Machine learning to predict vasopressin responsiveness in patients with septic shock.

STUDY OBJECTIVES: The objective of this study was to develop and externally validate a model to predict adjunctive vasopressin response in patients with septic shock being treated with norepinephrine for bedside use in the intensive care unit. DESIGN: This was a retrospective analysis of two adult tertiary intensive care unit septic shock populations. SETTING: Barnes-Jewish Hospital (BJH) from 2010 to 2017 and Beth Israel Deaconess Medical Center (BIDMC) from 2001 to 2012. PATIENTS: Two septic shock populations (548 BJH patients and 464 BIDMC patients) that received vasopressin as second-line vasopressor. INTERVENTION: Patients who were vasopressin responsive were compared with those who were nonresponsive. Vasopressin response was defined as survival with at least a 20% decrease in maximum daily norepinephrine requirements by one calendar day after vasopressin initiation, without a third-line vasopressor. MEASUREMENTS: Two supervised machine learning models (gradient-boosting machine [XGBoost] and elastic net penalized logistic regression [EN]) were trained in 1000 bootstrap replications of the BJH data and externally validated in the BIDMC data to predict vasopressin responsiveness. MAIN RESULTS: Vasopressin responsiveness was similar among each cohort (BJH 45% and BIDMC 39%). Mortality was lower for vasopressin responders compared with nonresponders in the BJH (51% vs. 73%) and BIDMC (45% vs. 83%) cohorts, respectively. Both models demonstrated modest discrimination in the training (XGBoost area under receiver operator curve [AUROC] 0.61 [95% confidence interval (CI) 0.61-0.61], EN 0.59 [95% CI 0.58-0.59]) and external validation (XGBoost 0.68 [95% CI 0.63-0.73], EN 0.64 [95% CI 0.59-0.69]) datasets. CONCLUSION: Vasopressin nonresponsiveness is common and associated with increased mortality. The models' modest performances highlight the complexity of septic shock and indicate that more research will be required before clinical decision support tools can aid in anticipating patient-specific responsiveness to vasopressin.

Effects of cyclophosphamide related genetic variants on clinical outcomes of adult hematopoietic cell transplant patients.

PURPOSE: Genetic variants may influence the pharmacokinetics and pharmacodynamics (PKPD) of cyclophosphamide (CY). CY plays a critical role in conditioning chemotherapy for hematopoietic cell transplantation (HCT), but its use is limited by toxicity. We explored the effect of genetic variants, potentially affecting PKPD of CY, and outcomes after HCT. METHODS: This observational pharmacogenomic study included 85 adults with hematologic malignancies who received reduced intensity conditioning with CY, fludarabine, and total body irradiation. We collected recipient DNA prior to HCT and evaluated 97 candidate variants in 66 genes and 3 metabolism phenotypes potentially involved in PKPD pathways of CY. In multivariable analysis we investigated the association between the genotypes and four clinical outcomes: Day 180 non-relapse mortality (NRM) and day 180 overall survival (OS), acute graft-versus-host-disease (aGVHD) grades 2-4, and engraftment. p values were not adjusted for multiple testing. RESULTS: The median recipient age was 63 years (range 21-75). Acute myeloid leukemia was the most common diagnosis (34%; n = 29). In multivariable analysis adjusted for exposure to phosphoramide mustard, the final active metabolite of CY, we identified 6 variants in 6 genes associated with at least one of the clinical outcomes. An ABCC4 variant (rs9561778) was associated with poor Day 180 NRM (p < 0.01), MUTYH variant (rs3219484) with higher Day 180 NRM and aGVHD (both p < 0.01), and SYNE1 variant (rs4331993) with better Day 180 OS and engraftment (both p ≤ 0.01). CONCLUSION: The present study suggests that genetic variants influencing the PKPD of CY may help identify patients at risk for inferior outcomes after HCT using CY-based reduced-intensity conditioning.

The Impact of Rivaroxaban and Apixaban on Tacrolimus Trough Levels.

The solid organ transplant community is slow to adopt the routine practice of using direct oral anticoagulants. Rivaroxaban and apixaban share common metabolic pathways with tacrolimus. This study aimed to clarify the impact of rivaroxaban/apixaban on tacrolimus troughs. Fifty solid organ transplant recipients with concomitant use of tacrolimus and rivaroxaban/apixaban were retrospectively assessed for changes in tacrolimus troughs and dose. Average dose-adjusted tacrolimus troughs and average tacrolimus total daily doses prior to and after rivaroxaban/apixaban initiation were compared. Subgroup analyses evaluating rivaroxaban and apixaban individually were performed. Rivaroxaban was prescribed to 18 recipients, and apixaban was prescribed to 32 recipients. Transplanted organs included kidney (n = 22), lung (n = 18), liver (n = 7), simultaneous pancreas and kidney (n = 1), and simultaneous kidney and liver (n = 2). The median doseadjusted tacrolimus trough and tacrolimus total daily dose prior to rivaroxaban/apixaban initiation was 2.15 ng/mL/mg (IQR 1.17, 3.37) and 4 mg (IQR 1.88, 6.25), respectively. The median dose-adjusted tacrolimus trough and tacrolimus total daily dose after rivaroxaban/apixaban initiation was 2.16 ng/mL/mg (IQR 1.24, 4.10) and 3.55 mg (IQR 1.5, 6.35), respectively. No significant difference was found between average dose-adjusted tacrolimus troughs or tacrolimus total daily doses before and after rivaroxaban/apixaban initiation or in the individual subgroup analyses for rivaroxaban/apixaban. It is unlikely that initiating rivaroxaban/apixaban affects tacrolimus troughs or requires tacrolimus dose adjustment. This study does not elucidate if tacrolimus affects rivaroxaban/apixaban pharmacokinetics or pharmacodynamics.

