Safety of Procalcitonin Guided Early Discontinuation of Antibiotic Therapy among Children Receiving Cancer Chemotherapy and Having Low-Risk Febrile Neutropenia: A Randomized Feasibility Trial (ProFenC Study)

Abstract In low-risk febrile neutropenia (LR-FN), the safety of early discontinuation of empiric antibiotics without marrow recovery is not well established. This study aimed to evaluate the safety of procalcitonin (PCT) guided early discontinuation of antibiotics in LR-FN. In this trial, children with LR-FN with an afebrile period of at least 24 h, sterile blood culture, and negative/normalized PCT were randomized at 72 h of starting antibiotics into two groups: intervention arm and standard arm. The antibiotics were stopped in the intervention arm regardless of absolute neutrophil count (ANC), while in the standard arm, antibiotics were continued for at least 7 days or until recovery of ANC (>500/mm3). The primary objective was to determine the treatment failure rates, and the secondary objective was to compare the duration of antibiotics and all-cause mortality between the two arms. A total of 46 children with LR-FN were randomized to either the intervention arm (n = 23) or the standard arm (n = 23). Treatment failure was observed in 2/23 (8.7%) of patients in the intervention arm compared to 1/23 (4.3%) in the standard arm [RR: 2 (95% CI: 0.19–20.6); p = 0.55]. The median duration of antibiotics in the intervention arm and standard arm were 3 days vs 7 days (P= <0.001). There was no mortality in this study. PCT-guided early discontinuation of empirical antibiotics in LR-FN is feasible. There was no significant difference observed in treatment failure between the early discontinuation of antibiotics vs standard therapy. The total duration of antibiotic exposure was significantly lesser in the discontinuation arm. Further, larger multicenter studies are needed to confirm the finding of this study.


Introduction
Febrile neutropenia (FN) is one of the most common oncological emergencies.Early initiation of antimicrobials along with adequate supportive care, are the two vital components of the management of FN. 1 There are well-established guidelines for the initiation of empirical antibiotics for children presenting with FN.The choice of inpatient versus outpatient administration of antimicrobials and route of administration is usually dictated by adopting a validated risk stratification strategy which classifies children into low-risk and high-risk based on clinical and laboratory findings. 2nappropriate antimicrobial usage and dwindling anti-microbial discovery resulted in antimicrobial resistance among microorganisms to an alarming extent. 3In order to tackle this serious problem; there has been an active campaign for antimicrobial stewardship in recent years.It involves stringent efforts to reduce antimicrobial overuse by tailoring the antimicrobial prescription to the particular clinical setting. 4One of the most important hindrances to antimicrobial stewardship is the reluctance of clinicians to withhold antibiotics when the diagnosis is uncertain due to the non-availability of highly sensitive and specific diagnostic tools.Another challenge is the de-escalation of empirical antibiotics when a particular organism is not isolated.Thus, in this setting, biomarkers with high sensitivity and specificity to diagnose a bacterial infection as well as to monitor therapeutic response is found to be useful as they provide objective data to substantiate clinicians' decision to start, de-escalate or withhold antibiotics.Procalcitonin (PCT) is one of the various biomarkers that have been extensively studied as a potential tool to be incorporated into antibiotic stewardship programmes (ASP). 5Meta-analysis of 41 studies exploring the predictive role of various biomarkers among 8319 FN episodes (n = 4843 patients) revealed that baseline PCT at a cutoff of 0.5 ng/mL proved to be the most suitable biomarker to predict adverse events. 6Hence, high PCT levels are often used to predict poor prognosis, and PCT levels can be incorporated as a component of the risk stratification scoring system. 7When the infectious process is controlled by the host immune response or antimicrobial therapy, the PCT level decrease by 50% over 24 h.Therefore, PCT can be used to monitor the response to therapy, and it can be used as an additional aid to decide about the safe discontinuation of empirical antibiotics when there is a clinical improvement. 4ecently published pediatric international guidelines for fever with neutropenia (2023) recommends that in LR-FN patients who have been clinically well and afebrile for at least 24 h, discontinue empiric antibiotics if blood cultures remain negative at 48 h if there is evidence of marrow recovery (strong recommendation, low-quality evidence) or consider discontinuation of empiric antibiotics despite no evidence of marrow recovery (conditional recommendation, moderate-quality evidence). 8This remains low to moderate-quality evidence; moreover, there is no well-established cutoff for defining marrow recovery.Even though there is some evidence supporting the early discontinuation of empirical antibiotics in LR-FN, the safety of such early discontinuation without marrow recovery is not explored extensively in clinical trials. 2,9he utility of PCT in the setting of pediatric FN is limited only to diagnosis and prognosis.The PCT-based therapeutic monitoring in pediatric FN, though reported, is not utilized for therapeutic decision-making.Hence, this study aimed to evaluate the safety of PCT-guided early discontinuation of antibiotic therapy among children receiving cancer chemotherapy and having LR-FN.

