The increasing emergence of bacterial strains with high VAN MICs (BS_H–VAN–M), such as Staphylococcus aureus, methicillin-resistant Staphylococcus aureus (MRSA), and Enterococcus faecium (The Micron Group, 2018; European Committee on Antimicrobial Susceptibility Testing (EUCAST), 2020), which are defined as any isolate displaying VAN MICs of ≥2 mg/L in the broth microdilution or of ≥1.5 mg/L in the E-test (Song et al., 2017), has posed significant public health threats and must be urgently addressed according to the antimicrobial resistance data released by the World Health Organization (WHO, 2017; Tacconelli et al., 2018). Infections due to these isolates not only cause growing concern that VAN is becoming disadvantageous (Sakoulas et al., 2004; Mavros et al., 2012) but also have severely limited treatment options.

Currently, some traditional drugs (e.g., daptomycin, linezolid, quinupristin-dalfopristin, tetracyclines, and chloramphenicol) and several newly approved agents (e.g., tedizolid, telavancin, oritavancin, and dalbavancin) for VAN-resistant Enterococcus (Barber et al., 2015), together with ceftaroline and tigecycline for MRSA (Gould, 2013), in monotherapy exhibit good activity. Unfortunately, however, they are limited due to either their significant shortcomings (Barber et al., 2015) [e.g., treatment-emergent side effects and drug interactions for daptomycin and linezolid (notably daptomycin) despite being well studied, higher morbidity and mortality and less tolerability for quinupristin-dalfopristin, and lack of sufficient documentation for tetracyclines and chloramphenicol] or the lack of sufficient clinical data. Worse, these drugs considered to be effective for BS_H–VAN–M are still not available even in tertiary hospitals in some regions (e.g., our hospital). Likewise, combination therapies of daptomycin or VAN with a β-lactam, or daptomycin with trimethoprim-sulfamethoxazole, or linezolid with daptomycin (or VAN or gentamicin or fosfomycin), which are considered synergistic for BS_H–VAN–M and may eventually prove to be more effective than monotherapy, particularly in “salvage” situations (Barber et al., 2015; Holubar et al., 2016; Chen et al., 2020), may also be limited since (1) the synergy of these combinations were derived mainly from retrospective studies or in vitro model, resulting in a lack of evidence of prospective randomized trials for a definitive answer on the synergy; and importantly (2) unavailability of daptomycin and linezolid in some hospitals is still an important factor that prevents this form of therapy. Antimicrobials, even novel and effective drugs, with a difficult balance of safety, efficacy, availability and other potential issues (e.g., cost and medical insurance protest), have often forced clinicians to rely on alternative options. Consequently, uncertainties remain for selecting the optimal treatment for BS_H–VAN–M infections.

If the dilemma without better options in the presence of BS_H–VAN–M infections can be broken through by optimizing the administration of the available drugs which have a currently reduced competency for such infections, it must be helpful for clinicians to treat such infections. Surprisingly, studies have shown that meropenem, with administration optimization, can still achieve sufficient PK/PD exposure against highly resistant bacterial isolates (Song et al., 2019), which has aroused substantial interest in exploring the “new usage for old drugs” to improve and maximize the performance of VAN against BS_H–VAN–M infections. If successful, this exploration will have important significance in clinic, especially considering the rapid progress against the MIC creep phenomenon, delays in the development of new alternatives, and lack of better treatment options. Due to the lack of outcome data, the updated guidelines regarding VAN therapy issued in 2020 by the American Society of Health-System Pharmacists (called the 2020 VAN guidelines) (Rybak et al., 2020b) encourage investigators to ascertain whether VAN can provide sufficient PK/PD exposure for the isolates despite their having an MIC susceptibility breakpoint as low as 2 mg/L. Given that positive studies on these isolates have currently rarely been performed and few attempts have been made to mine the competence of VAN against the increasing number of BS_H–VAN–M with MICs of ≥2 mg/L, this study aimed to construct a novel dosing strategy based on PK/PD modeling to exploit the maximum potentialities of VAN in the management of infections, especially for those resulting from isolates with MICs of ≥2 mg/L. We hope that our method will resolve the disadvantages of VAN in the growing presence of MIC creeping, especially in the absence of better alternatives.