Higher Fludarabine and Cyclophosphamide Exposures Lead to Worse Outcomes in Reduced-Intensity Conditioning Hematopoietic Cell Transplantation for Adult Hematologic Malignancy.

Reduced-intensity conditioning regimens using fludarabine (Flu) and cyclophosphamide (Cy) have been widely used in hematopoietic cell transplantation (HCT) recipients. The optimal exposure of these agents remains to be determined. We aimed to delineate the exposure-outcome associations of Flu and Cy separately and then both combined on HCT outcomes. This is a single-center, observational, pharmacokinetic (PK)-pharmacodynamic (PD) study of Flu and Cy in HCT recipients age ≥18 years who received Cy (50 mg/kg in a single dose), Flu (150 to 200 mg/m2 given as 5 daily doses), and total body irradiation (TBI; 200 cGy). We measured trough concentrations of 9-ß-D-arabinosyl-2-fluoradenine (F-ara-A), an active metabolite of Flu, on days -5 and -4 (F-ara-ADay-5 and F-ara-ADay-4, respectively), and measured phosphoramide mustard (PM), the final active metabolite of Cy, and estimated the area under the curve (AUC). The 89 enrolled patients had a nonrelapse mortality (NRM) of 9% (95% confidence interval [CI], 3% to 15%) at day +100 and 15% (95% CI, 7% to 22%) at day +180, and an overall survival (OS) of 73% (95% CI, 63% to 81%) at day +180. In multivariate analysis, higher PM area under the curve (AUC) for 0 to 8 hours (PM AUC0-8 hr) was an independent predictor of worse NRM (P < .01 at both day +100 and day +180) and worse day +180 OS (P < .01), but no associations were identified for F-ara-A trough levels. We observed lower day +100 NRM in those with both high F-ara-ADay-4 trough levels (≥40 ng/mL; >25th percentile) and low PM AUC0-8 hr (<34,235 hr ng/mL; <75th percentile), compared with high exposures to both agents (hazard ratio, 0.06; 95% CI, 0.01 to 0.48). No patients with low F-ara-ADay-4 (<40 ng/mL; <25th percentile) had NRM by day +100, regardless of PM AUC. The interpatient PK variability was large in F-ara-ADay-4 trough and PM AUC0-8 hr (29-fold and 5.0-fold, respectively). Flu exposure alone was not strongly associated with NRM or OS in this reduced Flu dose regimen; however, high exposure to both Flu and Cy was associated with a >16-fold higher NRM. These results warrant further investigation to optimize reduced-intensity regimens based on better PK-PD understanding and possible adaptation to predictable factors influencing drug clearance.

Tacrolimus Elimination in Four Patients With a CYP3A5*3/*3 CYP3A4*22/*22 Genotype Combination.

Cytochrome P450 3A5 (CYP3A5) and cytochrome P450 3A4 (CYP3A4) are the predominate enzymes responsible for tacrolimus metabolism. The presence of CYP3A4 and CYP3A5 genetic variants significantly affects tacrolimus clearance and dose requirements. CYP3A5*3 is a loss-of-function variant resulting in no CYP3A5 enzyme production. CYP3A4*22 is a variant that reduces production of functional CYP3A4 protein. Caucasians commonly carry these variant alleles but are very rarely homozygous for both CYP3A5*3 and CYP3A4*22. This report describes four kidney transplant recipients who carry a rare genotype combination (CYP3A5*3/*3 and CYP3A4*22/*22). These patients were identified from a larger cohort of Caucasian kidney transplant recipients (n=1366). To understand the significance of this genotype combination on tacrolimus troughs and doses, we compared these patients to recipients without this combination. Patients homozygous for both variants are at risk for profound reductions in metabolism of CYP3A substrates. A 342% and a 90.6% increase in the median dose-normalized trough was observed, when the CYP3A5*3/*3 and CYP3A4*22/*22 genotype combination was compared to the CYP3A5*1/*1 and CYP3A4*1/*1 genotype combination and the CYP3A5*3/*3 and CYP3A4*1/*1 genotype combination, respectively. These four individuals only required on average 2.5 mg/day of tacrolimus. Knowledge of these genotypes would be useful in selecting appropriate tacrolimus doses to avoid overexposure.