Study site, duration, and patients
This study was conducted from February 2020 to October 2021 at the Division of Pediatric Oncology, Department of Pediatrics, All India Institute of Medical Sciences (AIIMS), New Delhi, India, a tertiary health care center with a dedicated Pediatric Oncology program in a lower-middle-income country (LMIC).Children aged ≤18 years diagnosed with cancer receiving cancer chemotherapy (received at least one cycle of chemotherapy) were enrolled at 72 h of starting antibiotics if they; a) were being treated for LR-FN with empirical antibiotics; with an afebrile period for at least 24 h; with a sterile blood culture at 48 h; and with a negative or normalization of PCT at 48 h.The exclusion criteria included patients receiving intensive phases of chemotherapy, clinically documented infections (CDI) which require a prolonged duration of antibiotics, whose malignancy is not in remission, patients with grade-3 or grade-4 mucositis, with profound neutropenia, severe organ dysfunction, and undergoing hematopoietic cell transplantation (HCT).Patients were also excluded when there was non-availability of a reliable and compliant caretaker, and their residence was >1 h of reach from the hospital.Written informed consent was obtained from the parent/legally authorized representative (LAR) and assent (if the child > 7 years) in all the enrolled cases.

Outcome assessment
The primary outcome of the study was to assess the difference in treatment failure rate in the first 14 days of randomization with an afebrile period for at least 24 h and sterile blood culture, and a negative/normalization of PCT at 48 h.The secondary outcomes of the study were to compare the duration of antibiotic exposure in 14 days of randomization and to assess the difference in all-cause mortality at 28 days of randomization between the two arms.

Definitions
Definitions used in this study have been provided in Supplementary Table 1.

Study design
This study was a randomized-open-labeled -parallel feasibility trial.

Screening and enrollment
All children aged ≤18 years diagnosed with cancer and receiving cancer chemotherapy who were being started on empirical antibiotics for FN, were screened for enrollment.

Risk stratification of FN and investigations
Children with FN in this study were classified as low risk as per previously published guidelines and our Department's FN protocol. 2,11,12The baseline investigations, including complete blood counts (CBC), serum electrolytes, renal function tests (RFTs), liver function tests (LFTs), chest x-ray (CXR), blood culture and sensitivity (C/S), and other microbiological samples, if required, were sent routinely as per our department's policy before initiation of empirical antibiotics.Additionally, a 2 milliliter (ml) serum sample at the time of presentation of FN and at 48 h after starting antibiotics was also taken for PCT estimation.The PCT samples were processed in the Department of Reproductive Biology, AIIMS, New Delhi.The samples were processed using ARCHITECT B.R.H.M.S PCT assay, a chemiluminescent microparticle immunoassay (CMIA), for the quantitative determination of PCT with a measuring interval ranging from 0.02 to 100 ng/ml.

Treatment of FN
Children with LR-FN were started on IV antibiotics as per institutional protocol, which includes combination therapy with antipseudomonal penicillin [cefoperazone + sulbactam (if platelet count was >20,000/mm 3 ) or piperacillin + tazobactam (if platelet counts were <20,000/mm 3 )] and/or an aminoglycoside (amikacin).Patients received these antibiotics at age and weight-appropriate dosages on an outpatient basis from the daycare facility.

Monitoring of FN patients
All participants were managed in the outpatient setting.The parents were instructed to record the temperature at least three times daily and also whenever the child was febrile and bring the temperature record for review.The patient was closely followed up clinically, and a repeat sample for CBC and PCT estimation was sent at 48 h of starting empirical antibiotics.

Screening for eligibility for randomization
Patients were reassessed at 72 h of starting empirical antibiotics along with CBC, PCT, and blood culture reports.Following patients were excluded from the study at 72 h; 1) Patients who had not been afebrile for at least 24 h, as these patients were eligible for 2 nd line antibiotics if clinically indicated; 2) Patients with a positive blood culture who were considered high risk as they would require a longer duration of antibiotics; 3) Patients who had a higher PCT value (cutoff < 0.25 ng/ ml) and patients whose PCT had not normalized within 48 h of empirical antibiotics since these patients received standard therapy with the continuation of antibiotics.

