Impact of a Risk Stratification Tool on the Outcome of Liver Transplant Recipients Colonized With Carbapenem Resistant Enterobacteriaceae: an Observational Study

Patients undergoing orthotopic liver transplantation (OLT) are more susceptible to acquire colonization with multidrug resistant Gram-negative bacteria, in particular with carbapenem resistant Enterobacterales (CRE), than non-solid organ transplant (SOT) recipients or patients with other types of SOT. CRE colonization, either acquired before either emerged after OLT, has been associated with highest risk of developing CRE infection after OLT with a dramatic impact on patient survival. For these reasons, along with the improvement of infection control and antimicrobial stewardship policies, preventive strategies targeted to patients colonized with CRE have been suggested. However, several uncertainties have to be considered concerning the optimal timing for intervention, the correct patient stratification, and the choice of drugs. Regarding the timing for intervention, surgical prophylaxis is regarded as an option. However, it should be considered that of the overall burden of CRE carriage in OLT recipients, pre-transplant acquisition accounted for lower than one third of isolations in a large multicenter retrospective study. Furthermore, even when CRE carriage is present at transplantation, the colonization status is frequently recognized in the immediate post-operative period by results of rectal swabs done at surgery. Finally, current data about the efficacy of targeted prophylaxis in patients colonized with CRE are very limited and showed controversial results. Thus, in the majority of cases the unique window for preventive strategies is represented by the post-transplant course. This requires an active screening policy as well as the possibility of stratifying colonized patients according to their risk of developing CRE infection, and eventually to die, in order to target specific interventions in those patients who could most benefit, and to optimize essential resources as the new drugs. Indeed, since the availability of new drugs such as ceftazidime/avibactam, meropenem/vaborbactam and cefiderocol, the therapeutic management and outcome of patients colonized and/or infected with CRE has changed in both general population and SOT recipients. New drugs are associated with higher rates of clinical cure and patient survival, as well as with lower toxicity, compared with old regimens. However, rates of microbiological failure around 10%, consisting in persistent positive cultures or relapsing infections associated with development of further resistance, have been reported raising concerns about the need of optimizing the use of new drugs to avoid the loss of their efficacy.

To this end, the investigators have recently carried out a large multicenter observational study aimed at building a prediction model to stratify patients according to their risk of developing CRE infection after OLT. The study cohort consisted of 800 OLT recipients colonized with CRE, 25% were colonized at OLT, and 75% acquired colonization within two weeks after OLT. Infection rate was of 30% and was similar between the two groups as well as the infection severity. All-cause 6-month mortality rate among patients who developed CRE infection was 58% vs. 20% among carries who did not develop infection (p<0.001). Almost all infections occurred within 60 days after transplant. Thus, the investigators derived, and internally validated, a prediction tool able to stratify the risk of CRE infection at 30 and 60 days after OLT (CRECOOLT score) based on six variables easily to monitor such as CRE colonization detected in the 60 days prior to transplant, CRE colonization detected post-transplant, multisite colonization, need of prolonged (≥48 hours) mechanical ventilation, acute renal failure, and reintervention. The calculated cumulative incidence of CRE infection and nomogram-derived prediction for a low-risk and a high-risk patient were made available as app to be used bed-side. The investigators explored the potential clinical utility of our model using a decision-curve analysis to examine the "net benefit" of applying the prediction model across a range of CRE infection threshold probabilities. A theoretical risk-model guided strategy (i.e. empiric administration of CRE-active antibiotics) was compared against two default strategies- "treat all" and "treat none" suggesting that the model-directed intervention would show net benefit over default strategies when the overall CRE infection risk exceeded 10%.

Finally, the investigators have explored the risk of death using the same variables included in the CRECOOLT score by a multi-state model obtaining three risk groups: low, intermedium and high. The investigators have then estimated the potential impact of using ceftazidime-avibactam in each of such groups finding a significant protective role (HR 0.32) only in patients with intermedium or high risk of dying.