Effects of continuous renal replacement therapy on linezolid pharmacokinetic/pharmacodynamics: a systematic review

Although data considered for this review are only derived from studies of continuous treatments, wide variability in treatment modalities and operational parameters (such as blood, dialysate, replacement flows, etc.) was evident (see Table 1). Despite the wide variability observed, as well as treatment heterogeneity, extracorporeal clearance values for linezolid were similar across the different modalities: 1.2–2.3 L/h for CVVH, 0.9–2.2 L/h for CVVHDF and 2.3 L/h for CVVHD.

Although diffusive techniques should theoretically be characterized by higher extracorporeal clearance for low molecular-weight molecules (like linezolid) when compared with convective techniques, this effect was not observed across the studies assessed. Indeed, a number of factors might have influenced this finding, such as the variability in the flow set of the extracorporeal circuit and/or the specific geometrical characteristics of the various membranes, and the lack of a direct comparison of linezolid removal between the different techniques (diffusive vs convective). In fact, only one study [36] directly compared the PK linezolid parameters in CVVH and CVVHDF. Particularly, this study compared linezolid PK parameters in critically ill patients with sepsis and AKI treated with CVVH or CVVHDF at the same prescribed effluent dose (30 ml/kg/h) [36]. Unfortunately, the authors reported few data specifically for patients on CVVH or CVVHDF. Indeed, excluding the total drug clearance, no other PK parameters were reported or formally compared between the two groups. Furthermore, when comparing the total body clearance, extracorporeal clearance was not detailed (i.e. CVVH or CVVHDF clearance). A 20% reduction in total linezolid clearance (5.9 vs 4.5 L/h, p = 0.39) was observed in the group of patients treated with CVVH as compared to CVVHDF, making this finding in line with the concept of higher extracorporeal clearance achievable for small molecules through diffusive techniques. However, although a total effluent dose of 30 ml/kg/h was reported in the method section for both groups of patients, the CVVH group was in fact treated with a 20% lower total effective effluent dose (Qeff) compared to the CVVHDF group (26 ml/kg/h vs 32.17 ml/kg/h) (See Table 1). As the extracorporeal clearance is defined as the sieving coefficient (SC) (or saturation coefficient, SA, for CVVHDF) multiplied by Qeff, it is not surprising that the total linezolid clearance was 20% lower in the CVVH group as compared to CVVHDF (See Table 1). Therefore, in these conditions, it is not possible to infer that removal by diffusive techniques is more efficient than by convective techniques.

The analysis of the extracorporeal drug removal and PK derangements due to CRRT should also take into account the characteristics of the membrane used, in terms of surface area, composition and pore diameter [38].

Higher extracorporeal clearance of linezolid has been documented in patients treated with larger filters (1.2 vs 0.9 m²) [35]. Particularly, although these two groups of patients were treated with the same modality and treatment setting (CVVH in post-dilution, with a blood flow of 185–189 ml/minute and effluent dose of 27.7 ml/kg/h), extracorporeal clearance was higher when the larger filter was used (2.34 L/h vs 1.63 L/h) (see Table 1). Furthermore, membrane composition may also influence drug removal by adsorption. As the most common membranes used in the ICU, such as polymethylmethacrylate or polyacrylonitrile, also have unselective adsorption properties, further studies should take this issue into consideration. In fact, although membrane drug adsorption is commonly underrated, it is usually rapid and not reversible, leading to a reduction of antibiotic concentrations [39]. For example, Kraft et al. [32] and Carcelero et al. [24], using the same treatment modality (CVVHDF) and the same operational setting (effluent flow 2.6–2.7 L/h), observed the same extracorporeal clearance of linezolid (2.05–2.19 L/h), despite the former having used a polysulfone membrane with a twofold surface area (1.6 m²) compared to the acrylonitrile membrane used by the latter (0.9 m²) (see Table 1). In this case, the reduced transmembrane clearance obtained by the smaller acrylonitrile surface might have been compensated by its higher adsorption properties; as a consequence, the total extracorporeal clearance was similar to that observed with polysulfone membrane with by a broader surface area but lower adsorption capacity.

In the case of linezolid, a molecule with quite a low radius as compared to the large cut-off of the hemodiafilters commonly used in the ICU (high flux and/or HCO membranes), pore size is likely to be scarcely relevant. For example, Villa et al. observed a mean SA equal to 0.80 with HCO membranes used in CVVHD [9]; this value was similar to SA or SC reported in other studies in which patients with AKI were treated with hemodiafiltration or hemofiltration with standard high-flux membranes. In particular, Kraft et al. calculated SA values ranging from 0.77 to 0.81 during CVVHDF [32] and similar results were reported by several authors for the SC obtained during CVVH [24, 33‐35]. The removal of linezolid calculated by Villa et al. during HCO-CVVHD was similar to that reported in the literature for high-flux membranes [9], even though an increase in transmembrane linezolid loss should be expected with HCO if convective clearance is applied during the treatment (see aforementioned).