Randomization and allocation concealment
Those patients with negative/normalization of PCT at 48 h, an afebrile period of at least 24 h, and sterile blood culture were randomized into two groups, the intervention arm (IA) and the standard arm (SA).In the intervention arm, empirical antibiotics were discontinued.In the standard arm, antibiotics were continued for at least 7 days or until recovery of neutropenia, i.e.ANC >500/mm 3, which was ascertained by repeat CBC every 2 nd /3 rd day or as needed till 14 days of randomization.
We used computer-generated block randomization with variable block sizes to ensure an equal number of participants in each group.The randomization blocks, thus generated, were kept with an individual not involved in the enrollment of the subjects, administration of the intervention, or the analysis of the data.The randomization data was provided to an individual in the Department of Biostatistics, AIIMS, New Delhi, who labeled the envelopes as serial numbers corresponding to a given randomized patient number.The principal investigator (PI) opened the next serially numbered envelopes upon enrollment of the patient.Allocation concealment was ensured by using randomization and identical opaque sealed envelopes.The PI kept all the opened envelopes, along with the patient allocation details and date of opening the envelopes, for future reference.

Follow-up of enrolled patients
Patients in both arms were followed up clinically for 14 days after randomization for primary and secondary endpoints and for 28 days for all-cause mortality.The follow-ups were either physical or telephonic, depending upon the patient's scheduled visit for chemotherapy.

Sample size
The reported failure rate with standard therapy of continuation of empirical antibiotics till ANC recovery is varied from 5.4% to 31%, [13][14][15][16][17] with heterogeneity in an inpatient or outpatient settings of antibiotic administration as well as oral or intravenous (IV) route of antibiotics administration.The majority of the studies reported a failure rate between 10-20% among children receiving standard therapy with the continuation of empirical antibiotics till ANC recovery.Since this was a feasibility trial of PCT-guided discontinuation of antibiotics in children with LR-FN, hence, we decided to enroll a minimum of 20 patients in each arm.

Statistical analysis
Data analysis was done using STATA/SE 14.2 (Stata Corp, College Station, Tx).Descriptive statistics were used to analyze the demographics and clinical characteristics of the patients.Frequencies (%) were obtained for categorical variables.Mean (±SD) and median (IQR) were calculated for continuous variables.Comparison between categorical variables was performed using the Pearson Chi-square test or Fisher's exact test and between continuous variables using the t-test or Wilcoxon rank-sum (Mann-Whitney) test.All analyses were done on an intention-to-treat (ITT) basis.Two-sided p < 0.05 was considered statistically significant.

Results
Between February 2020 and October 2021, a total of 175 patients who were started on empirical IV antibiotics were screened for eligibility.Out of those 175 patients screened, 111 patients did not meet the inclusion criteria, as they had one or multiple features that classified them as high-risk FN, which rendered them ineligible for this study.Out of the remaining 64 children who were followed up and screened for randomization at 72 h of starting antibiotics, 18 children did not fulfill the eligibility criteria for randomization (n = 6 patients had an afebrile period of <24 h; n = 11 patients did not achieve either negative PCT or normalization of PCT and, n = 1 patient had both of the above-mentioned features).
The remaining 46 patients were enrolled in the study and randomized at 72 h of starting antibiotics.Twenty-three patients were randomly allocated to the intervention arm whose antibiotics were stopped at 72 h.The remaining 23 patients were randomly allocated to the standard arm and received standard therapy with the continuation of antibiotics for at least 7 days or until recovery of ANC.None of the patients was lost to follow-up.Adherence to the study protocol was observed in 100% of patients.All 46 randomized patients were included in the final analysis.The CONSORT diagram of the study is presented in Figure 1.

Baseline and clinical characteristics
The baseline characteristics were all well-balanced between the two arms (Table 1).The baseline clinical and laboratory characteristics of the enrolled patients (n = 46) captured at the presentation of FN are shown in Table 1.The clinical and laboratory characteristics at the time of randomization (72 h after starting antibiotics) are shown in Table 2.
At the time of randomization, 19/46 (41%) patients had marrow recovery.In the standard arm, 48% of patients, while in the intervention arm, only 35% had marrow recovery (p = 0.369) at randomization.