To our knowledge, this study is the first to design OTSI method to exploit the maximum potentialities of VAN. Our data suggests great superiority and satisfactory PK/PD exposure of the dosing strategy of OTSI relative to that of TIII, especially for BS_H–VAN–M of ≥2 mg/L and ≤4 mg/L, supporting that VAN in OTSI mode is still powerful for BS_H–VAN–M of ≤4 mg/L and thus providing a temporary solution when better treatment options are unavailable.

Currently, it is still difficult to find an antimicrobial with a desired balance of safety, efficacy, availability, cost, and other potential issues for BS_H–VAN–M despite the emergence of some current novel drugs (Gould, 2013; Barber et al., 2015). Importantly, the lack of clinical trials and experimental studies evaluating antimicrobial efficacy against BS_H–VAN–M has forced clinicians to rely on alternative regimens extrapolated from PK/PD models, retrospective studies, and case reports (Yim et al., 2017). Currently, clinical data on the continued use of VAN in the setting of known BS_H–VAN–M are quite rare. Some clinical and simulated studies have focused mainly on MRSA isolates with an MIC of 2 mg/L (still within the susceptibility range).

Clinically, despite the achievement of a target C_trough of 15 mg/L, VAN performed worse against MRSA bacteremia, intraabdominal infections and pneumonia caused by isolates with an MIC of 2 mg/L than against those with an MIC < 2 mg/L (Moise-Broder et al., 2004b; Sakoulas et al., 2004; Hidayat et al., 2006). Harigaya et al. (2009) demonstrated that VAN did not have bactericidal activity against MRSA pneumonia due to infection by isolates with an MIC of 1 mg/L regardless of whether 1–1.5 g of VAN was administered twice daily (i.e., 2–3 g/day) and of the assumptions of 100% VAN penetration in the lung and an AUC_0–24/MIC of 350 as the therapeutic target. Likewise, for MRSA meningitis due to strains with MICs of 2 mg/L, Sipahi et al. (2013) found that six (five of the six had MRSA isolates with a VAN MIC of 2 mg/L) out of eight patients (75%) with MRSA meningitis who received 500 mg VAN every 6 h (i.e., 2 g/day) as a 1 h infusion failed to achieve the clinical outcome, suggesting that VAN, at its standard daily dose and conventional dosing strategy, is unsatisfactory for such infections. Interestingly, our data indicated that even for bacteremia due to isolates with MICs of 2 mg/L, VAN at the standard 2 g/day can still produce adequate PK/PD exposure when administered using the regimen of 1.2 g LRRI + 0.8 g LRCI, 1.8 g LRRI + 0.2 g LRCI, or 1.95 g LRRI + 0.05 g LRCI, while worse outcomes were achieved when TIII was used; these results suggest the competence of VAN for such infections and the superiority of OTSI relative to TIII. Consistently, however, for pneumonia and meningitis due to isolates with MICs of 2 mg/L, VAN performed poorly (well only for those resulting from isolates with MICs of ≤1 mg/L) even at 6 g/day and with the OTSI strategy. The occurrence of this outcome may be due to the fact that VAN has poor profiles in terms of lung interstitial fluid and cerebrospinal fluid penetration. More recently, a review research for the penetration of VAN (15 mg/kg IV 1 h) into bone showed an average VAN concentration of 3.8 mg/L and 4.5 mg/L in cancellous and cortical bone, respectively (Thabit et al., 2019), suggesting that VAN reaches concentrations that exceed the MIC₉₀ against S. aureus (1 mg/L) and thus implying that VAN will have good performance on bone infections due to such bacteria. However, in a case report, Lee et al. (2016) found that the infusion of even 1 g of VAN over 1 h every 12 h (i.e., 2 g/day) was unsuccessful for an infected patient who had lumbar osteomyelitis, discitis, and epidural abscess with persistent MRSA bacteremia resulted from isolates with an MIC of 1 mg/L. The occurrence of this result may be associated with the physiological and pathological status of the patient, slow bactericidal activity of VAN, inappropriate dosage regimen of VAN or a combination. However, our data showed that even though relatively poor bone penetration coefficient (0.57) of VAN was applied, 2 g/day VAN with a rational OTSI strategy (e.g., 1.2 g LRRI + 0.8 g LRCI) could still yield desirable PK/PD exposure for this complicated infection, and 4 g/day VAN with a rational OTSI regimen (e.g., 2 g LRRI + 2 g LRCI) performed well for this case due to isolates with an MIC of up to 2 mg/L. This implied that for population in which VAN has good profiles of bone penetration, these optimal dosage regimens will show better PK/PD exposure performance, and this was also the case for VAN against other infected tissues. Unexpectedly, for bacterial pericarditis, 2 g/day VAN with a rational OTSI (e.g., 1.2 g LRRI + 0.8 g LRCI) achieved optimal PK/PD exposure for isolates with an MIC of up to 4 mg/L, indicating good activity at infected sites with strong VAN penetration and enrichment (e.g., the pericardium).

