Prospective validation of a model-informed precision dosing tool for vancomycin treatment in neonates

ABSTRACT We recruited 48 neonates (50 vancomycin treatment episodes) in a prospective study to validate a model-informed precision dosing (MIPD) software. The initial vancomycin dose was based on a population pharmacokinetic model and adjusted every 36–48 h. Compared with a historical control group of 53 neonates (65 episodes), the achievement of a target trough concentration of 10–15 mg/L improved from 37% in the study to 62% in the MIPD group (P = 0.01), with no difference in side effects.

. Logistic or linear regression with generalized estimating equations and exchangea ble correlation structure was used to compare the groups.
A total of 48 patients with 50 treatment episodes in the study and 66 with 85 treatment episodes in the control group were recruited (Fig. 1).A total of 115 SS-C trough (defined as measurement up to 1.5 h before the next scheduled infusion) in the study and 95 in the control group were retained in the final analysis (additional 8 and 56 SS-non-C trough , respectively, in the sensitivity analysis).
Patients in the study group had significantly lower birth weight, gestational age, and body weight but higher creatinine levels compared with the control group (Table 1).
The administered doses were similar with a median (range) of 22.8 (15.0-67.7) in the study (in two episodes, optimized doses were considered inadequate and not implemen ted) and 26.9 (12.6-46.1)mg/kg of body weight/day in the control group, respectively.Dosing intervals were significantly shorter in the study vs the control group [median (range) 12 (6-18) h vs 12 (8-24) h; P < 0.001].
After the first and all other optimized doses, the TA% C trough 10-15 mg/L and TA% C trough 8-17 mg/L were significantly greater in the study compared with the control group (Table 2).Sensitivity analyses were in line with the main analyses (Table 2).Of the AUC/MIC calculated based on the first SS-C trough , 53.1% (26/49) in the study compared with 43.5% (20/46) in the control group, but 84.4% (54/64) and 69.4% (34/49), respec tively, when including all measured SS-C trough , were within the target range of 400-700 mg•h/L (Fig. 2).
All patients survived.No one had acute kidney injury (12).Otoacoustic emission test was not passed in ≥1 ear in 9/45 (20%) and 5/60 (8.3%) patients in the study and control groups, respectively (P = 0.06).All adverse events in the study group are listed in Table S1.
Using MIPD software, we showed that MIPD significantly improved the attainment of C trough (by 59% of the narrow target 10-15 mg/L and 38% clinically acceptable range 8-17 mg/L), outperforming that found in other neonatal studies, with similar side effect profile (4, 6, 7).
We considered both C trough and AUC/MIC for the optimization target.Even if the correlation between C trough and AUC/MIC is only moderate or poor, the surrogate marker, C trough 10-15 mg/L should mostly ensure adequate exposure (AUC/MIC 400-700 mg•h/L) of the drug (13,14).
Although the predictive performance of the numerous published neonatal vancomy cin popPK models varies widely, bias can be significantly reduced with the incorporation of a single TDM result already (9,15).We included all available TDM data in each subsequent dose optimization, although studies have not found additional value in including more than two prior TDM points in MIPD dose optimization (9,16).
Our study has some limitations.First, we used a historical control group that may bias the results due to differences in study populations (e.g., lower gestational age associated with a higher variability of PK and carrying higher risk of hearing disturbance).However, simulation and prospective intervention studies have shown improved TA% with the MIPD software (4,6).Thus, conducting a randomized controlled trial of MIPD may raise  ethical issues.Second, a large part of the study population and the population for the selection of the best-performing model in our MIPD tool development were recruited from one center, possibly affecting the generalizability of our results to other institutions.Third, dose optimization was performed by one researcher.Therefore, we may have missed some user-related real-life issues.
In conclusion, MIPD software appears to be a promising tool to improving PKPD target attainment of vancomycin therapy in neonates without undermining safety, especially when a narrow therapeutic window is targeted.

TABLE 1
Demographic characteristics of the study population b a At the start of the treatment; NS, not statistically significant.b Data are presented as median (range) if not stated otherwise.

TABLE 2
Target attainment of measured steady-state trough concentrations in the range of 10-15, 8-17, and 5-20 mg/L and of sensitivity analysis, including in addition predicted steady-state trough concentra tions for dosing intervals, where non-trough concentrations were available, after the first optimized and any adjusted dose a a Data are presented as N in target/N available (percent in target).NS, not significant (P-value > 0.05); C trough , trough concentration.