Primary outcome
Recurrence of fever A total of 5 patients had a recurrence of fever during the follow-up for 14 days [D2 (SA), D5 (SA), D5 (IA), D10 (SA), and Day 14 (IA)], and only one patient had a recurrence of fever within 72 h of randomization from the standard arm, who required escalation of antibiotics.

Hospitalization
A total of two patients from the intervention arm required hospitalization for antibiotic administration, and both patients were hospitalized within 14 days of randomization (D5 and D4 of randomization).No patient from the standard arm required hospitalization.

Secondary outcomes: Antibiotic exposure and all-cause mortality
The median duration of antibiotics in the intervention arm and standard arms was 3 days vs 7 days (P= <0.001).The median antibiotic exposure from its starting (3 vs 7 days, P= <0.001) and from randomization to 14 days was lesser (0 vs 4 days, P= < 0.001) in the intervention arm.The mean antibiotic-free period was significantly higher (13.56 vs 9.87 days, P= < 0.001) in the intervention arm in the 14 days after randomization (Table 3).None of the study subjects died due to any cause during the follow-up of 28 days.

Discussion
The CCLG (Children's Cancer and Leukemia Group) supportive care group supports SIOP-endorsed international pediatric FN guidelines and recommends to use of a combination of two oral antibiotics in the LR-FN. 18Since 2005, the CCLG was undertaking audits of FN care in the UK prior to the use of NICE (CG151) guidelines. 19he NICE guidance recommends discontinuation of antibiotic therapy irrespective of the ANC in patients whose FN responds to treatment. 19The NICE guidelines also recommend switching from IV antibiotics to oral administration after 48 h in patients whose risk of developing septic complications is reassessed as a low-risk. 20n the current study, in patients with LR-FN who remained afebrile for at least 24 h with negative blood culture and negative or normalization of PCT levels, there was no difference observed in treatment failure between the discontinuation of antibiotics at 72 h and the standard practice of continuing the antibiotics till marrow recovery.Klaassen et al., 17 conducted a trial among children with LR-FN and documented that there is no significant difference in the readmission rates between the early stoppage arm and continuation arm, similar to our findings.The criteria used to define LR-FN in this study were similar to those utilized in the current study.In their study, 77% of patients had marrow recovery at the time of randomization, with significantly higher readmission rates among those without marrow recovery.In the current study, 41% of patients had marrow recovery at the time of randomization.
Kumar et al., 14 compared the clinical success rate (defined as the proportion of patients remaining afebrile till ANC ≥ 500/mm 3 ) between the early stoppage arm and continuation arm, and there was no significant difference in success rate between the two arms (94.7% vs 94.6%; p = 0.682).The criteria for LR-FN were similar to that used in the current study, except that Kumar et al., 14 included those with profound neutropenia and those receiving intensive chemotherapy like acute myeloid leukemia (AML), whereas we excluded them.Interestingly, all four events of treatment failure reported in their study occurred in those with profound neutropenia at baseline.Kumar et al., randomized the study subjects when they became afebrile for at least 24 h.The median duration of antibiotics at the time of randomization was 60 h in both arms, which is comparable to the antibiotic exposure in our study prior to randomization.But some proportion of patients might have received more than 72 h of antibiotics at the time of randomization, which is the time point of randomization in our study.
In the current study, we used treatment failure as the primary outcome, which was a composite of unfavorable outcomes used in previous studies (recurrence of fever, hospitalization, and infection-related mortality), in order to capture the entire consequence of our intervention.Similar composite endpoints combining clinical, laboratory, or microbiological criteria were utilized to define clinical failure in the two previous studies conducted by Santaloya et al. 21,22 We identified that the treatment failure in both arms in the current study did not differ significantly [RR: 2 (95% CI: 0.19-20.6)].Similarly, Stern et al., 23 in a systematic review, did not find any significant differences in the rates of clinical failure between early discontinuation and continuation arm (RR 1.23, 95% CI 0.85 to 1.77; very low-certainty evidence).Our study results tend to correlate with the results of previous studies in this regard.The summary of treatment failure in various pediatric studies is depicted in Table 4.
The total duration of antibiotic exposure from initiation to 14 days of randomization was significantly lesser among the intervention arm (3 days vs 7 days; P= < 0.001).Similar results were obtained in a study conducted by Santolayo et al. 22 among children with LR-FN (3 days vs 7 days; P= < 0.0001).But Santolayo et al. 22 included the children with respiratory viral infections, whereas we excluded patients with any focus of infection who would require prolonged antibiotics. 24The difference in the median duration of antibiotic exposure between the two arms in the current study was four days, well within the range documented in the systematic review done by Stern et al. 23 The comparison of antibiotics duration with other pediatric studies is shown in Suppl.Table 2.