Apart from the key role of CRRT in determining PK antibiotic changes, even the critically ill status may be playing a major role [19]. Indeed, increased drug clearance was reported in patients with low SOFA scores [36]. Commonly, in patients with sepsis and AKI, who are undergoing CRRT and are on vasopressor support (as in the study by Roger et al.) [36], lower SOFA scores are mainly correlated with improved neurologic, respiratory and/or liver function. As 40% of the drug is usually metabolized and inactivated by the liver, it is not surprising that lower SOFA scores were associated with higher linezolid clearance.

The results were similar in our previous study performed in patients with sepsis and AKI treated with standard doses of linezolid and hemodialysis with HCO membranes [9], in which impaired liver function was associated with decreased corporeal drug clearance. As the total clearance (CL_tot) is the sum of the extracorporeal (CL_HCO) and non CRRT clearance, a subsequent increased ratio between extracorporeal and total linezolid clearance (CL_HCO/CL_tot) was observed in critically ill patients with AKI and severe liver dysfunction. In these conditions, the CL_HCO/CL_tot value is likely to be higher than 30%, only apparently identifying a treatment in which the extracorporeal clearance is critical for the total drug removal, potentially affecting its PK/PD parameters [9, 40]. In these conditions, in line with Roger’s results, the severity of the condition may influence the drug disposition.

Cmax and Cmin values are highly variable among patients considered in this review, as they ranged from 12.4 to 28.6 mg/L and from 1.7 to 14 mg/L, respectively (Table 1). Although Cmax values are similar or higher than those reported in the literature for healthy subjects, patients with renal impairment and ICU patients [13, 16], the Cmin values were lower than those reported for the same populations. Similarly, total clearance and volume of distribution values observed in the overall population were also highly variable (from 3.64 to 11.3 L/h and from 26.5 to 67.89 L, respectively) (Table 1). Patients’ comorbidities and organ dysfunction may significantly affect these parameters. For instance, the large Vd calculated by Roger et al. in the CVVH group might be explained by the likely greater fluid overload observed in these patients with respect to the CVVHDF group. Indeed, the net ultrafiltration fluid removal set in the CVVH group was about four times that in the CVVHDF group (net ultrafiltration 1100 vs 3620 ml respectively, see Table 1).

All these observations suggest that the clinical features of critically ill patients with sepsis and AKI treated with CRRT may play a key role in the achievement of appropriate PK/PD parameters for linezolid [36]. When considering pathogens with an MIC value of 2.0 mg/L for linezolid, two of the reviewed studies reported AUC/MIC ratios constantly below 85, and therefore not adequate for a clinical cure. What is more, when considering pathogens with an MIC value of 4.0 mg/L, only one study reported the optimal AUC/MIC ratio [15]. Therefore, the failure to achieve the optimal PK/PD targets for linezolid might increase the likelihood of emergence of microbial resistance and consequent clinical failure [36]. Furthermore, increased dosing of linezolid in critically ill patients with sepsis and AKI undergoing CRRT could expose patients to adverse effects. Thrombocytopenia has been reported as a frequent side effect of linezolid, especially in patients with renal dysfunction, in patients who develop heparin-induced thrombocytopenia (HIT) during anticoagulation treatment for CRRT, and in patients whose blood is exposed to low compatibility membranes.

Another partial explanation for our results may be related to linezolid drug-drug interactions. Drug-drug interaction with linezolid has been reported by several authors [14, 41‐47] Rifampin can markedly decrease the linezolid serum concentration (thus affecting the probability of target attainment for clinical success). On the contrary, the majority of drugs interacting with linezolid, such as P-glycoprotein and CYP isoform inhibitors (i.e. protein pump inhibitors, calcium channel blockers, macrolides and so on), increase linezolid serum levels and therefore do not contribute to increased risk of clinical failure. Unfortunately, no information on pharmacological treatments is available for patients in the studies analyzed for this systematic review.

In this context, if optimal linezolid PK/PD targets are not achieved in patients with sepsis and AKI treated with CRRT, clinical failure of antibiotic treatment is more likely, especially in the case of infections sustained by microorganisms with higher MIC values (e.g. Enterococci). Therefore, although conflicting opinions are reported on the use of TDM for linezolid therapy monitoring in critically ill patients with sepsis, it is however, to be recommended for these patients, especially if they develop AKI and eventually CRRT is to be started [2].