This study was the first study to assess CISNE and MASCC score performance based on malignancy type. Previous studies by Koppaka et al.[11] and Moon et al.[7] were only conducted on patients with solid malignancy. Other research by Bayonas et al.[2] was conducted on patients with solid malignancy in 90% of cases. The CISNE score itself was originally developed by Bayonas et al.[2, 12] in a solid malignancy population, and thus its performance in hematologic malignancy is unknown. The incidence of neutropenia in hematologic and solid malignancy display different characteristics; for instance, neutropenia periods are longer, and positive culture results are more often found in patients with hematologic malignancy.[13] In our study, there were no significant differences in the characteristics of patients with solid tumors and hematologic malignancy subjects. The efficacies of CISNE scores and MASCC scores were assessed based on each malignancy type.
Furthermore, the types of solid tumor in this study differed from the types of solid tumor in previous studies. In a study by Bayonas et al.[2] and Moon et al.,[7] the majority of solid tumor were breast cancer. In this study, the majority of FN occurrences in solid tumors were head and neck tumors (46.32%), followed by bone and soft tissue tumor (17.89%). These findings showed a difference in the prevalence of tumors in our study location, because as the national referral hospital we received many end-stage cancer patients with different types of tumor distribution.[14, 15]
In hematologic malignancy subjects, 87.67% of patients had non-Hodgkin's lymphoma. We excluded patients with acute leukemia because febrile neutropenia in acute leukemia patients, according to the IDSA guidelines, are in the high-risk category, regardless of their MASCC scores. Aplasia in patients with acute leukemia is influenced by many factors, including those associated with the disease (e.g., leukocyte function defects, or humoral immunity deficiencies); patient-dependent factors (e.g., age, comorbidities, or malnutrition); or the effects of chemotherapy (e.g., prolonged aplasia due to high-intensity chemotherapy, bacterial colonization due to aplasia, recurrent antibiotic use, or central catheter–related infection) therefore we exclude acute leukemia in our study.[16]
By the time of FN diagnosis, 55.36% subjects had documented infections, a greater number than in previous studies, which recorded an infection rate around 30% in FN patients.[17] In subjects with documented infection, 51.5% had pneumonia, a common comorbidity in neutropenic patients. Vehreschild et al.[18] showed that 36.4% of the neutropenic patients in their study had pneumonia. In addition to pneumonia, some subjects also displayed gastrointestinal infection in the form of acute gastroenteritis and diarrhea. These conditions in neutropenic patients are also known as neutropenic enterocolitis, which displays an incidence estimated around 5.6%. Such infections can occur as a direct result of taxane-class chemotherapy, which can damage the mucosal surface.[19]
The median of the absolute neutrophil count (ANC) was 212 cells, which categorized in severe neutropenia. The risk of infection and complications in patients with FN becomes higher as ANC values drop.[20, 21] Procalcitonin levels were examined in 64.2% of subjects, with a median value of 2.66 ng/ml. Procalcitonin is a significant parameter to distinguish infections as a cause of FN. A systematic study by Sakr et al.[22] obtained a cut-off value of 0.5–2 ng/ml for indicating life-threatening conditions in neutropenic patients. Increased procalcitonin levels were more pronounced in patients with bacteremia or sepsis, and the mean value of 2.66 ng/mml illustrated a serious condition encountered in our subjects.
Complications were found in 45.83% of subjects, a greater prevalence than previous studies. For example, in an experiment by Ahn et al and Coyne eatl.,8,[9] complications occurred in 20.8% and 25.7% of subjects, respectively. This difference might occur because of the different settings from which subjects were recruited. Coyne et al. and Ahn et al. recruit subjects from the emergency wing triage so that clinically stable patients were included in the subject. In contrast, in this study, we did not include outpatient data. This was also illustrated in the percentage of subjects classified in the low-risk category by MASCC and CISNE scores. In this study, based on MASCC scores, 58.3% of participants were included in the low-risk category. These results differed from those found by Ahn et al and Coyne et al., in which 89.1% and 73.5% of subjects were at low risk, respectively. The population in this study showed a higher prevalence of complications. In those who appeared to have a complication, 83.1% were diagnosed with sepsis, and 72.7% underwent shock. This is higher than the previous study that estimated the prevalence of this sepsis in FN patient ranges from 7 to 45%.[21]
This study analyzes the performance of CISNE scores compared with MASCC scores in predicting complications in FN patients with solid and hematologic malignancies after chemotherapy. In patients with solid malignancies, CISNE scores performed well, with an AUROC of 0.89. Two previous studies have analyzed the performance of CISNE scores in patients with solid malignancies. Research by Koppaka et al.[11] also showed that CISNE scores had an AUROC of 0.846, whereas in Moon et al.[7], CISNE scores had an AUROC of only 0.66. In our study, the population of solid malignancies was dominated by head–neck malignancies, whereas previous studies showed head–neck malignancies in only 5% of populations. Also, these research subjects were recruited in different settings. Previous studies gathered subjects from emergency wings and outpatient, whereas in our study, subjects were recruited from hospitalized FN patients. The CISNE score in the study by Bayonas et al.[2] was divided into three groups: low, moderate, and high risk, with a sensitivity of 80.1% and a specificity of 75.6%. Koppaka et al. used cut-off point 2 to divide patients into low risk and high risk groups with a sensitivity of 80.6% and specificity of 72.8%, a positive predictive value of 35.2%, and a negative predictive value of 95.3%. We conducted an analysis to assess the cut-off point and obtained a sensitivity of 74.5%, a specificity of 86.4%, a positive predictive value of 86.4%, and a negative predictive value of 74.5%, and cut-off point of 2.
This is the first study to analyze the effectiveness of CISNE scores in predicting complications of FN after chemotherapy in patients with hematologic malignancy. The analysis showed that the CISNE scores perform well, with an AUROC of 0.91. Previous research by Mohindra et al.[10] was not specific to hematologic malignancy, but this group also conducted a subgroup analysis of the hematologic malignancy population and obtained an AUROC of 0.67. We performed the cut-off analysis to get the best diagnostic value. We obtained a cut-off point of 2 with a sensitivity of 78.9%, a specificity of 91.4%, a positive predictive value of 90.9%, and a negative predictive value of 80%. There was a significant difference between the performances of the CISNE and MASCC scores in predicting complications in patients with FN that had resulted from hematologic malignancy chemotherapy. These results showed that CISNE scores effectively predict complications in febrile neutropenia patients, both in those with solid and hematologic malignancies. These differences between CISNE and MASCC scores may result from different score components. CISNE scores assess patient conditions with ECOG performance, whereas MASCC scores assess patients' burden of disease. The use of ECOG performance can give a more objective assessment than the burden of disease, which is highly dependent on clinician assessment. These objective assessments resulted in less misclassification of patients.[6]
Our study showed that the use of CISNE scores to substitute MASCC scores in risk classification of FN patients, both solid and hematologic malignancies, can reduce misclassification. Further research on the role of CISNE scores in the selection of antibiotic therapy may be needed. We found that re-classification of CISNE scores into two groups (low risk and high risk) with a cut-off point of 2 was more effective in predicting complications, thus helping clinicians determine patient therapy.
Limitations
Our study limitation was a retrospective study using secondary data obtained from medical records; therefore some subjective score components did not have standard parameters. This source of medical record was also the cause why the demographic data of the type of tumor is different from the national or regional characteristics