Etiology of Febrile Episodes in Patients With Acute Myeloid Leukemia: Results From the Hema e-Chart Registry FREE

Several studies have attempted to analyze the epidemiologic patterns of infectious complications in patients with acute myeloid leukemia (AML), although the power of these studies was frequently limited by their retrospective design.^1- 2

A lively debate currently focuses on the potential of Internet-based software and associated technologies, such as electronic medical records, to simplify reporting, to improve database content, and to optimize the timing of interventions.^3- 4 Because a well-designed surveillance system is essential for the systematic collection, analysis, and dissemination of data, we designed an electronic medical record system.

From March 2007 to March 2009, all newly diagnosed AML candidates for conventional chemotherapy from 17 hematology units in Italy were enrolled in the Hema e-Chart and followed-up prospectively to assess the incidence of febrile events (FEs). Data were entered prospectively into electronic case report forms using previously described methods.⁵ Enrolling a patient in the Hema e-Chart registry had no impact on the standard clinical practice of each hematology unit.

In the 747 adults with AML treated with conventional chemotherapy, 528 FEs were registered: 208 of them (39.3%) were classified as fever of unidentified origin (FUO), while there were 21 (3.9%) disease/treatment-related fevers (DTRF). The remaining 299 FEs (56.6%) were associated with infections.

Bacterial infections were identified in 223 cases. After we excluded 20 cases of mixed infections, 203 patients had a bacterial infection only. Thirty-one patients had possible bacterial pneumonia (ie, no etiological agents were isolated and the patients were successfully treated with antibiotic therapy only). Overall, 192 patients had a microbiologically proven bacterial infection, and in 20 of them, 1 or more other agent was identified as a concomitant cause of fever. Among the 192 FEs with an identified bacterial agent, 147 infections were caused by a single agent (83 from gram-positive [56.5%] and 64 from gram-negative [43.5%] bacteria). Two or more bacterial agents were associated in 45 episodes.

Of 528 patients, 93 invasive fungal diseases (IFDs) were identified (17.6%), with an incidence of 12.4% in the examined population. Molds and yeasts were detected in 80 (86%) and 12 (13%) cases, respectively. Invasive fungal diseases were proven in 24 cases (25.7%), probable in 20 cases (21.6%), and possible in 49 cases (52.7%). In all cases of possible IFD, the suspected etiological agent was a mold. Proven or probable cases comprised 12 yeast (all Candida spp), 31 mold (all Aspergillus spp), and 1 dimorphic fungus (Histoplasma capsulatum) infection. In 73 of 93 IFDs, a single fungal infection was detected; in the remaining 20 cases, a mixed infection was diagnosed.

Viral infections resulted to be the cause of fever in 4 cases only, with a very low incidence in the overall population (4 of 747 AML cases [0.5%]).

During the study period, only 34 FEs were recognized as the cause of death, with an overall incidence in AML cases of 6.4% (34 of 528). The highest attributable mortality rate (11 of 73 [15.1%]) was observed in the IFD group. Overall, 10 deaths were registered in the FUO group (4.8%), 10 in the bacterial group (4.9%), and 3 in the mixed-infection group (14.3%). No deaths were observed either in the viral infection or in the DTRF group. Of 21 patients with mixed infection, 3 patients died (attributable mortality rate [AMR], 14.3%) (Table).

Our study shows that FUO and fungal and bacterial infections were most frequently observed in AML cases. The etiology of FUO, detected in over 38% of episodes, remains unknown despite the use of currently available diagnostic tools, and FUO-AMR is as high as 5%. This result is important considering the potential progressive expansion of preemptive therapy use in routine practice.

Bacterial infections are the most frequent cause of infectious complications in cases of AML. In our survey, the ratio of gram-positive to gram-negative bacteria approached one. Strikingly, more than 25% of proven bacterial infections were polymicrobial and 9% were mixed infections, of which most were fungal and bacterial.

The incidence of IFDs observed confirmed a trend of increased frequency of mold infections relative to yeast infections; the ratio in our study was 5:1.¹ Considering the high number of observed FUO episodes, we cannot exclude that the incidence of IFDs might be higher. The AMR due to IFDs is decreasing, confirming recent data that report a similar trend.⁶

The incidence of FUO is still high and remains one of the most challenging issues faced by hematologists. Whether viral infections in AMLs are really rare or underestimated is yet to be verified.

Hema e-Chart allows the prospective collection and analysis of targeted high-quality data derived from clinical experience and provides information on the epidemiologic patterns of infectious complications in AMLs, which are rarely studied in prospective multicenter observational trials. Our registry appears to be very useful particularly in the early phases of AML, when infectious complications are more frequent and thus affect the schedule of chemotherapy.