Likewise, in the simulated studies, poor VAN exposure at an MIC of 2 mg/L was also observed with a PTA of 90% as an acceptable PK/PD exposure target. By MCSs, Mohr et al. (Mohr and Murray, 2007) determined the PTAs for the targeted AUC_0–24/MIC of ≥400 using the PK data derived from the study conducted by Jeffres et al. (2006) and indicated that the PTA would be 100% at an MIC of 0.5 mg/L but 0% at an MIC of 2 mg/L despite that a high dose of VAN (i.e., C_trough > 15 mg/L but no detailed information on the simulated regimens) was used. Coincidentally, using MCSs, del Mar Fernández de Gatta Garcia et al. (2007) reported that 3–4 g/day VAN administered via the TIII strategy for an assumed MIC of 1 mg/L would reach a PTA of 90% at the targeted AUC_0–24/MIC of 400, thus questioning the dosage of 2 g/day as a standard schedule for such isolates. Interestingly, Setiawan et al. (2019) found that 2 g/day VAN with TIII obtained a PTA of 100% at an MIC of 0.5 mg/L but only 84.41% at an MIC of 1 mg/L; moreover, even 4 g/day VAN administered with TIII failed to achieve the optimal PTA for an MIC of 2 mg/L. Conversely, however, our data supported that at the above PK/PD target, 2 g/day VAN, with the regimen of 1.2 g LRRI + 0.8 g LRCI, 1.8 g LRRI + 0.2 g LRCI, or 1.95 g LRRI + 0.05 g LRCI, still produced a PTA of ≥90% at an MIC of 2 mg/L, suggesting the achievement of sufficient PK/PD exposure. Surprisingly, 4 g/day VAN, with the regimen of 2.0 g LRRI + 2.0 g LRCI, yielded a PTA of ≥90% (91.28%) for an MIC of up to 4 mg/L but failed with the regimen of 4 g/day administered with TIII. In addition, when administered via TIII and even at a high dose of 5 g/day, VAN cannot provide sufficient PK/PD exposure for Enterococcus isolates with MICs of 4 mg/L regardless of the infection, although these isolates are currently considered to be susceptible to VAN at this breakpoint. These findings show the superiority and feasibility of OTSI for the competence mining of VAN against BS_H–VAN–M. Of note, the achievement of such regimens for these MICs might be more agreeable for bacteremia, mastitis and pleuritis based on VAN tissue permeability and the derivation of the AUC_0–24/MIC target of 400 [derived mainly from MRSA bloodstream infections (Rybak et al., 2020b)].

However, the exact MIC values are often unavailable in empirical therapy. Fortunately, our data provided the CFRs for the targeted bacterial populations. In empirical therapy, 3–4 g/day VAN with an OTSI strategy, such as the preferred regimen of 2 g LRRI + 1 g LRCI or 2 g LRRI + 2 g LRCI, may be preferable for most common infections, including bacterial pericarditis, mastitis, pleuritis, bacteremia and sternal osteomyelitis, as these regimens produced a CFR of ≥90% for all of the targeted bacterial populations, including E. faecalis, S. aureus, S. epidermidis, and S. bovis, with the sole exception of E. faecium. However, for pneumonia and meningitis, even at the aggressive regimen of 6 g/day (2 g LRRI + 4 g LRCI), VAN was sufficient for infections due to only S. aureus and S. bovis; in theory, for such pathogen populations, the standard 2 g/day may result in similar PK/PD exposure and reduced nephrotoxicity relative to that at 6 g/day based on our data. Table 5 summarizes some preferred VAN regimens for common infections based on our analysis.

