Imipenem, a broad-spectrum carbapenem antibiotic, possesses potent antibacterial activity against a diverse range of gram-positive and gram-negative bacteria, as well as anaerobic organisms (Dingle and Pitout, 2022). Its bactericidal action is mediated through the inhibition of bacterial cell wall synthesis (O’Donnell and Lodise, 2022). In the clinic, imipenem is widely utilized for the management of severe bacterial infections, including sepsis, pneumonia, intra-abdominal infections, and urinary tract infections (Chang et al., 2023; Fratoni et al., 2022; Heo, 2021; Marino et al., 2025; Titov et al., 2021). The primary objective of its dosing regimen is to maximize the duration during which plasma concentrations exceed the pathogen’s minimum inhibitory concentration (MIC) (Song et al., 2025). Consequently, imipenem is commonly coadministered with cilastatin, a renal dehydropeptidase inhibitor, to prevent renal metabolism and inactivation, thereby ensuring that a sufficient amount of intact drug reaches the target site and maintains excellent clinical efficacy (Zhanel et al., 2018).

Despite its strong antibacterial potency, the clinical use of imipenem is accompanied by a range of noteworthy adverse effects. An analysis of 2,574 adverse event reports related to imipenem revealed that over half of the events involved individuals aged over 60 years. Previously unreported adverse reactions, including brain atrophy and delirium, have also been identified (Jia et al., 2025). Among the most concerning adverse effects is dose-dependent central neurotoxicity (Sutter et al., 2015). Reported potential targets for carbapenem-induced neurotoxicity include the GABA_A receptor, glutathione S-transferase Pi, glutathione S-transferase Mu 1, and glutathione S-transferase A2 (de Lima et al., 2025). The incidence of this complication is significantly elevated in patients receiving high doses, those with renal impairment leading to drug accumulation, and individuals with preexisting central nervous system disorders (Miller et al., 2011). Furthermore, similar to other β-lactam antibiotics, imipenem may induce gastrointestinal disturbances, allergic manifestations such as skin rashes, and occasional eosinophilia (Foong et al., 2016; Hamao et al., 2025; Sivanandy et al., 2024). Notably, although cilastatin is employed to inhibit renal organic anion transporters (OATs) and prevent imipenem-induced nephrotoxicity, imipenem/cilastatin administration has been associated with alkaline urine, polyuria, crystalluria, and elevated plasma levels of urea, creatinine, and uric acid, indicating that potential nephrotoxic risk still requires close monitoring (Huo et al., 2019; Tahri et al., 2018). A more serious long-term challenge is the emergence and dissemination of carbapenem-resistant strains driven by inappropriate and extensive use, particularly when drug exposure remains at subtherapeutic levels for prolonged periods, exerting substantial selective pressure and posing a significant threat to global public health (Lechtig-Wasserman et al., 2021).

Given the complexities associated with its clinical application, the pharmacokinetic (PK) profile of imipenem is highly important. This antibiotic exhibits time-dependent bactericidal activity, the efficacy of which is commonly quantified by %fT_>MIC, the percentage of the dosing interval during which plasma concentrations exceed the MIC (Fratoni et al., 2023; Khan et al., 2025). Accordingly, a precise understanding of its PK behavior and the maintenance of adequate drug exposure duration are essential for optimizing therapeutic outcomes and preventing the emergence of resistance. In clinical practice, significant interindividual and intraindividual variability in imipenem PK has been observed (Zou et al., 2019). This variability may be influenced by multiple factors, including patient age, body weight (BW), renal function, comorbid conditions, and drug‒drug interactions. For example, impaired renal function can lead to reduced drug clearance and subsequent drug accumulation (Bricheux et al., 2018). However, in patients with obesity, the volume of drug distribution tends to increase, potentially necessitating adjustments to the dosing regimen to maintain effective plasma drug concentrations (Chen et al., 2024). Furthermore, variations in drug concentrations across different infection sites can impact treatment efficacy (Alikhani et al., 2025). Such PK variability presents substantial challenges for appropriate clinical use, highlighting the urgent need to develop individualized dosing strategies that ensure both efficacy and safety.

