Since the turn of the twenty-first century, the reports emerged indicating that GNB have gained importance again but this time with increased resistance to available broad-spectrum agents [24, 25, 29, 31, 34, 38, 41, 44, 47, 50, 67, 88, 98, 104, 105, 152, 154, 158, 198]. The latest published trial of EORTC-IATG including high-risk patients [54] found that 47 % of all bacteremic episodes were due to GNB. Whereas in the two recent EORTC trials [107, 109] in which patients were considered having low-risk neutropenia, the incidence of Gram-negative bacteremias rose to 59 and 50 %, respectively. But one should note that these trials included more patients with solid tumors and the frequencies of patients with acute leukemia were 32 and 51 %, respectively. Reemergence of Gram-negative bacterial infections has also been reported from other cancer centers [92, 152, 154]: Collin et al. [53] observed a decline in ratio of GPB/GNB from 2.7 to 2.3 between 1991 and 1997. Similarly, Mikulska et al. [139] reported from a single center in Italy with patients undergoing HSCTs that the ratio of GPB/GNB decreased from 2.4 in 2004 to 1 in 2007 (p = .043). Unpublished data from the most recent European Conference on Infections in Leukaemia (ECIL) group also indicated that several European cancer centers recently observed a similar increase of GNB in neutropenic cancer patients; however, GPB are still a majority in bloodstream infections (BSIs) [140].

If one compares epidemiology between different countries, in general, US centers reported that GPB are still the predominant microorganisms in cancer patients [210]. Elsewhere, many surveillance studies provided data that the incidence of GNB has become numerically higher than GPB in hematological cancer or hematopoietic stem cell transplant (HSCT) recipients [8, 41, 86, 88]. Interestingly, GNB dominance or equivalence to GPB was usually reported from outside the United States including studies from Brazil [198], Guatemala [23], Japan [50], and several southeastern Asian [25, 48, 49, 98] and Mediterranean countries [44, 86, 88, 104, 105, 158]. But similar reports from some of the Western European countries have also been available [67, 179]. The reasons for this geographical variation are not clear but may be related to the lack of widespread quinolone prophylaxis in at least some of these centers reporting more infections with GNB [86, 105], and in those still using this type of prophylaxis, emerging fluoroquinolone (FQ)-resistant E. coli bacteremias can be accounted for this epidemiology [44, 108]. Another theory which has originally been proposed about the spread of carbapenemase-producing enteric GNB may also be applicable to cancer patients in those countries where environmental sanitation is poor and resistant Gram-negative pathogens can be found in abundance in water and food [7]. Thus, via oral-fecal route, cancer patients may get colonized in the outpatient settings and when they become neutropenic with compromised gastrointestinal mucosa due to chemotherapy, they can acquire bacteremia with these resistant pathogens. It is also noteworthy that in some of these centers, GNB have been the leading pathogens during the last two decades [54, 55, 91, 110]. In our center in Ankara, we observed a similar picture [88, 105]. In a retrospective evaluation of bacteremic episodes in hematological cancer patients with febrile neutropenia, we analyzed 3,703 neutropenic episode in 2,098 patients between 2005 and 2009 [105]. Overall, single Gram-negative bacteremia were detected in 61 % of these episodes. Except 2005 (incidence of GNB, 48 %), GNB were the main type of microorganisms during all study years (range 58–66 %). A previous study dated back to 1997 reported similar figures [88].

Catheter-related bacteremic episodes in neutropenic patients are usually caused by coagulase-negative staphylococci (CNS) [1]. Gram-negatives, albeit much rare, can also be responsible for this picture [47].