P. aeruginosa, especially those with CR/DTR resistance phenotypes, remains one of the most challenging pathogens to treat globally. The high genome plasticity of the P. aeruginosa complicates the management, necessitating contemporary evidence to support treatment practices for this constantly evolving pathogen20. Singapore’s status as an international travel and medical hub has seen the import and dissemination of antimicrobial resistance to the country. For this reason, the local CRPA epidemiology encompasses well-known epidemic multi-drug-resistant clones amidst a diverse pool of non-epidemic lineages19. Here, we identified potential treatment options for the management of a genomically diverse CRPA cohort, primarily composed of carbapenemase-producing strains, using iACT. Additionally, we shared our real-world clinical experience in treating these challenging infections with a personalised approach.
Based on iACT results, polymyxin-containing combinations exhibited the highest bactericidal rates. Compared to the other test agents, where the agent’s MIC far exceeds the corresponding simulated test concentration, polymyxin MICs observed in CRPA were lower than or equivalent to the simulated concentration (2 mg/L) utilised in iACT21, increasing the probabilities of enhanced bacterial killing when combined with a synergistic partner agent. Therefore, a polymyxin-containing combination might be a reasonable empiric combination prior to the availability of iACT results or where iACT is not available. Unfortunately, most patients afflicted with CRPA infections tend to have relative or absolute contraindications to polymyxins or are unable to tolerate the regimen for the entire treatment duration, which is often prolonged.
Alternative regimens identified in this study included fosfomycin-containing combinations. Specifically, fosfomycin + aztreonam was the polymyxin-sparing regimen that displayed the highest bactericidal rates in our study. This may be related to the epidemiology of CRPA in Singapore, where MBL-producing P. aeruginosa are encountered frequently19. Aztreonam is not appreciably hydrolysed by MBLs, owing to poor and unproductive binding with the enzymes22. The addition of fosfomycin here exerted an additive/synergistic effect, resulting in bactericidal kill. This combination has also shown promise in various other studies where the aztreonam + fosfomycin combination demonstrated collective synergistic/additive rates above 80%, even restoring aztreonam susceptibility for some P. aeruginosa23,24,25.
Given fosfomycin’s and aztreonam’s wide tissue penetration, regimens containing this combination with/without levofloxacin were the fosfomycin-based regimen of choice targeting various infections in this study. IV fosfomycin disodium 8 g every 8 h was the most commonly employed dose in our patient cohort. Unfortunately, this dosing regimen came with the propensity for electrolyte derangement, particularly hypernatremia and hypokalemia, warranting closer monitoring. A good clinical response was observed in most cases utilising the fosfomycin + aztreonam combination in our study. The encouraging results both in vitro and in vivo should prompt further investigation of this combination, especially for the management of MBL-producing CRPA in general.
Apart from polymyxin-containing regimens that were able to target a large proportion of the isolates tested, iACT results suggested that the in vitro activities of antibiotic combinations are highly strain-specific and unpredictable. Corresponding to this strain specificity, we observed that some antibiotic regimens utilised prior to iACT were found to be non-bactericidal (or even non-inhibitory) upon in vitro testing. These antibiotic regimens were empirically selected by the treating physicians based on previous experience and/or anecdotal reports combined with routine antibiotic susceptibility testing results. Prolonged use of these inactive regimens could potentially lead to eventual antibiotic resistance or therapeutic failure26,27.
The iACT service provided several benefits to the treatment of CRPA infections. All patients were eventually treated with bactericidal combinations with optimal pharmacokinetic/pharmacodynamic profiles, leading to a high proportion of patients with clinical response at the end of therapy. The role of a clinical ID pharmacist in this service was critical in therapy optimisation through individualising the antibiotic combination(s) according to patients’ clinical status, co-infections, and co-morbidities with guidance from the iACT results. CRPA infections are often complicated, requiring prolonged therapy durations. In our experience, patients frequently have complex comorbid conditions resulting in turbulent and protracted infection courses. With the iACT service, the patients are monitored throughout their infection episodes, allowing timely personalised therapy optimisation in response to dynamic changes during the infection course (e.g. organ function changes, adverse drug reactions, concomitant infections, etc). Furthermore, the iACT encompassed >100 different combinations, greatly enhancing the treatment availability and flexibility in responding to these changes. This is accompanied by therapeutic drug monitoring, where available, to ensure that dose adjustments result in adequate antibiotic levels at infection sites.
Simultaneous testing of multiple varied combinations has also allowed the majority of patients to be treated with polymyxin-sparing regimens. Polymyxins are generally disfavoured by clinicians in view of the well-known toxicity profiles of the agent in our cohort and its poor tissue penetration28,29. Prolonged treatment durations exceeding 2 weeks, particularly in those with poor underlying kidney function, increase the risk of polymyxin-associated adverse drug reactions30. In this study, iACT has allowed physicians to minimise polymyxin duration with the use of bactericidal non-polymyxin-containing regimens where activity was confirmed by the test, resulting in the avoidance of adverse drug reactions without compromising the clinical outcomes.
