Invasive fungal infections in neutropenic enterocolitis: A systematic analysis of pathogens, incidence, treatment and mortality in adult patients

Neutropenic enterocolitis is the most important and a highly life-threatening abdominal infection in neutropenic patients. This complication most frequently occurs after intensive chemotherapy in acute leukaemias. The pooled incidence, derived from all available studies, was calculated to be 5.3% in patients hospitalised for haematological malignancies, for high- dose chemotherapy in solid tumors or for aplastic anaemia [1]. The literature provides generally poor evidence on the treatment of neutropenic enterocolitis.

In a previous systematic review, we could not find prospective trials nor case control studies on any part of therapy [1]. The most intensively discussed therapeutic question in the past decades is whether patients with neutropenic enterocolitis should receive conservative or surgical therapy. In contrast, the differential options of medical management consume much lesser space in the discussion sections. Gramnegative bacteria constitute the most important group of causative pathogens [2‐5]. There is a consensus that immediate empirical broad spectrum antibiotic therapy is mandatory, but it is unclear which antibiotic is the drug of choice for empirical therapy. After all, application of a modified recommendation of the current IDSA guidelines for the treatment of neutropenic patients with unexplained fever can be regarded as reasonable to select appropriate antibiotics [1, 6].

The role of fungi as causative organisms is even more obscure. In conventional reviews on neutropenic enterocolitis, sometimes citations concerning detection of Candida spp. in blood cultures can be found [7‐9]. In case series and cohort studies, frequencies of causative fungi are either not given or differ largely from 0% up to more than 19% in autopsy studies [10]. In a paediatric study, 16% of the organisms, cultured premortem from blood, were fungi. Moreover, in this report fungal pathogens accounted for 53% of new microorganisms seen at autopsy [11]. Mortality has been reported to be up to 100% in patients with fungaemia [2]. The latter studies suggest a high relevance of fungal pathogens. Changes in the intestinal flora, due to frequent use of broad spectrum antibiotics in neutropenic patients, might support fungal invaders. Surprisingly, the need for an empirical antimycotic therapy is nearly never discussed in reports describing patients with fungal neutropenic enterocolitis. Similarly, guidelines give no clear recommendations for the management of these patients. IDSA guidelines for the use of antimicrobial agents in neutropenic patients with cancer [6], generally, make very few recommendations concerning abdominal infections. Particularly, no statement regarding the indication for antimycotic therapy in neutropenic enterocolitis is given. The use of empirical antimycotic therapy is not recommended as a necessary component of the first-line antimicrobial therapy of neutropenic patient with unexplained fever. Hughes et al. state that, although clinicians disagree as to when, and even if, amphotericin B therapy should be introduced empirically, most believe that the patient who remains febrile and profoundly neutropenic for >5 days, despite the administration of broad-spectrum antibiotics in adequate dosages, is a candidate for antifungal therapy.

On the other hand, there are clinically documented infections in neutropenia in which first line empirical antimycotic therapy can be regarded as standard. The guidelines of the German Society of Haematology and Oncology recommend that in febrile neutropenic patients with severe neutropenia (i.e. lasting for more than 10 days) and lung infiltrates, initial antimicrobial therapy should include amphotericin B [12]. The markedly poorer prognosis of both neutropenic enterocolitis [1] and pneumonia [13] compared to fever of unknown origin (FUO) might suggest that early antifungal could be also beneficial for patients suffering from neutropenic enterocolitis.

For these reasons, we conducted this analysis. Following a systematic search, we extracted and summarised all detail data (if possible, single patient data) from the complete literature to enhance evidence regarding the question whether a routine empirical antimycotic therapy should be administered as a parallel with broad-spectrum antibiotic therapy.

The following clinical questions were posed in this systematic review: What is the frequency of invasive fungal infections in neutropenic enterocolitis? What is the mortality of fungal neutropenic enterocolitis? Does the frequency and the microbiological distribution of invasive fungal infections support empirical antimycotic therapy? Do reported results of antimycotic therapy in documented fungal neutropenic enterocolitis help with the selection of appropriate drugs?

A MEDLINE search, as of 15 March 2005, with the search terms "neutropenic enterocolitis or neutropenic colitis" and "typhlitis and (neutrop* or granulocyt*)" yielded 286 articles. This represented an extension to a previously reported search strategy [1]. After individual review of these titles and abstracts, we excluded 87 articles describing paediatric patients, five articles describing Clostridium difficile colitis and 55 other articles, not dealing with neutropenic enterocolitis, reviews without primary data or duplicate publications. After this selection, 139 articles were remaining. Additional hand search in reference lists of these reports and reviews revealed further 47 relevant articles, whose reference lists were also screened. Finally a total of 186 articles were found to be relevant and their full text versions were studied. Among these we found, again, no study on neutropenic enterocolitis that fulfilled the criteria for evidence levels 1a to 3b according to Phillips et al. [16]. We found no systematic review, no randomised controlled trial, no good quality cohort study and no good quality case control study on any part of therapy, particularly, on the role of antifungal treatment. Eighty-four reports of evidence level 4 (case series or poor quality cohort or poor quality case control studies) were detected. One-hundred and two reports of evidence level 5 (expert opinion) were found.