To address these challenges, population pharmacokinetic (popPK) modeling offers a valuable methodological approach. PopPK utilizes sparsely collected data from routine clinical settings and applies nonlinear mixed-effects models to simultaneously estimate typical population parameter values (fixed effects) as well as interindividual and residual variability (random effects) (Yao et al., 2025). This enables a deeper understanding of the key factors driving PK differences within specific patient subgroups. Although many studies have been conducted to develop popPK models for imipenem, these investigations are dispersed across diverse patient populations. The final model structures and the significant covariates identified often vary, and a systematic integration and comparison are currently lacking. Therefore, this review aims to comprehensively synthesize and systematically evaluate published popPK studies on imipenem. It focuses on assessing and analyzing the structural model employed, the key covariates identified along with their quantitative influences, and comparing the application and validation of these models across various special populations.

Imipenem is a beta-lactam carbapenem antibiotic characterized by its broad-spectrum antibacterial activity and high stability against beta-lactamases. It is frequently employed as a critical therapeutic agent for severe infections caused by multidrug-resistant pathogens (Armin et al., 2023; Barbier et al., 2023; Kaya et al., 2022). Given its narrow therapeutic window (high doses of imipenem are prone to induce neurotoxicity and nephrotoxicity), pathogens with increased drug resistance and significant interindividual variability in pharmacokinetics, popPK modeling is essential for its optimal use (Chen et al., 2020a; Huo et al., 2020; Martínez Delgado et al., 2022; Slama, 2008). However, although several popPK studies have been published, the sources of PK variability are unclear. Therefore, this review is the first to synthesize recent advances in imipenem popPK and summarize the covariates that significantly affect imipenem exposure.

This systematic review included a total of 18 studies from various countries, five of which were conducted in China. Notably, only three studies enrolled more than 100 patients, whereas the smallest study included only 10 participants. As popPK models rely on group data to estimate parameters, studies with small samples often lack statistical power, making it difficult to accurately identify and quantify the influence of key covariates on imipenem PK. In addition, small sample sizes may significantly constrain the robustness and generalizability of the developed models (Berisha and Liss, 2024). Restricted sample representation can introduce bias, thereby limiting the model’s capacity to reflect the true variability in pharmacokinetic profiles across real-world patient populations, particularly among those with high heterogeneity. Therefore, larger-scale studies are warranted in the future to increase the reliability of individualized dosing strategies for imipenem.

Among the studies included in this review, NONMEM was the most extensively utilized tool for popPK analysis. The two-compartment model is most frequently employed to characterize the PK of imipenem, which closely aligns with its in vivo distribution properties. Following administration, imipenem rapidly distributes into highly perfused tissues and moderately perfused organs, a phase described by the intercompartment clearance (Q) between the central (V₁) and peripheral (V₂) compartments. This is followed by a comparatively slower elimination phase, governed predominantly by the total CL. The two-compartment model effectively captures the characteristic feature of the drug concentration‒time curve, which exhibits an initial rapid decline (distribution phase) followed by a slower decline (elimination phase), striking an optimal balance between model complexity and biological plausibility.

Although the two-compartment model is most commonly applied, some investigations have explored the use of one- or three-compartment models. The one-compartment model benefits from structural simplicity and fewer parameters, facilitating easier fitting and convergence. However, its major limitation lies in the inability to accurately depict the pronounced distribution phase of imipenem, often resulting in underestimation of early plasma concentrations and potential bias in AUC estimation. In contrast, a three-compartment model could theoretically offer a more refined characterization of the slow distribution phase into deep tissues, such as adipose or poorly perfused regions, potentially enabling a more precise depiction of the in vivo processes (van de Bool et al., 2015). This advantage becomes particularly relevant in studies investigating imipenem PK in nonplasma compartments, for example, peritoneal fluid (Ikawa et al., 2008) and the dialysate of CRRT (Li et al., 2020). By employing a three-compartment model, researchers are better able to quantify drug disposition in these specific biological compartments, which often exhibit kinetics distinct from both central and shallow peripheral compartments. Nevertheless, this model demands extensively rich sampling data for reliable identification of all the parameters. Given the typically sparse nature of clinical data, the three-compartment model often exhibits unstable convergence, high uncertainty in parameter estimates, and a tendency for overfitting (Rambiritch et al., 2016).

