes
2-compartment Yes Yes No
model No (up to 3) Yes Yes (up to 3) Yes
Customizable Yes Yes No Yes Yes No No
 Therapeutic Drug Monitoring Optimization Software
Software
Single seat license: Single license: $186ª Clinical use:
Cost $150/$200* Executable version: Not available Not available Institution⁺: $802ª $2,000-$15,000/year Free
Personal use license: Free Non-clinical use:
$75/$125* Maintenance/support: $1,000-$3,000/year
Institutional license: Web version: Single- $36ª/year
$390/$550* $1,000/year Institution- $202ª/year

Adane (2015)
Winter (1994) Neely (2014) Buelga (2005) Thomson (2009) Haeseker (2014) Rodvold (1988) Masich (2020)
Matzke (1985) Glatard (2015) Goti (2018) Goti (2018) Zahálková (2011) Crass (2018) Roberts (2011)
References Ambrose (2004) Macdonald (2008) Sabourenkov Buelga  (2005) Adane (2015 Buelga (2005)
Bauer(1982) Hurst (1990) (2019) Colin (2019) Llopis –Salvia P (2006) Matzke (1984)
Birt (1990) Carreno (2017) Tan  (1990)  Bauer (1982)
Burton (1989) Neely  (2014) Im (2012)
Burton revised (1991) Osterkamp (1995)
Matzke variation (1991) Devine (1974)
Moellering (1981) Brater (1986)
Rodvold (1988) Dolton (2010))
Abbott (1992) Hong (2015)
Creighton (1992) Dunn (2019)
VA (1992) Krogstad (1980)
Uaamnuichai (1987) Cutler (1984)
Rodvelt (1989) Blouin (1982)
Burton (1991) Garaud (1984)
Ito (1993) Rotschafer (1982)
Fernandez (1996)
Teramachi. (2002)
Llopis-Salvia (2006)
Thomson (2009)
Yamamoto (2009)
Sanchez (2010)
Purwonugroho (2012)
Medellin-Garibay (2016)
 Interactive Case Study
• 54-year-old female patient was admitted to the hospital for septic shock
and initiated on vancomycin.
• Vancomycin 750 mg every 12 hr was initiated

Yes
No

• Is empiric vancomycin dose optimal? 33

Interactive Case Study:
Precise PK Demonstration

34
 Interactive Case Study • CrCl 70.6mL/min
• ^Short-infusion model
Population Non-Bayesian, Least-Squared Method^ Bayesian*
PK*
Rodvold First First Peak Steady-State First First Peak Steady-State
PK Parameter 1998 Trough and Trough Peak and Trough and Trough Peak and
Model Trough Trough
Measured 6.9; 5:30 6.9; 5:30 10; 5:30 6.9; 5:30 6.9; 5:30 10; 5:30
Concentrations 34; 8:00 44; 8:00 34; 8:00 44; 8:00
Volume of Distribution 29.40 29.40 19.81 15.96 29.42 25.32 22.76
(L)
Clearance (L/hr) 3.47 3.97 3.33 2.49 3.46 3.17 2.56
Half-Life of Elimination 5.89 5.10 4.10 4.40 5.89 5.54 6.16
Phase (hr)
Predicted Steady-State 431 378 451 603 432 473 585
AUC24 (mg-hr/L)
Observed Steady-State 10 10 10 10 10 10 10
Trough (mg/L) 
Predicted Steady-State 6.9 6.7 6.3 9.2 6.9 7.1 10.0
Trough (mg/L)

35
 Key Takeaways
• To ensure use of minimum effective dose of vancomycin:
– Bayesian-derived AUC24 (rather than trough)-guided dosing and monitoring is
recommended for adults and suggested for pediatrics
– Monitor based on Bayesian method with at least one (preferably two) concentrations
that do NOT have to be at steady-state
– Effectiveness target of AUC24 /MICBMD ~ 400 to 600 (assuming MIC 1 mg/L) based on
adult outcomes data
– Avoid AUC24 ≥ 700 to 800 mg-h/L to minimize nephrotoxicity
– Review the PK models to ensure they fit your institution’s patient population as this
will be used as a priori data
– Therapeutic drug monitoring, especially with Bayesian estimation, is required as
vancomycin population PK model can only account for ~50% of the inter-individual
variability
 Acknowledgement
Dr. Michael Rybak for sharing his slides and pictures.
Contact Information
Jennifer Le, PharmD, MAS, BCPS-ID, FIDSA, FCSHP, FCCP
Professor of Clinical Pharmacy
University of California, San Diego Skaggs School of Pharmacy and Pharmaceutical Sciences
Email: jenle@health.ucsd.edu @idpedle, @kidpharmacology
Website: http://pharmacy.ucsd.edu/faculty/le.shtml