Introduction

Patients with hematologic malignancies may develop neutropenic fever after chemotherapy in up to 80 % of the cases. The clinical presentation may vary from undocumented clinical infection up to bacteremia, septic shock, and death [1–5]. Due to the heterogeneity of the clinical presentation and the diversity of the complications associated with febrile neutropenia, the Multinational Association of Supportive Care in Cancer (MASCC) developed a prognostic score for patients with febrile neutropenia. This score divides patients into two mutually exclusive risk groups and could serve as a basis for selecting therapeutic strategies [5]. C-reactive protein (CRP) is a global marker of inflammatory activity; high values had been associated with an increased risk of mortality [6, 7]. We have previously found that the mean value of CRP during the first 5 days of the febrile neutropenia episode was useful to identify patients more likely to die due to infection (unpublished observations).

The MASCC risk model has been mainly used among patients with solid neoplasms, the information regarding its usefulness as a prognostic tool in the hematological malignancies setting is more limited [5, 8]. CRP is an inexpensive and widely available laboratory test that can improve the accuracy of the MASCC risk score for assessment of prognosis in this group of patients. One previous study has suggested that this combination could be a convenient option that improves patient risk stratification [7].

In this study, the authors wanted to assess the prognostic usefulness of the MASCC risk score in association with the mean CRP value to identify patients with febrile neutropenia and hematologic malignancies and a higher risk of death.

Methods

A retrospective cohort study assessing survival rates based on the MASCC score and the CRP values of patients diagnosed with febrile neutropenia at the Pablo Tobón Uribe Hospital in Medellin, Colombia was performed. The study was conducted between May 1, 2009 and December 3, 2012.

Patients older than 15 years, with febrile neutropenia and a diagnosis of hematologic neoplasms or aplastic anemia who appeared in the institutional records of the hematology group were included in the study. Conversely, patients with incomplete information or with a follow-up of less than 30 days were excluded. Each episode of febrile neutropenia occurring in the same patient was included in the analysis.

Mortality rates were analyzed. These rates were calculated from the moment in which the febrile neutropenia was diagnosed until the 30th day after such episode. The data found in the clinical records was used to identify the presence of febrile neutropenia, the clinical variables, the MASCC score, the CRP values upon diagnosis, and the mean CRP values within the first 5 days of the episode. The MASCC score was calculated for all episodes of febrile neutropenia as suggested by the authors. Patients were considered to be at low risk when their scores were above 21 points and at high risk when their scores were equal to or less than 20 points (Appendix 1) [5]. Febrile neutropenia was defined in accordance with the recommendations of the Infectious Diseases Society of America as a single oral temperature higher than 38.3 °C or a temperature higher than 38.0 °C if sustained for more than 1 h in a patient with an absolute neutrophil count below 500 cells per mm3 (or expected to fall below this level in the next 48 h) [6]. As for the CRP, it was defined as high when it had a value above 15 mg/dL measured either at the moment in which febrile neutropenia was diagnosed or as the mean value during the first 5 days after diagnosis. The CRP was measured through latex agglutination using the immunoassay technique and was expressed in milligrams/deciliter (mg/dL).

After confirming the prognostic value of the mean CRP as well as that of the MASCC score, we evaluated the usefulness of this association to identity four different risk groups: low-risk MASCC score and mean CRP <15 mg/dL, low-risk MASCC score and mean CRP >15 mg/dL, high-risk MASCC score and mean CRP <15 mg/dL, and high-risk MASCC score and mean CRP >15 mg/dL.

Statistical analysis

A sample size of 236 patients was estimated (using Yates correction) after considering the following aspects: a 30-day mortality rate of 30 % for patients with high CRP values or with a high-risk MASCC score along with a mortality rate of 10 % for patients with low risk or low CRP values. In addition, the significance level was 0.05, the power was 90 %. The hypothesis and tests of independence were analyzed using either the chi-square test or Fisher’s exact test as required. The association between the analyzed variables and the need for being transferred to the intensive care unit was determined with a logistic regression using relative risk and odds ratio. The Kaplan-Meier nonparametric method was used to analyze survival rates. The comparison of the groups’ survival rates was performed using the log-rank test. The association between dependent variables was determined through the hazard ratio using the Cox regression method. The significance level was set at 5 %, and the confidence intervals were calculated at 95 %. The statistical analysis was conducted using the Stata 11.2 software.

This study was previously approved by the Research and Ethics Committee of the Pablo Tobón Uribe Hospital.

Results

A total of 237 patients with febrile neutropenia were analyzed between May 1, 2009 and December 1, 2012. See Table 1 for the characteristics of the study population. For 73.8 % of the cases, the onset of febrile neutropenia occurred during the hospitalization period. Furthermore, 5.9 % of the episodes occurred among patients that were having hematopoietic stem cell transplantation. Most patients had a diagnosis of acute leukemia (68 %); they were followed by patients with lymphoma (23 %) or other neoplasms (9 %).

Table 1 Characteristics of the study population (N = 237)
Full size table

The MASCC score classified as low risk, 56.5 % of the cases and as high risk, 43.5 % of them. The baseline values for CRP were above 15 mg/dL for 26 % of the patients. Similarly, for 36 % of the cases, the mean value during the first 5 days of the febrile episode was above 15 mg/dL. In 26 % of the cases, admission to the ICU was necessary. The reasons for admission were either shock or respiratory failure in all of them. Finally, 9.3 % of the patients died within 30 days after diagnosis.