Despite acquisition of target AUC_0–24/MIC by OTSI method, we were unable to demonstrate a microbiological or clinical superiority of these recommended VAN OTSI regimens based on MCSs as little data are currently available, either in vivo or in vitro experiments. However, in a study of in vitro pharmacodynamic model and MCSs (Eguchi et al., 2010), experimental verification of the efficacy of optimized two-step infusion therapy (OTIT), for meropenem but not for VAN, was performed. Surprisingly, the in vitro bactericidal effect of the optimal meropenem OTIT regimens is consistent with the bactericidal effect predicted by their respective PTA of these regimens derived from MCSs, especially for Pseudomonas aeruginosa isolates with meropenem MIC of ≤4 mg/L, which partly indicated the feasibility of MCSs method in predicting the efficacy of the regimens although the research object of this study was meropenem rather than VAN. However, despite this significative case of MCSs prediction, the recommended VAN OTSI regimens based on MCSs cannot be considered certainly effective against the target bacteria given the difference in action profiles among antibiotics and in resistance mechanisms among bacteria. In addition, modification of VAN delivery in severe infections may not be sufficient, by itself, to change the clinical outcome for critically ill patients, and the bacterial status like bacterial biofilm widely formed among various bacteria were also a barrier to effective anti-infection.

This study has some limitations. First, the desired regimens acquired by MCSs were not validated by experimental outcomes, which somewhat limits their generalization. However, further studies on this topic will be summarized in our next study. Second, the AUC_0–24/MIC target of >600 (e.g., 667 and 1,000) utilized in this study may be too high to increase VAN-induced nephrotoxicity. Regarding this, we need to correctly understand (1) the derivation of the value of 600, and (2) that these targets simulated herein are just used to simulate concentrations in sites other than the plasma (e.g., lung interstitial fluid and cerebrospinal fluid). Moreover, the dosing interval and a C_trough value of 20 mg/L designed for OTSI method ensure a theoretically low probability of drug accumulation; notwithstanding this optimal design, however, therapeutic monitoring of VAN should be recommended as a routine test, especially for patients with long medication time. Third, in consideration of the problem of VAN stability and low flow rate (especially in the LRCI phase), some experts consider that OTSI may not be implemented well clinically. However, the concern on the VAN stability is not a factor preventing the implementation of OTSI method since VAN displayed well chemical stability even during 24-h infusion (Masse et al., 2020). And, the low flow rate can be performed well with the recently developed microcomputer pumping method. Moreover, we can appropriately increase the total solvent volume of VAN to improve the velocity problem. Fourth, the PK models selected in this study may have been biased for assessing target attainment in different populations. Indeed, the inherent interindividual variability in the PK data contributes to the prediction bias. However, we believe that the use of these PK models together with PK/PD analysis based on MCSs still offers a definitive outcome to some extent since (1) the PK variability, which is often extensive, was considered by the lognormal distribution patterns settings of CL_van and V_d in MCSs, and (2) the PK models used in this study performed well with satisfactory precision and bias prediction errors when compared with some established models in the external validation evaluation (Sánchez et al., 2010). Finally, at the same PK/PD target, whether regimens with desirable PK/PD exposure to MRSA are also satisfactory against Enterococcus and whether regimens with acceptable PK/PD exposure for ordinary patients are also adequate for critically ill patients remains unknown, thereby limiting the extrapolation of optimal regimens to other BS_H–VAN–M with different resistance mechanisms and to critically ill patients with marked PK variability. Notwithstanding its limitations, the study provides some tentative options for BS_H–VAN–M, especially in the absence of better alternatives. However, prospective validation of these limitations is desirable.

When faced with the daily challenge of infections due to isolates with high MICs, we should try to reduce the gap between the available medical evidence for using VAN and the dearth of alternative options, some of which have not been sufficiently explored and/or whose efficacy in certain situations remains debatable. The question of whether VAN can continue to be used for isolates with MICs of ≥2 mg/L (this study supports VAN, at an allowable daily dose and with a reasonable OTSI, for isolates with an MIC of up to 4 mg/L) is still not obsolete, and this study provides a new resource for understanding how PK/PD modeling shapes the power of VAN to meet the growing challenges of BS_H–VAN–M infections. However, in the absence of control trials, the continued appraisal of VAN for use, along with the optimal dosage regimens in clinical experience, will provide additional important information on its utility against BS_H–VAN–M.

The original contributions presented in the study are included in the article/supplementary material, further inquiries can be directed to the corresponding author.

XS performed the model simulation and wrote the manuscript. MZ, YW, and YP conceptualized and supervised the manuscript. All authors approved the final version of the manuscript.

The authors declare that the research was conducted in the absence of any commercial or financial relationships that could be construed as a potential conflict of interest.