Does age play a role in fever and neutropenia events and complications: A comparison of adolescents versus younger children with cancer at a tertiary care pediatric hospital, a pilot project

Abstract Background Adolescents and young adults with cancer (AYA) are a complex group of patients. The development of fever and neutropenia (FN) is a potentially lethal complication of chemotherapy. Risk stratification of patients with FN has become increasingly valuable allowing for early intervention and to guide treatment type and duration appropriately. There are risk stratification guidelines that exist, but most are validated in young children with cancer (YCWC). AYA are frequently shown to have more numerous and severe side effects from chemotherapy. Aims This study aimed to identify whether age contributes to the incidence and severity of FN. Methods and Results Patients diagnosed with a malignancy in a 5‐year period at our institution were included from ages 0–18 years. We reviewed details of their FN events, including duration of hospital admission, source (bacterial/fungal), PICU admission and duration, positive blood cultures and mortality. Adolescents with cancer (AWC) had a trend of being 1.56 times more likely to have FN events (CI 95% 0.936–2.622, p = 0.087). Assessment of the duration of PICU stay showed that AWC were 4.9 times more likely to have longer admissions (CI 95% 0.998–24.067, p = 0.050). There was no significant difference between the two groups in the rate of PICU admission, positive cultures, identification of a bacterial or fungal source, hospital admission duration or mortality from FN. Conclusion This study demonstrated a trend towards AWC being more likely to develop FN events. When such events occur in this group, the severity of them may be heightened as evidenced by longer duration of PICU admission.


| INTRODUCTION
Adolescents and young adults with cancer (AYA) are commonly defined as those who fall within the age range of 15-29 years, with certain institutions extending this definition to include those up to 39 years of age. 1,2 This group faces numerous healthcare challenges that disproportionately impact their overall health outcomes and wellbeing compared to their pediatric and adult counterparts with the same diagnoses. 1,2 Some of these disparities in care arise from their unique biologic, epidemiologic, and psychosocial factors in association with their underrepresentation in clinical trials. [3][4][5] AYA have additionally been noted to suffer more side effects from therapy compared to younger children. 5,6 Additional consideration must also be given to the discontinuity of care faced by AYA who are often treated at pediatric institutions until 18 years of age then transitioned to adult institutions beyond this, creating both physical and academic barriers to studying this population. Given the varying treatment protocols and approach to the management of complications between adult and pediatric institutions, the treatment of AYA is overall less uniform. All these aspects have contributed to the lag in improvement in cancer mortality for the AYA population, in contrast to the more drastic improvements made in younger children with cancer (YCWC). 5,6 These unique factors highlight the need for further recognition of cancer associated risks in order to limit adverse health outcomes and improve overall survival for this group, by creating more standardized treatment guidelines. [3][4][5] A common oncologic emergency is fever and neutropenia (FN) which is associated with increased morbidity and mortality. [7][8][9] The presence of neutropenia can result in an increased risk of life-threatening infections and chemotherapy dose reductions and delays, which may negatively impact treatment outcomes. 10,11 Given that about half of the pediatric oncologic population develop at least one episode of FN when treated with chemotherapy, routine emergency hospitalization upon presentation and empirical antibiotic treatment has significantly reduced mortality over the years to around 1%. 12 However, FN presentations vary among patients with delayed recognition resulting in worse outcomes prompting the need for improvements in patient risk assessments. 13,14 Risk stratification strategies and clinical decision rules for presentations of FN have been adopted internationally in order to provide appropriate individualized treatment for patients. 15 Various risk factors have been identified in predisposing patients to the development of FN, some of which include chemotherapy type and dose intensity, advanced age, comorbidities, elevated C-reactive protein, hypotension, leukemia as the cancer type, thrombocytopenia, and acute malnutrition. 16,17 For patients who meet criteria of low risk, a reduced intensity treatment regimen as an outpatient, has been shown to be safe, improve quality of life and reduce costs of care. 18 However, such stratification guidelines are based on young pediatric patients, rather than AYA and even within such regulations there remains an ongoing uncertainty as to the safety and efficacy of their applicability in all clinical settings. 18,19 Due to the uncertainty of these risk indices, treatment of FN with intravenous antibiotics, irrespective of underlying risk, is most common. 20,21 Further investigations are required to identify the feasibility of using risk factors, such as patient age, to risk stratify patients presenting with FN in order to further advance management, risk assessment and even consideration of preventative measures for cancer patients with this risk factor.
The main objective of this study is to assess the difference in quantity of FN presentations and severity between adolescents with cancer (AWC) and YCWC who present to our pediatric institution.
Given that our institution only treats patients up to 18 years of age, our study group of focus will be the AWC. The results of this study will be particularly pertinent to pediatric oncology clinicians who typically treat up to 18 years of age in their institutions.
The primary aim of this retrospective chart review was to compare the proportion of AWC to the proportion of YCWC who experienced FN following initial malignancy diagnosis to determine whether age is a risk factor for FN events. The secondary objective of this study was to identify the differences in severity of FN events in AWC and YCWC based on secondary variables including the duration of hospitalization, incidence of a bacterial or fungal cause of the event, incidence of a resulting PICU admission, duration of PICU admission, incidence of positive blood cultures and incidence of mortality due to FN.

