Validation of an Index of Neutropenia (D-index) in Febrile Neutropenic Cancer Patients

Overview

The main objective of this study is to test prospectively the performance of an algorithm stratified by an index based on neutrophil counts in association with galactomannan assay and image tests to start an antifungal early therapy (empirical/preemptive) in neutropenic patients. Ths specific objectives are to determine the overall incidence of invasive fungal infections, use of antifungal agents, duration of hospitalization and mortality in this cohort, and to evaluate if this strategy is associated with a reduction in the expected use of antifungal agents if a classical empiric antifungal strategy was used, without an increase in the incidence of invasive fungal infections. This is a prospective, non randomized, non comparative study. Patients aged ≥ 18 years are eligible if they have acute leukemia, myelodysplasia or other baseline disease submitted to chemotherapy or to allogeneic stem cell transplantation with an expected duration of neutropenia (neutrophil count \<500cells/mm³) of at least 10 days. Exclusion criteria are patients with and a past history of or invasive mold infection and those who do not want to participate. The study has no comparator arm. However, the investigators intend to determine if the algorithm based on the D-index would result in a 50% reduction in the use of antifungal agents, if all patients with persistent fever and neutropenia received empiric antifungal therapy. Based on our database of \~2,000 episodes of febrile neutropenia, 36% of patients had persistent fever between days 4 and 7 of antibiotics and would receive empiric antifungal therapy. A total of 105 patients will be needed to demonstrate a 50% reduction in antifungal use if the investigators compared this cohort with a matched control historical cohort (alpha = 5%, beta = 20%).