This is the first analysis on the relevance of fungal pathogens in neutropenic enterocolitis of adults, based on a systematic analysis of the literature (including our own primary data). The approach was to extract the complete detail data from the studies and to increase evidence by pooling analyses. The maximum number of clearly described patients in a single study was 5, which is to low to perform conclusive statistics concerning frequency, mortality, spectrum of causative species or therapy. A discussion about this question is lacking in the literature, especially in contrast to the extensive debate concerning the role of surgery.

The frequencies of invasive fungal infections in neutropenic enterocolitis were lower than expected. In our own patients from two studies, we observed fungal neutropenic enterocolitis in 11.8% (2/17) [4, 23]. Autopsy studies reported frequencies up to more than 19% [10]. However, several selection and detection biases of these methods of frequency calculation have to be considered. On one hand, fungal neutropenic enterocolitis is probably underestimated, since histology, as the most specific diagnostic method, is rarely obtained in living patients. On the other hand, patients with more severe courses may be selected because of the inclusion of surgical or pathological studies that did not describe consecutive patients. Furthermore, the diagnostic criteria in studies concerning neutropenic enterocolitis are generally heterogeneous. However, in most reported cases of invasive fungal infection, neutropenic enterocolitis was confirmed histopathologically (Table 1).

Fungal infections other than Candida spp., particularly Aspergillus fumigatus, are rare exceptions. Unfortunately, only in a minority of reported cases, a species confirmation is available. Among these patients, Candida albicans occurred most frequently, but in some patients, fluconazole-resistant species such as Candida glabrata and Candida krusei have been observed. In approximately half of the patients, diagnosis of invasive fungal infection is established by blood culture and in the other half by histology. Bacteria are clearly the most important group of pathogens in neutropenic enterocolitis. Rates of bacteraemia between 34% [27] and 82% [28] have been reported in larger series. In most patients, development of neutropenic enterocolitis is probably a multifactorial process. In our analysis, the term "causative pathogen" does not mean "exclusively causative". We recognize that mostly secondary infection of injured mucosa is responsible for deterioration, complications and poorer outcome of the patients. Furthermore, from a clinical and practical point of view, the treatable factor infection and not hardly treatable factors such as toxic damage should be stressed.

The mortality of fungal neutropenic enterocolitis was very high (81.8%). This pooled mortality rate may be biased and slightly overestimated due to inclusion of 8 patients from autopsy case series. Less severe cases may be moderately underrepresented. However, it must be emphasised that the course of fungal neutropenic enterocolitis is not invariably fatal. Eight patients survived and the description of a 100% mortality rate reported by Starnes et al. and cited by others [2, 29] can not be confirmed. It was striking, that in the majority of cases with fungal neutropenic enterocolitis, authors did not report or discuss the use of antimycotics. One might speculate that this obviously low grade of consideration of antifungal therapy in neutropenic enterocolitis might have contributed to the high mortality rate.

Infections from Aspergillus spp. are very rare exceptions (despite observation of two patients including one with jejunal ulceration by Aspergillus in one of our studies [4]). This rarity may be explained by the fact that Aspergillus infections are typically acquired by inhalation and invasive growth in the respiratory tract.

Evidence concerning empirical antifungal therapy of neutropenic enterocolitis is still very limited. We found that neither prospective nor high quality retrospective studies concerning this issue in neutropenic enterocolitis are available. In 33 patients with fungal neutropenic enterocolitis, in reports of evidence levels 4 and 5, it was not described that any antifungal therapy was given. In 20 patients with proven fungal neutropenic enterocolitis more or less precise information regarding empirical or microbiologically induced antifungal therapy was provided. Only in two of these patients empirical antimycotic therapy (fluconazole) was clearly reported [21, 22]. Many other patients probably received antimycotics only after detection of fungaemia. Unfortunately, patient numbers and data quality are not sufficient to analyse, statistically, the success of antimycotic therapies, especially empirical use. Our analysis is limited by lack of additional individual information from the authors on antifungal therapy.

Possibly, many authors see no need for empirical antifungal therapy. Only few authors of the relevant studies recommended empirical antifungal therapy [20, 24, 30]. Girmenia et al. suggested a semi-empirical selective therapy with fluconazole based on the detection of Candida mannoproteinaemia [24]. Micozzi et al. pointed out that empirical administration of fluconazole in all febrile patients could represent a way to prevent neutropenic enterocolitis [20]. Wach recommended that conservative therapy of neutropenic enterocolitis should include antimycotic drugs but did not provide more specific recommendations [30]. D'Amato et al. stated that antifungal treatment should be considered [31].