Association between the MASCC risk score, CRP, and the mortality rates

Upon categorizing the variables composing the MASCC model, we found that 103 patients (43.5 %) had moderate or severe symptoms at the time of diagnosis, 66 (27.9 %) had arterial hypotension, 3 (1.3 %) had chronic obstructive pulmonary disease (COPD), 17 (7.2 %) had a history of previous fungal infection, 103 (43 %) showed dehydration requiring parenteral fluids, 175 (73.8 %) were hospitalized at the moment of the episode, and 186 (76.8 %) were younger than 60. The variables of the MASCC score that better identified patients with a higher risk of mortality were symptom burden, hypotension, and dehydration requiring intravenous fluids (Table 2). Upon analyzing the survival rates within 30 days based on the MASCC score, we found a survival rate of 97 % for the population with a low-risk MASCC score. Patients with a high-risk score were found to have a survival rate of 87 % (p < 0.001) (Fig. 1). Likewise, when compared with low-risk patients, high-risk patients according to the MASCC score showed a greater probability of dying (RR 2.21; CI 95 % 1.74–2.79, p < 0.001) or being transferred to the intensive care unit due to severe infection (RR 3.55; CI 95 % 2.73–4.62, p < 0.001).

Table 2 MASCC risk score variables and 30-day mortality risk
Full size table
Fig. 1
figure 1

30-day survival according to the MASCC risk score

Full size image

Upon conducting a univariate analysis, the following predictive variables for mortality were found: a high-risk MASCC score, a baseline CRP value greater than 15 mg/dL, a mean CRP value greater than 15 mg/dL, documented infection, bacteremia, infection with a resistant germ, and admission to the intensive care unit. After multivariate analysis, the following variables retained prognostic power: a high-risk MASCC score, a mean CRP greater than 15 mg/dL, and admission to the intensive care unit (Table 3).

Table 3 Univariate and multivariate analysis of variables associated with mortality
Full size table

Four groups were established by the prognostic model based on mean CRP values and MASCC risk score. These groups had different death risks. The rate of survival within 30 days was 100 % for patients with low-risk MASCC scores and a mean CRP <15 g/dL, whereas the same rate was only 64 % for patients with high-risk MASCC scores with high CRP (Fig. 2).

Fig. 2
figure 2

Survival rates at 30 days according to the mean CRP values and the MASCC risk score

Full size image

Discussion

The MASCC risk score was initially designed as a tool to identify low-risk subjects with febrile neutropenia that can be managed as outpatients. We found that this risk score can be used for assessing the risk of mortality and the need for being transferred to the ICU among Colombian patients with hematologic neoplasms and febrile neutropenia. When trying to optimize the prognostic models, including CRP values along with the MASCC score was extremely useful, especially for a better stratification of the high-risk group.

Klastersky et al. documented that out of 551 patients identified as being at low risk, 6 % experienced serious medical complications and 1 % died. As for the group considered to be at high risk (205 patients), 39 % of the patients had serious complications and 14 % died [5]. These results are consistent with ours but in this cohort, fewer patients had a low-risk MASCC score (56 vs 73 %). This could be due to the fact that the present study only included patients with hematologic neoplasms. Furthermore, most febrile neutropenia episodes (68 %) were secondary to acute leukemia, and 73 % of them occurred in hospitalized patients; for this reason, not all the individual variables had the same weight for the prognosis assessment. The individual variables of the MASCC scale with the highest statistical and prognostic weight for this study were symptom load, hypotension, and dehydration requiring intravenous fluids. In different parts of the world, the MASCC scale has been validated as a prognostic tool for patients with febrile neutropenia. Moreover, various treatment guidelines recommend its application so that low-risk patients can be managed as outpatients [7–13].

Upon reviewing the literature on the usefulness of inflammatory markers as predictors of bacteremia or severe infections, we found several studies that evaluated CRP, procalcitonin, interleukin-6, or interleukin-8 [14–22]. Two findings are worth mentioning. First, most of these studies were carried out among pediatric patients [14–17, 21], whereas the present study included adults with diverse hematologic malignancies. Secondly, there is no clear association between the levels of CRP measured at the time of onset of neutropenia and the severity of the infection or bacteremia [16–22]. It is well known that CRP is a sensitive indicator of inflammation, albeit its specificity is low. In a previous observation, we found that specificity could be improved by using mean CRP value (measurements obtained during the first 5 days of each neutropenic episode) instead of a lone baseline value. This has not been reported elsewhere and could be a useful tool in resource-poor settings such as ours.

Our study has several limitations. The retrospective nature of the design and the fact that the study was carried out in a single institution could lead to bias. On the other hand, the external validity of our findings is high as the MASCC risk score can be used in almost every setting, and CRP is widely available. We recommend validating these findings in prospective studies with different populations in order to confirm their usefulness.

In conclusion, the combination of the MASCC risk score and mean CRP value during the first 5 days of a neutropenic episode successfully identifies a group of patients with hematologic malignancies and a high risk of death.