Substantial interindividual variability in PK parameters was observed across studies utilizing a two-compartment model (CL: 4.79–15.31 L/h; V₁: 7.2–32.67 L; V₂: 2.9–41.23 L; Q: 0.996–24.3 L/h). Some studies focused on the same clinical group, but the intra-group variability in PK parameters was also large. For example, the CL in adult critical ill patients varied from 4.79 to 14.6 L/h. This heterogeneity can be attributed primarily to the considerable pathophysiological diversity among the investigated patient populations, which included critically ill subjects with or without ECMO support, elderly patients, children with hematological malignancies, burn patients with or without receiving CVVH, and critically ill patients with suspected ventilator-associated pneumonia. These groups differ markedly in terms of fluid balance, organ function, and hemodynamic status. Notably, the highest CL value (15.31 L/h) in two-compartment model was reported in burn patients with or without CVVH (Por et al., 2021). This elevation is explained by the hyperdynamic circulatory state characteristic of major burns, which involves increased cardiac output and enhanced renal blood flow, thereby accelerating the elimination of imipenem from the kidney (Stanojcic et al., 2018). Furthermore, CVVH contributes to drug elimination by providing an additional clearance pathway (Corona et al., 2022). Similarly, markedly enlarged volumes are frequently documented in critically ill patients, particularly those with sepsis or severe burns (Dickinson and Kollef, 2011; Pruskowski, 2020; Roberts and Lipman, 2009). This phenomenon is associated with capillary leakage, tissue edema, and expansion of the extracellular fluid volume (Roberts and Lipman, 2009). ECMO support may also further increase the apparent distribution volume due to drug adsorption to the circuit (Ahsman et al., 2010; Jelliffe, 2016). It is important to identify and successfully incorporate covariate that can explain the inter-individual variability in the final model.

The majority of studies included in this review predominantly involved critically ill patients. This focus can be attributed to the complex clinical presentations, rapid disease progression, and high mortality rates observed in this population, which consequently make them a priority for clinical intervention and research (Reignier et al., 2025). However, other patient groups also warrant considerable investigation, particularly elderly, infant, and child patients. Elderly patients often present with multiple comorbidities, declining physiological function, and altered PK profiles, predisposing them to adverse outcomes and complicated clinical courses (Medellín-Garibay et al., 2022). Neo et al. reported that carbapenems induced seizures in 2.4% of elderly patients, a prevalence substantially higher than the 0.2%–0.7% range documented in the literature (Neo et al., 2020). In contrast, neonates and children are characterized by ongoing growth and development, resulting in significant differences from adults in terms of organ function, immune status, and drug response. These distinctions lead to unique disease manifestations, therapeutic requirements, and prognostic features. In support of this, a study by Pevzner et al. demonstrated that imipenem/cilastatin was associated with more pronounced nephrotoxicity in neonates, underscoring the need for greater caution in antibiotic selection and dosing in this vulnerable group (Pevzner et al., 2023). Despite the urgent clinical needs in these specific populations, only a limited number of popPK studies targeting these populations were identified in the present review. This indicates a significant research gap, highlighting the necessity for future investigations to prioritize the pharmacokinetics of imipenem in these distinct patient subgroups.

Renal function is known to be the primary determinant of imipenem clearance (Gorham et al., 2022). Consequently, popPK analyses are needed to quantify the impact of renal impairment and other covariates in populations with varying degrees of kidney function. Renal function can be estimated by different equations, commonly CLcr by Cockcroft-Gault equation, eGFR by MDRD and CKD-EPI equations. However, these equations differ in accuracy, especially for special populations like the critically ill, obese and elderly. Among the studies included in this review, CLcr was identified as a statistically significant covariate in ten investigations, eGFR in three, and Scr in one. This distribution reflects the prevalent utilization of CLcr as a routine biomarker for renal function assessment and dosage individualization in clinical practice. When renal function decreases and glomerular filtration capacity becomes impaired, Scr, a waste product of muscle metabolism, cannot be effectively eliminated. Consequently, elevated Scr levels directly indicate diminished renal filtration function. However, Scr is considered a relatively delayed indicator because of the considerable functional reserve of the kidneys; a significant reduction in the GFR must occur before Scr increases noticeably (Romano et al., 2013; Yoo et al., 2019). Furthermore, Scr levels are influenced by factors such as age, sex, muscle mass, and diet (Levey et al., 1999). Therefore, Scr is commonly employed in conjunction with age, sex, and ethnicity for estimating the GFR or CLcr via established equations to better reflect renal functional status in clinical settings (Brown et al., 2013; Inker et al., 2021; Levey et al., 2006; Winter et al., 2012). Recently, Mitton et al. reported that the GFR estimated via the CKD-EPI equation is not related to the plasma level of imipenem in critically ill patients (Mitton et al., 2022). This indicated that the accuracy in estimating renal function of different formulations may introduce bias in covariate effect estimation, and further influence the performance of final model and accuracy of dosing recommendation. It suggests that the plasma concentration of imipenem in critically ill patients cannot be predicted solely on the basis of GFR and that therapeutic drug monitoring (TDM) is a safe and effective approach to ensure precision dosing. Additionally, future studies should include more samples and covariates to facilitate the development of a more precise imipenem popPK model.