Concern for safety is the major factor that precludes early discontinuation of antibiotics for LR-FN prior to hematologic recovery.We tried to address the safety concerns associated with such an early discontinuation strategy in our study.Even though there are no well-validated risk stratification criteria for pediatric FN, we used a risk stratification model that combines various patient, disease and episode-related factors. 2 In addition to clinical and routine hematological parameters utilized for risk stratification, we also incorporated a biomarker (PCT) to improve the performance of risk stratification criteria and identify patients with increased risk of infectious complications.We wanted to strengthen the clinical decision rule with the addition of a PCT-guided protocol to ensure the safety of de-escalation.This might be important in our setting, where the culture-positive rate is very low (6%). 24Incorporating a PCT within the risk stratification can identify a subset of patients with a high risk for serious bacterial infection.
Haddad et al., 25 compared the outcomes of early discontinuation of antibiotics (<7 days) with that of prolonged antibiotic therapy (>7 days) among those with FN with low PCT levels at baseline or decreasing PCT levels.In this study, both groups had similar clinical outcomes.Lima et al. 26 compared the utility of PCT-guided protocol to decide on the duration of empirical antibiotics among adults with hematological malignancy and FN, with the standard practice of continuing antibiotics until hematological recovery.The empirical antibiotics were stopped in the intervention arm when the patients became afebrile (for two consecutive days if non-neutropenic and for three consecutive days if neutropenic) or when PCT levels were either negative (<0.5 ng/ml) or normalized (≥ 90% reduction from baseline) for two consecutive days.In this study, no significant difference was documented in clinical success rate, infection relapse, hospitalization duration, and mortality between the two groups.
Of the four previously conducted randomized studies among children examining the difference in treatment failure between the early discontinuation of antibiotics and standard therapy, meaningful conclusions with adequate power could not be drawn from three single-center studies. 14,17,21,22This reflects the need for well-designed, adequately powered multicenter randomized control studies using clear and consistent definitions and risk stratification, to throw some light on this subject.This may help in reducing the cost and appropriate utilization of resources, which are of major concern in resource-limited settings.More importantly, such early de-escalation could contribute toward antimicrobial stewardship resulting in the judicious use of antibiotics against rising antibiotic resistance and dwindling antibiotic development.
Our study's strength is that one of the first randomized studies incorporated PCT-guided protocol in an attempt at early discontinuation of empirical antibiotics among children with LR-FN.A sincere attempt to address the safety concern for early de-escalation strategy using a stringent risk stratification model strengthened with the addition of a sensitive biomarker, PCT, also adds to the strengths of this study.This study used a clearly defined outcome which is a composite of previously reported poor outcomes associated with early stoppage of antibiotics, which upscales the value of this study.Strict adherence to the protocol, a predefined follow-up plan, and careful monitoring enhance the weight of this study.
Our study is not without limitations.The major limitation of our study is poor accrual during the stipulated study period.This is partly due to stringent eligibility criteria for enrollment and partly due to the COVID-19 pandemic-related restriction of services and modified department's policy. 27Secondly, we didn't exclude those with marrow recovery at the time of randomization, which might decrease the clinical difference between the two arms.
In conclusion, this study reports that in patients with LR-FN who remained afebrile for at least 24 h with negative blood culture and negative or normalization of PCT levels, there was no significant difference observed in treatment failure between the early discontinuation of antibiotics and the standard practice of continuing the antibiotics till marrow recovery.The total duration of antibiotic exposure from initiation to 14 days of randomization was significantly lesser among the intervention arm.Further, well-designed, adequately powered, randomized control studies are needed to support this study's findings.

Table 1 .
Baseline clinical and laboratory characteristics of patients at the time of screening (at the presentation of febrile neutropenia) (n = 46).

Table 2 .
clinical and laboratory characteristics of patients at the time of randomization (72 h of starting antibiotics) (n = 46).

Table 3 .
comparison of treatment failure and duration of antibiotics between intervention and standard arm (n = 46).

Table 4 .
comparison of treatment failure of the current study with previous other pediatric studies.