Carbapenem-resistant K. pneumoniae is an increasingly prevalent pathogen which poses serious threat to immunocompromised hosts, particularly patients with hematologic malignancies(3, 8, 12, 13). Infections especially bloodstream infection caused by CRKP can cause significantly high mortality, for the reason that it expresses multidrug resistance phenotypes so that antimicrobial treatments appear to be quite intractable. In our study, the 28-day mortality of CRKP BSI among patients with hematologic malignancies reached 77.3%. It is higher compared to that reported in previous works, in which mortality rates ranged from 50–60%(7–10). Accurate detection of high-risk CRKP BSI patients and timely implementation of infection control to contain them play a crucial role in better survival.
As is evident from the Table 2, the median age of the non-survival group was older than that of the survival group. PBS > 4, pneumonia and septic shock were more common in the non-survival group. All of the survivors received appropriate initial anti-infection therapy, while only 11(64.7%) patients in the non-survival group. A further study of the above 11 patients found that all had experienced severe neutropenia after the occurrence of BSI. Based on this information, we speculated that advanced age, PBS > 4, severe neutropenia, pneumonia and septic shock were death-associated risk factors, while appropriate initial anti-infection therapy is protective against mortality. Our results have been proved by Trecarichi et al(9, 10, 14–19). In this study, these CRKP isolates from the patients with hematologic malignancies showed high resistance to most commonly used antimicrobial agents except tigecycline and polymyxin B, which was in accordance with the results of previous reports(2). Figure 4 showed that the antibiotic resistance rates to several agents (fluoroquinolones, monocyclic amides, aminoglycosides, sulfonamides, tigecycline) of the non-survival group were all significantly higher than that of the survival group, which may directly lead to the limited option of active antibiotics for patients infected with such pathogens and indirectly led to the death. Figure 2(b) revealed that tigecycline combined with polymyxin B could improve survival. The optimal treatment for invasive infections caused by CRKP has not yet been established(10). But on the basis of observational studies, combination therapy that includes > 1 drug with vitro activity against the isolates is significantly associated with survival(10, 19–21). Mario Tumbarello reinforced concerns that monotherapy regimens utilizing drugs with substantial potency or pharmacokinetic shortcomings for BSIs may be associated with increased morality(19).
When is appropriate time to empirically combat CRKP in patients undergoing hematologic malignancies remains an unavoidable difficult problem. Given the limited therapeutic options, prevention of CRKP infection in this vulnerable population is of paramount importance. Recently, several studies concerning risk factors for CRKP BSI have reported mostly in regard to CRKP carriers and patients with immunosuppression or in the ICU. There also have been similar studies in patients with hematologic malignancies, but the number of participants included is relatively small and there has not risk prediction score for it yet. The findings of our study showed that CRKP rectal colonization, severe neutropenia and IMV were independent risk factors for CRKP BSI in patients with hematologic malignancies. In accordance with the present results, previous retrospective clinical studies have demonstrated that CRKP rectal colonization and mechanical ventilation are vital prediction factors for CRKP BSI(22, 23). Malignancy hematologic patients whose immune defense function is impaired due to use systemic immunosuppressants(24), intestinal microecology is out of balance caused by exposure to a wide range of antibiotics and intestinal mucosa is damaged by chemotherapy or acute GVHD(25, 26), are susceptible to develop BSI because these factors make pathogenic bacteria implanted in intestine easy to break through mucosa barrier into blood escaping from effective containment and removal. The CRKP rectal colonization rate was 7.4% in our study, significantly higher than that reported by Cattaneo et al(8, 27), confirming that this is a growing problem in endemic areas of our country and great attention must be paid to it. IMV makes it vulnerable to the occurrence of ventilator-associated pneumonia, coupled with the breakdown of the blood-gas barrier, pathogens are easy to pass through the barrier into blood stream. In addition, previous researches have revealed that urinary catheterization, admission to ICU, having a CVC, prior receipt of antibiotics, chemotherapy/radiation therapy, renal diseases are also independent risk factors for CRKP BSI(22, 28–30). The discrepancies between our results and these findings might be due to different experimental designs, basic characteristics of the study population, sample sizes or statistical methods. In this study, we targeted on the population with hematologic malignancies, so some factors might have no statistical significance and need to be further verified.
CRKP BSI results from a combination of factors, so the establishment of relevant prediction model should be based on the screening of high-risk factors. This retrospective study analyzed more than 40 clinicopathological factors that might be associated with CRKP BSI. A risk prediction equation was built so as to judge the BSI risk rapidly in the clinical and implement early intervention. We found that the prediction probability of CRKP BSI was 48%, 86% and 94% in patients with total risk score of 5,7and 8 respectively. It has important therapeutic implication if the risk score ≥ 5. The score effectively stratified the risk of CRKP BSI, demonstrated by an AUROC of 0.753(95% CI = 0.648–0.858) which was higher than using CRKP rectal colonization alone (AUROC = 0.673; 95% CI = 0.537–0.808).
There were several limitations of this study that should be mentioned. First, it was a single-center retrospective study, therefore, the results may not necessarily be applicable to other settings; there might be cases of CRKP BSI with negative blood culture results were not included for analysis. Second, the date when the first positive blood culture was drawn was considered as the onset of BSI, it might be inconsistent with the real situation which may have happened earlier and thus affected the outcomes. Third, although our prediction model has a good fit and predictive ability recognized separately by Hosmer-Lemeshow test and ROC curve, our study did not perform external validation of the prediction model which is one of our goals for further study.