In the study by Li et al., conventional renal biomarkers such as CLcr, GFR, and Scr were not identified as statistically significant covariates (Li and Xie, 2019). This observation is explained by the fact that their research focused on critically ill patients undergoing CRRT. In this population, drug clearance is substantially influenced by extracorporeal support, thereby limiting the ability of traditional renal function indicators to accurately reflect the actual elimination rate of medications. Alternatively, the model developed by Li et al. (2020) incorporates residual diuresis as a meaningful variable. The clinical relevance of utilizing residual diuresis for dose adjustment has been supported by previous PK investigations. For example, Yu et al. introduced residual diuresis as a significant covariate when developing a popPK model for vancomycin in critically ill patients receiving CRRT (Yu et al., 2023). Similarly, Ulldemolins et al. demonstrated its considerable impact on optimizing meropenem dosing regimens in a comparable patient population (Ulldemolins et al., 2015). Collectively, these findings underscore that residual diuresis serves as an essential and nonnegligible parameter for guiding individualized drug therapy in patients receiving CRRT.

BW was identified as another key covariate characterizing imipenem distribution and elimination and was incorporated into the final model in six of the reviewed studies. For hydrophilic antibacterial agents such as imipenem, PK behavior is closely associated with extracellular fluid volume and renal function, and BW serves as a fundamental metric for estimating body composition and normalizing renal function (Alobaid et al., 2016; Meng et al., 2017). This parameter is particularly relevant in pediatric populations or patients with significant weight fluctuations, where it forms the basis for individualized dosing regimens (Gade et al., 2018; Natale et al., 2017). Consequently, it is frequently included in pharmacokinetic models to increase its predictive accuracy.

In addition to BW and renal function markers, age is commonly considered a demographic covariate, although its significance varies across populations (Sanghavi et al., 2024). This review revealed that age is a meaningful covariate in models involving children (Dong et al., 2019) and neonates (Dao et al., 2022). During early development, age serves as a surrogate for dynamic changes in body size, body composition, and renal function. Keij et al. provided age-specific dose recommendations for pooled popPK studies of intravenous and oral amoxicillin in neonates (Keij et al., 2023). In contrast, the influence of age is often supplanted by more direct physiological or biochemical indicators. Therefore, most adult popPK models do not retain age as a significant covariate.

Notably, most of the included studies relied solely on internal validation, such as bootstrapping, GOF plots, and VPC, whereas only three investigations performed external validation (Por et al., 2021; Truong et al., 2025; Wang et al., 2024). If only internal data were used for validation, the good predictive performance of the model would only be reflected in its own center, making its generalizability to other patient groups uncertain. Therefore, future research should prioritize rigorous external evaluation of these models through multicenter data to verify their predictive ability across diverse clinical settings.

In conclusion, of the 18 studies systematically evaluated in this review, popPK models for imipenem have been successfully established across different patient subpopulations. A two-compartment model was predominantly employed to characterize the imipenem popPK model. Notably, markers of renal function were consistently identified as the most significant covariates influencing imipenem exposure, which aligns with imipenem’s primary renal elimination. BW and patient age were also demonstrated to substantially impact PK parameters, necessitating their consideration during therapy. Future research should focus on quantifying the effects of under investigated covariates. Importantly, rigorous external validation was conducted to verify the predictive robustness and general applicability of these models in diverse clinical environments.