Neutropenia is a major risk factor for invasive fungal infections, particularly those caused by moulds, such as Aspergillus spp., Fusarium spp. and the Zygomycetes. Among neutropenic patients, these invasive mould infections (IMI) occur almost exclusively in patients with profound neutropenia (\<100 neutrophils/mm3) lasting more than 10-15 days \[1-3\]. Empiric antifungal therapy is considered standard of care in patients with persistent or recurrent fever and neutropenia. However using fever as the sole criterion for starting empiric therapy, a significant number of patients will receive antifungal therapy unnecessarily, particularly among recipients of fluconazole prophylaxis, because persistent fever in these patients is commonly due to various factors, such as uncontrolled occult bacterial infection, viral infection, drug fever and others \[4\]. An alternative to empiric antifungal therapy is the preemptive treatment. In this strategy, other markers (such as serology, radiology and clinical data) are added to persistent fever \[5,6\]. One problem of such strategy is that the biomarkers currently in use (galactomannan \[GM\], 1,3-beta-D-glucan and PCR) have high sensitivity. If one of these markers is used to trigger the start of antifungal therapy, still a considerable number of patients will use antifungal agents unnecessarily. However, a positive biomarker early in the course of neutropenia most likely represents a false-positive result. By contrast, a positive biomarker test late in the course of neutropenia is more likely to be a true positive. A problem is that no cutoff value of duration of neutropenia is validated to help clinicians to identify these 2 possibilities. A clinical parameter that evaluated the dynamics of neutropenia, combining intensity and duration, could be a good tool to identify patients at high risk for IMI. This tool could be used to stratify patients, helping clinicians to select appropriate candidates for early antifungal therapy (empiric or preemptive) in persistently febrile neutropenic patients. We developed an index (called D-index) that uses data from white blood cell counts (WBC), and combines intensity and duration of neutropenia. This index was tested in patients with acute myeloid leukemia, and showed a good discriminatory performance in identifying patients with IMI and without IMI. A cutoff was derived, and showed good sensitivity, specificity and a very high negative predictive value (97 to 99%) \[7\]. The high negative predictive value of this index is very similar to that obtained with GM and 1,3-beta-D-glucan in febrile neutropenic patients \[8,9\]. With this regard, the D-index could be of help to assess the pre-test probability of IMI and help to interpret the results of positive serum GM and / or 1,3-beta-D-glucan: patients with low values of D-index would be best interpreted as having false-positive results of these tests. Similarly, positive biomarkers in the presence of a high D-index would be more likely to be true positives, since the combination of different tests may increase their positive predictive value. This is a prospective cohort study. The study will be submitted to the IRB and written informed consent will be obtained from all participants. Patients aged ≥ 18 years are eligible if they have acute leukemia, myelodysplasia or other underlying hematologic malignancy receiving chemotherapy with an expected duration of neutropenia (\<500/mm3) \>10 days. This includes allogeneic (but not autologous) hematopoietic stem cell transplantation. Patients with past history of IMI will be excluded, since for these patients secondary prophylaxis is advised. Patients will be enrolled at initiation of chemotherapy and no later than 48h from the first day of neutropenia. The same patient will be able to be enrolled more than once, but if he develops IMI, he will be no longer eligible for a subsequent inclusion. All patients will receive antimicrobial prophylaxis with fluconazole (400mg daily). Chest imaging (X-ray or CT scan) will be performed at the enrollment. During the period of neutropenia induced by chemotherapy/ conditioning regimen, patients will be surveyed for the development of fever and will have the D-index calculated day by day. At the time of febrile neutropenia, at least 2 blood cultures will be collected. Other microbiologic exams will be obtained if clinically indicated. The patients will then receive a broad-spectrum β-lactam antibiotic (cefepime). Vancomycin will be added for patients who have clinically suspicion of catheter-related infection or a blood culture yielding Gram-positive bacteria not yet identified. Serum will be obtained 3 times a week for Aspergillus GM assay (Platelia Aspergillus; Bio-Rad) and Aspergillus sp. PCR (in-house real time qualitative and quantitative PCR). A positive GM will be defined as an index ≥ 0.5. In case of persistent fever or a new fever, blood cultures will be collected and CT scans of the chest and the sinuses will be performed. Results of GM will be checked and the patients will be stratified according to the D-index as high, moderate or low risk for IMI. Based on this analysis, antifungal therapy (caspofungin 70 mg day 1 and 50 mg/d thereafter) will be started. In high risk patients (D index ≥ 5,800), the trigger for anti mold treatment will be the presence of any image on CT scan or any positive GM. In the moderate risk (D-index between 3,000 and 5,800), the trigger will be the presence of persistence of a positive GM in consecutive tests. In this stratum, chest CT scan will be defined as indicative of antifungal initiation only in the presence of one of the following: nodules \> 1 cm, with or without halo sign, wedge-shaped images, or cavitation. For low-risk patients (D-index \<3,000), antifungal treatment will be started only in the persistence of positive GM result in consecutive tests. In situations in which antifungal treatment is not started, reassessment of these parameters will be made at least twice a week with the evaluation of the D-index and biomarkers, and once a week with CT scans. In patients with lung infiltrates, a bronchoalveolar lavage (BAL) will be taken, provided the patient is not hypoxemic or severely thrombocytopenic and refractory to platelet transfusions. In the presence of findings in sinuses CT scans suggestive of sinusitis, nasal endoscopy will be performed. BAL and sinus aspiration samples will be processed for direct examination, culture, GM and Aspergillus PCR. Other diagnostic procedures will be performed when clinically indicated. Autopsy will be performed in all fatal cases after family authorization. Invasive fungal infection will be classified according to the Revised Definitions from the European Organization for Research and Treatment of Cancer/Invasive Fungal Infections Cooperative Group and the National Institute of Allergy and Infectious Diseases Mycoses Study Group (EORTC/MSG) Consensus Group \[10\]. To assess the performance of risk stratification based on the D-index, the overall use of antifungal treatment (al well as stratified by D-index) will be recorded, and reported by descriptive statistics. The outcome of each risk stratus (low, moderate and high) will be compared regarding: incidence of suspected and documented mold infection, use of anti mold therapy, duration of hospitalization and death rate. In addition, the proportion of patients receiving early antifungal treatment will be compared with empiric antifungal use in a historical matched-control cohort. Patients from the 2 cohorts will be compared using propensity score procedure. Propensity score analysis attempts to identify patients who are similar except for their treatment. The analysis will be performed as follows: first, the probability of any patient in the database to receive empiric antifungal therapy will be estimated using a multivariate logistic regression model that incorporates baseline characteristics, underlying diseases and co-morbidities. Next, we will match from the 2 cohorts patients with the nearest propensity score (the score generated by the multivariate equation)\[10\]. Based on our database of \~2,000 episodes of febrile neutropenia, 36% of patients had persistent fever between days 4 and 7 of antibiotics and would receive empiric antifungal therapy. A total of 105 patients would be needed to demonstrate a 50% reduction in antifungal use if we compared this cohort with a matched control historical cohort (alpha = 5%, beta = 20%).