Effects of cardiopulmonary bypass on the disposition of cefazolin in patients undergoing cardiothoracic surgery

Abstract The aim of the study was to evaluate the disposition of plasma unbound cefazolin in patients undergoing cardiothoracic surgery with cardiopulmonary bypass (CPB). Adult patients undergoing cardiothoracic surgery with CPB were enrolled in the study. Cefazolin sodium was given intravenously before skin incision (1 g) and at the beginning of CPB (2 g). Thereafter, an additional dose (1 g) was given every 4 hours. Seven to ten blood samples were collected before and during surgery. Plasma total and unbound (ultrafiltrated) cefazolin concentrations were analyzed using an HPLC‐UV method. Plasma protein binding was analyzed with the Langmuir model. Twenty‐seven patients (aged 70 ± 12 years, body weight 62 ± 12 kg, mean ± SD) with GFR >30 mL min−1 completed the study. There was a significant (P < 0.001) increase in median plasma unbound fraction of cefazolin from 21% before skin incision to 45% during CPB (P < 0.001), which was accompanied by a significant (P < 0.001) reduction in median plasma albumin concentration from 36 to 27 g L−1. Plasma concentrations of unbound cefazolin exceeded the assumed target thresholds of 2 μg mL−1 in all samples and of 8 μg mL−1 in all but one of 199 samples. The increased plasma unbound fraction of cefazolin would be attributable to dilutional reduction of serum albumin at the beginning of CPB and to saturable plasma protein binding of cefazolin. These data reveal CPB may alter the plasma protein binding and possibly distribution of cefazolin. Further studies are warranted to reappraise the protocol of antimicrobial prophylaxis with cefazolin in patients undergoing surgery with CPB.


| INTRODUCTION
Surgical site infection (SSI) is one of the major postoperative complications and is associated with increased morbidity and mortality. 1,2 Particularly, SSIs (such as mediastinitis) that develop in patients undergoing cardiothoracic surgery are associated with a high mortality. Previous meta-analyses revealed that antimicrobial prophylaxis used in cardiothoracic surgery is associated with better surgical outcomes. 3,4 While cardiothoracic surgery is often performed with cardiopulmonary bypass (CPB), there is a paucity of knowledge about changes that may occur in the pharmacokinetics of drugs during CPB. As a result, consensus has not been attained about the choice and dosing protocols of antibiotics for these patients. 5 The guidelines of the American Society of Health-System Pharmacists recommend that traditional antimicrobial prophylaxis protocols for general surgery should not be changed for patients undergoing cardiothoracic surgery with CPB, unless further clinical outcome data obtained from well-designed studies are available. 1 Nevertheless, recent studies argued that conventional regimens of cefazolin prophylaxis for these patients may not be reliable in maintaining a target plasma threshold (for example, total drug concentration of 40 μg mL −1 , assuming normal protein binding of 80%-86% 6,7 during CPB) at intraoperative trough and/or at wound closure in patients with normal renal function undergoing cardiac surgery with CPB. 7 CPB may alter pharmacokinetics of antimicrobial agents by multiple mechanisms. For instance, approximately 1.3 L of crystalloid solution is administered to prime the bypass circuit, and a balanced crystalloid solution is administered prior to and during CPB to expand the circulating volume for maintaining cardiovascular stability, due to blood loss during operation. As a result, substantial dilutional reduction of serum albumin levels (<30 g L −1 ) occurs, and it may increase the volume of distribution and alter plasma protein binding of drugs that have a small volume of distribution and extensive plasma protein binding. In addition, heparin is routinely administered intravenously at 300 to 400 U kg −1 into the CPB circuit for anticoagulation. A heparin-induced increase in plasma free fatty acids may competitively displace drugs from their albumin-binding sites. 8 Furthermore, CPB may affect systemic elimination of drugs by altering hepatic and/or renal blood flow. 9,10 Current guidelines recommend intravenous cefazolin sodium as the drug of choice for antimicrobial prophylaxis in patients undergoing surgery, due to its excellent activity against common pathogens causing SSI. 1,11 Since cefazolin binds extensively to albumin (80%-86%) 6,7 and has a small volume of distribution (10 L per body), 6 its disposition may be particularly susceptible to CPB-induced physiological changes. Indeed, previous studies reported a 28%-50% decrease in plasma total cefazolin concentration during CPB. 12,13 Collectively, there is a paucity of knowledge about the impact of CPB on the disposition of cefazolin. Here, we report the disposition of total and unbound cefazolin throughout cardiothoracic surgery performed with CPB, with reference to the target plasma unbound drug concentration.

| Antimicrobial prophylaxis
Cefazolin sodium for injection (Nichiiko, Toyama, Japan) at a dose of 1 g was infused intravenously over 30 minutes. The infusion was started approximately 60 minutes before skin incision. Thereafter, additional dose was administered every 4 hours. In addition, 2 g of cefazolin was added to the priming solution of the CPB circuit. (Figure 1).

| Blood Samplings
Blood samples were collected before the preoperative dose was given and 30 minutes after the completion of infusion. Blood samples were also taken immediately before and 30 minutes after the start and the end of CPB, and at wound closure. When an additional dose of cefazolin was needed for lengthy operation with CPB, blood samples were also obtained immediately before and  and 50 μL of the mixture was injected into the HPLC column. We found no significant absorption of cefazolin to ultrafiltration filters or device. The lower limit of detection with a signal to noise ratio of 3 or greater was 0.1 μg mL −1 . Intra-and interday coefficients of variation of cefazolin assay were less than 5.8% and 6.5%, respectively.