| METHODS
We conducted a retrospective chart review of all patients diagnosed with a malignancy from January 1, 2013 to January 1, 2018 at the Children's Hospital of Eastern Ontario (CHEO). CHEO is a tertiary care, medium size institution in Ottawa, Canada that treats children from age 0-18 years old. This study was approved by the Research Ethics Board at CHEO. All malignancy types were included in the study, with no age restriction. Patients who presented with a relapse from a previous cancer diagnosis or who did not receive chemotherapy as part of their treatment regimen were excluded as their risk factors for severity and number of episodes of FN were felt to be greater or lesser, respectively.
For each enrolled patient, data on their sex, date of diagnosis, age at the time of diagnosis, primary oncologic diagnosis, treatment protocol(s), date of treatment completion, and current mortality status were collected from their health record. A FN event was identified at any time when the patient presented with an absolute neutrophil count (ANC) of less than 500 cells/mm 3 and fever according to institutional definition. Fever was defined as an axillary temperature of 38 C or greater measured once or a temperature between 37.6-37.9 C measured twice at timepoints an hour apart. If measured orally, patients were considered febrile at 38.5 C or greater measured once, or a temperature between 38.0-38.4 C measured twice after an hour's time. For each FN episode, information was collected on the date of presentation and age of the patient at that time. Severity was assessed by consideration of our secondary outcomes, which are the duration of hospital admission, the discovery of a source for the FN (bacterial/fungal), a resulting PICU admission and duration, the presence of a positive blood culture and mortality from the FN event.
Multiple FN episodes per patient were counted and assessed. The hospital length of stay variable was only calculated based on admissions where patients presented to the emergency department with FN, rather than recorded FN events that developed during a hospital stay for an alternate admitting diagnosis. The PICU admission and length of stay variables were calculated based on admissions to the PICU that were directly related to a FN event. The average length of stay for these variables was calculated with inclusion of a length of stay of zero.
Patient characteristics and outcomes were described using descriptive statistics. Continuous variables were summarized as means, standard deviations, and range, and categorical variables were summarized as frequencies and percentages. Due to the varying treatment durations and therefore the varied risk for the number of FN events during the course of individual patient treatments, the rate of FN was also quantified and described, which was calculated as the total number of each type of FN over the treatment duration in years.
Patients were sub-categorized by age into the AWC if they were 15 years or older up to 18 years old at the time of their cancer diagnosis or in the YCWC group if they had an age of less than 15 at the time of their cancer diagnosis. Quasi-poisson models were fitted to assess the relationship between age groups and FN complications, while controlling for key covariates of gender and diagnosis type (7 levels, leukemia set as reference category). In these models, the number of FN events was specified as the outcome, while the natural logarithm of treatment duration was specified as an offset parameter.
Effect sizes were expressed in terms of adjusted rate ratios and asso- Patients had varying cancer diagnoses of which the majority were leukemias (41.1%), followed by lymphomas (18.5%), primary brain tumors (15.1%), neuroblastoma (6.4%), sarcomas (6.0%), Wilms tumor (4.9%) and germ cell tumors (1.9%). The remaining diagnoses (6.0%) were grouped together due to their rarer occurrence. The proportion of varying diagnoses was significantly different (p < 0.001) between the AWC and YCWC groups, with lymphoma (36.8%) predominating in the AWC group and leukemia (39.2%) predominating in the YCWC group. The treatment duration (mean ± SD) for the YCWC of 1.85 ± 1.39 years was significantly longer (p < 0.001) than that of the AWC, which was 1.01 ± 0.93 years. Additional demographic and treatment characteristics are summarized in Table 1.
The mean frequency and mean rate of FN events and severity markers in the AWC and YCWC groups is summarized in      Table 3.
Among all patients enrolled in the study, there was 10.9% of mortality, made up entirely of the YCWC group, but no death was attributable solely to a FN event.