Combination therapy has been used as a tool for the prevention or delay of the emergence of resistance31. Although initial clinical response was observed in the majority of the patients, the CRPA re-infection rate was 13%, suggesting the potential emergence of resistance. P. aeruginosa is known to adapt readily to the host environment and can result in persistent colonisation, especially in biofilm-producing strains, leading to recurrent infections32. While antibiotic therapy forms the cornerstone of infection management, inadequate source control and host immune function are integral factors for successful treatment. Poor antibiotic penetration and slow growth within biofilms that may have formed on indwelling devices and in chronic wounds may have impaired both the killing effect of the aggressive combination antibiotic therapy administered, leading to the emergence of resistant phenotypes. This is further complicated by the compromised immune status as seen in several of the cases here. Resistant P. aeruginosa infections remain challenging to manage and may be refractory to even active combination therapy, especially in immunocompromised individuals with indwelling devices/chronic wounds, which can serve as potential sources of recurrent infection. Without the complete removal of such reservoirs, combination therapy can at best suppress the infection temporarily in such a scenario.
The iACT-guided personalised treatment strategy is not without challenges. Current in vitro testing methods rely on viable cell counting, which requires a minimum of 48 h from the time of isolate receipt at the pharmacy laboratory. This is compounded by an inherent time delay involved in pathogen isolation and identification, which may lead to considerable delay in the time of receipt of iACT therapy. Hence, due to current limitations of iACT, such a strategy may be of limited benefit to critically-ill patients who require immediate active therapy. Nevertheless, the data generated in this study have identified selected combinations with a high proportion of bactericidal activity against our local CRPA clones, which could be used to inform the empiric treatment of severely ill patients while awaiting test results. This treatment strategy is also relatively resource-intensive and requires skilled laboratory personnel and specialist-trained pharmacists in the conduct and interpretation of the test. Therefore, the cost and resources associated with this approach may limit its access.
Although the study had multiple promising findings, it had several limitations. Firstly, this was a non-controlled study with recognised inherent biases and a small sample size. It is uncertain if patients receiving iACT-guided personalised treatment fared better compared to those receiving unguided therapy. The favourable clinical outcomes we observed in this study could also be influenced by confounders, including disease severity and host factors such as co-morbidities/immune status and availability of non-antibiotic interventions. In addition, P. aeruginosa can exist as commensals (e.g. on skin, urine, airways), making it difficult to distinguish between colonisation and infection in febrile patients with polymicrobial cultures. At the time of iACT initiation, treatment targeting the implicated bacteria was initiated, leading to clinical improvement due to the susceptibility of the bacteria to the chosen treatment. When iACT results were available, the clinicians deemed P. aeruginosa to bear little clinical significance after assessment, choosing to manage the patients conservatively or continuing the therapy which were not active against CRPA. Test-guided treatment plans were reserved for clinical deterioration. Bias could have been introduced inadvertently through the exclusion of such cases from the outcomes analyses. Furthermore, the exclusion of patients who were unable to receive iACT-guided therapy, such as those who died shortly after test initiation, from the analyses could have inflated the perceived effectiveness of the test. However, the poor outcomes observed in these patients were mostly related to underlying co-morbidities (e.g. malignancies) and were less likely to be contributed by the test itself. This represents the current practical limitation of the test-guided approach, where benefit is limited in critically-ill patients with poor short-term survival due to the time delay to test initiation. A randomised controlled study is currently in progress to establish the differences between test-guided and conventional approaches and to validate the findings in this study. Nevertheless, this is one of the biggest real-world cohorts investigating the in vitro bactericidal activity of an extensive number of combinations against diverse CRPAs, potentially informing empiric treatment of CRPAs. The clinical outcomes of patients with varying disease severity and infection types receiving these bactericidal combinations were prospectively followed up, supporting the feasibility and safety of utilising personalised in vitro test-guided treatment of CRPA infections.
Secondly, the generalisability of the study results to settings with different CRPA molecular epidemiology or to other Gram-negative infections e.g. Enterobacterales/Acinetobacter baumannii may be limited. However, the results here could provide base knowledge for combination activities against CRPA with various different genotypes and resistance mechanisms that could be useful for future investigations.
In conclusion, personalised iACT-guided combination antibiotic treatment achieved positive clinical and microbiological outcomes in our patients, resulting in high response rates and corresponding low mortality. The current study demonstrated the potential of iACT-guided therapy through the selection of bactericidal antimicrobial combinations coupled with individualised dose optimisation using pharmacokinetic/pharmacodynamic principles in combating infections caused by CRPA, even in difficult-to-treat MBL-producing isolates such as the locally circulating ST308 clones. Where iACT is not available, polymyxin combinations are the most reliable in exhibiting bactericidal activity. Polymyxin-sparing regimens, like aztreonam or fosfomycin combinations, are useful to prevent polymyxin-related adverse side-effects if prolonged duration of therapy is expected. The findings in this study should be confirmed with subsequent controlled studies.