| Target plasma cefazolin concentrations
We set plasma unbound cefazolin concentrations of 2 and 8 μg

| Comparisons between measured plasma
unbound cefazolin concentrations and those estimated using the Langmuir model Decroix et al. 19 revealed that cefazolin is bound exclusively to albumin and that binding is saturable. Using the Langmuir model (see below), they found that cefazolin has an affinity constant (K a ) of 16 600 ± 1600 mol L −1 and one saturable binding site (n = 0.73 ± 0.02). The Langmuir model is formulated as follows: F I G U R E 1 Cefazolin dosing and blood sampling schedules for patients undergoing cardiothoracic surgery with cardiopulmonary bypass (CPB). The down arrows (↓) represent intravenous administrations of cefazolin with doses (1 or 2 g), and the arrowheads (▼) represent blood samplings. The horizontal box represents the duration of CPB. The times of skin incision and would closure are indicated by up arrows (↑). In a patient whose course of surgery was complicated before CPB initiation, the second dose of cefazolin was administered before the initiation of CPB ASADA ET AL.
where C b and C u are protein-bound and unbound drug concentrations, respectively, at equilibrium; and C prot is the concentration of binding protein. Since plasma protein binding of cefazolin has been shown to be attributed exclusively to albumin, 19 C prot may be replaced by plasma albumin concentration (C alb ). Rearranging Equation 1 yields the following equation: where C tot is total drug concentrations. Note that the dissociation constant (K D ) equals 1/Ka by definition. We assume that CPB alters plasma albumin concentration but not the binding affinity of albumin to cefazolin. We also assume that the equilibrium of drug-protein binding is attained rapidly. Solving Equation 2 for Cu, the following equation is obtained 19 : By substituting C tot , C alb , and K D into Equation 3, we estimated Cu for each plasma sample. Here, C tot and C alb were measured in the plasma samples collected from patients during surgery, and K a was obtained from the literature. 19 Then, we compared plasma unbound cefazolin concentrations (Cu) that were estimated by Equation 3 and those actually measured.

| Statistical analyses
Demographic variables of the participants are presented as mean ± SD with range. Data of unbound fraction of cefazolin and plasma albumin concentration are presented using box and whisker plots.
Nonparametric multiple comparisons of plasma unbound fraction of cefazolin and plasma albumin concentration before skin incision, during CPB, and at the end of CPB were performed using the Steel-Dwass' test for pair-wise comparisons. Correlation between observed unbound cefazolin concentration and estimated concentration was analyzed by the least-squares linear regression. A P value less than 0.001 was considered statistically significant.   cefazolin measured before skin incision was 21%, but increased during CPB to 45%, and thereafter showed a tendency of returning to the basal level at wound closure (25%) (Figure 3). Significant (P < 0.001) differences were observed between before skin incision and during CPB, and between during CPB and at wound closure. In addition, the median plasma albumin concentration before skin incision was 36 g L −1 , but decreased significantly (P < 0.001) during CPB to 27 g L −1 , and increased again to 34 g L −1 at wound closure.

| Patient characteristics and adverse drug reactions
Significant (P < 0.001) differences were observed between before skin incision and during CPB, and between during CPB and at wound closure ( Figure 3).

| Comparison between estimated and measured plasma unbound cefazolin concentrations
A total of 199 pairs of data set were available for this analysis. A significant (P < 0.001) linear correlation with a slope close to unity was observed between the estimated plasma unbound cefazolin concentrations (C u,obs ) and the actually measured concentrations (C u,measured ): C u,obs = 1.05・C u,measured + 1.49 (r = 0.86).

| DISCUSSION
In the present study, we found that the median plasma unbound fraction of cefazolin showed an abrupt and profound increase during CPB (45%) compared to that at preoperative period (21%) (Figure 2). We speculate that these changes may be attributable to dilutional reduction of plasma albumin levels and saturable plasma protein binding of cefazolin. 20 Especially, the CPB circuit was primed with approximately 1.3 L of crystalloid solution, and fresh frozen plasma was subsequently administered to compensate blood loss during operation (Table 1). Since cefazolin binds extensively and exclusively to albumin (80%-86%), 6   assuming that plasma protein binding of the drug during surgery with CPB was the same as that of healthy subjects (80%-86%). 6,7 Our study suggests that they might have underestimated the aneurysm open repair surgery and reported that the penetration of unbound drug from plasma to ISF was 85% (78% to 106%). 28 Thus, the plasma concentration of unbound cefazolin would reflect the drug concentration at the target site.
In conclusion, our results indicate that altered plasma protein binding of the drug during and after CPB should be taken into consideration when modeling the disposition of unbound drug concentrations. Individualization of antimicrobial prophylaxis with cefazolin for adult patients undergoing CPB may be performed using a comprehensive pharmacokinetic modeling taking into account CPBinduced physiological changes.

ACKNOWLEDG EMENTS
The study was supported by Japan Society for the Promotion of Science (JSPS) KAKENHI Grant Number JP17H00452. The authors appreciate the staff of operation room for their assistance in the study.

DISCLOSURE
The authors have stated explicitly that there are no conflicts of interest in connection with this article.