| DISCUSSION
We observed that compared to YCWC, AWC show a trend towards being more likely to develop FN given an adjusted rate ratio of 1.6 for the incidence of FN in AYC to YCWC. Additionally, when such events occur in this age group, there is a trend towards increased severity of FN as shown by the higher rates of PICU admission, with a rate of 0.07 in the AWC group compared to 0.03 in the YCWC group, and longer admission to PICU, when admission is secondary to their FN presentation, as shown through the statistically significant finding of a rate ratio of 4.9 when comparing duration of PICU admission in the AWC to the YCWC.
Given our findings that suggest that age may play a role in FN, and the known harmful effects of neutropenic sepsis, it is imperative that this group of patients is prioritized for their own set of guidelines that account for age in risk stratification. 10,11 Risk assessment guidelines are essential in establishing a standard for the extent of necessary investigations, type of empirical antibiotic therapy to be used with regards to the formulation in which they are provided, duration of antibiotic therapy, neutrophil recovery necessary prior to discharge or discontinuation of antibiotics, the need for prophylaxis throughout chemotherapy treatment and determination of whether treatment is conducted inpatient or outpatient. 14 There are numerous risk stratification tools (i.e., Rackoff, 23 Alexander, 24 PINDA, 25 SPOG, 26 Ammann 27 and MASCC 28  the YCWC population with a median age of 5-6 years of age. 18 Additionally, countless studies have shown that all rules perform inadequately in their ability to appropriately discriminate AYA patients as low or high risk. 19 The inconsistent performance of these tools across cohorts has left various panels aimed at developing guidance statements to advise that individual sites select which rules to use based on their available resources and population data, rather than evidence based superiority of rules. 29,30 The findings of this study do not suggest that all previous risk factors studied in the YCWC group be abandoned for the AYA population; they do however support the implementation of age into such existing guidelines in order to individualize them to this unique population.
Our findings which suggest that adolescence as an individual variable may increase both the incidence and severity of FN in the pediatric oncological population provide an opportunity in which age can be used to risk stratify younger patients with cancer to a lower risk group. 19 Similar approaches to this have been shown to be effective in the adult oncology population who present with FN but are classified as low risk. 31 These results also bring up the consideration for the use of prophylactic antimicrobials for AWC patients in future research as well as clinically. Further refinement of risk factors could help decrease the incidence of these mentioned events. With the addition of our findings that PICU admission duration in the AWC is significantly longer, early identification and treatment of this age group in particular may not only contribute to decreasing the likelihood of serious complications, but will also be more cost effective for institutions.
Age may play a role in the incidence and severity of FN due to both biological and social characteristics that are distinct to the adolescent population. As our study did not further classify the groups based on cancer diagnoses, but the distribution of varying malignancies did significantly vary between the groups, with lymphomas pre- social and demographic characteristics may also further inform on any differences noted between the different age groups.
In conclusion, this study highlights that differences in FN event occurrence and severity exist between AWC and YCWC. Further

CONFLICT OF INTEREST
The authors have stated explicitly that there are no conflicts of interest in connection with this article.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions.

ETHICS STATEMENT
This study was approved by the Research Ethics Board at the Children's Hospital of Eastern Ontario.