In the NCCN practice guidelines for fever and neutropenia "consideration of antifungal coverage" is recommended for febrile neutropenic patients with abdominal pain or diarrhea [32]. In the IDSA guidelines for the use of antimicrobial agents in neutropenic patients with cancer, no specific recommendation concerning fungal neutropenic enterocolitis is provided [6]. However, one might have the opinion that these patients fall within the scope of the sections of these IDSA guidelines referring to the treatment of neutropenic patients with unexplained fever. They state that, although clinicians disagree as to when, and even if, amphotericin B therapy should be introduced empirically, most believe that the patient who remains febrile and profoundly neutropenic for >5 days despite the administration of broad-spectrum antibiotics in adequate dosages, is a candidate for antifungal therapy [6].

So, should empirical antimycotic therapy in neutropenic enterocolitis be recommended? If the answer is yes, when should antimycotic therapy be started? Which drugs should be used? We believe that application of the above recommendations for FUO to our subgroup of patients is justified on principle. All patients who remain febrile and profoundly neutropenic for >5, days including those with neutropenic enterocolitis should receive amphotericin B empirically. Some patients, who are already febrile several days before clinical signs of neutropenic enterocolitis develop, will receive amphotericin B earlier as 6 days after diagnosis of neutropenic enterocolitis.

The question whether the frequency and the microbiological distribution of invasive fungal infections support immediate empirical antimycotic therapy can not be answered clearly from the literature data. If the moderately low frequency of fungal pathogens is taken into consideration, clinicians might not be willing to start empirical antimycotic therapy with amphotericin B only based on the diagnosis of neutropenic enterocolitis, immediately on the first day of fever. Therefore, an earlier start of empirical antimycotic therapy (like in patients with pulmonary infiltrates) would not be preferred by many.

However, some arguments could be made in favour of an empirical use of fluconazole in neutropenic patients with enterocolitis: (1) The drug has a low toxicity. (2) Candida albicans is the leading fungal pathogen. (3) Patients with neutropenic enterocolitis and a positive Candida mannoprotein test treated with fluconazole have been shown to have a good prognosis [24]. (4) Fluconazole is an effective drug in clinically stable, non-neutropenic patients with candidaemia [33].

However, fluconazole is not a safe choice for treating non-albicans candidaemia, especially when Candida krusei (primary resistance) or Candida glabrata (dose-dependent sensitivity) have been identified [34]. Many epidemiological studies indicate that the proportion of non-albicans Candida species is clearly increasing and around 50–60 % [35‐37]. Also, the empirical use of fluconazole must be restricted to patients who did not receive azoles as antifungal prophylaxis, since no improvement of antimycotic coverage by fluconazole can be expected. Currently, a major part of patients will be pretreated with itraconazole. Our own meta-analysis suggests such an antimycotic prophylaxis in long term neutropenic patients [38], although this practice is still a matter of debate [39] and although its impact on the incidence of neutropenic enterocolitis is unclear. Furthermore – and this a very strong argument against fluconazole for neutropenic enterocolitis – many patients suffer from multiple infections or are at least at high risk to develop a mould infection which would not be covered by fluconazole. All substances that have demonstrated efficacy in valid clinical trials of empirical antifungal therapy for FUO are active against Aspergillus spp. like liposomal amphotericin B [40], itraconazole [41] and caspofungin [42].

Unfortunately, no data are available on the use of itraconazole or other novel antimycotics, such as voriconazole and caspofungin, in patients with proven fungal neutropenic enterocolitis nor on the empirical use of these agents in this entity. A prospective multicenter study comparing empirical therapies, antibacterial agent vs. the same antibacterial agent plus antimycotic, is warranted.

In our opinion, in patients with evidence of invasive fungal neutropenic enterocolitis (in most cases by detection of candidaemia) administration of amphotericin B, caspofungin or voriconazole are appropriate choices. However, in contrast to amphotericin B [4, 22], we found no paper reporting experiences with caspofungin or voriconazole in fungal neutropenic enterocolitis, but only encouraging results from candidaemia trials [43, 44]. Of these latter drugs, caspofungin is indicated for patients with renal dysfunction.

In patients with neutropenic enterocolitis, fungal pathogens play a relevant, but secondary role compared to bacteria. The frequency of invasive fungal neutropenic enterocolitis is probably around 5% and its mortality around 70–80%. The frequency may be underestimated, since only few histological specimens are obtained in living patients and the sensitivity of the diagnostic methods is limited. Early empirical antimycotic therapy may be therefore considered in neutropenic enterocolitis. However, evidence concerning therapy is very poor, since most authors did not report or apply antifungal therapy. A clear recommendation for empirical antimycotic therapy outside the context of prolonged fever can not be made.