Adverse Drug Reactions in Pediatric Oncohematology: A Systematic Review

Introduction: Adverse drug reactions (ADR) are an important cause of morbidity and mortality in pediatric patients. Due to the disease severity and chemotherapy safety profile, oncologic patients are at higher risk of ADR. However, there is little evidence on pharmacovigilance studies evaluating drug safety in this specific population. Methods: In order to assess the incidence and characteristics of ADR in pediatric patients with oncohematogical diseases and the methodology used in the studies, a systematic review was carried out using both free search and a combination of MeSH terms. Data extraction and critical appraisal were performed independently using a predefined form. Results: Fourteen studies were included, of which eight were prospective and half focused in inpatients. Sample size and study duration varied widely. Different methods of ADR identification were detected, used alone or combined. Causality and severity were assessed frequently, whereas preventability was lacking in most studies. ADR incidence varied between 14.4 and 67% in inpatients, and 19.6–68.1% in admissions, mainly in the form of hematological, gastrointestinal and skin toxicity. Between 11 and 16.4% ADR were considered severe, and preventability ranged from 0 to 74.5%. Conclusion: ADR in oncohematology pediatric patients are frequent. A high variability in study design and results has been found. The use of methodological standards and preventability assessment should be reinforced in order to allow results comparison between studies and centers, and to detected areas of improvement. Systematic Review Registration: https://www.crd.york.ac.uk/prospero/display_record.php?RecordID=96513, identifier CRD42018096513.


INTRODUCTION
Adverse drug reactions (ADR) have been defined by the World Health Organization (WHO) as "a response to a drug which is noxious and unintended, and which occurs at doses normally used in man for the prophylaxis, diagnosis, or therapy of disease, or for modification of physiological function" (WHO, 1972).
ADR are an important cause of morbidity and mortality in patients of all ages, including pediatric population, and are considered a public health problem worldwide (Impicciatore et al., 2001;Clavenna and Bonati, 2009;Thiesen et al., 2013;Durrieu et al., 2014;Ramos et al., 2021). Children are more susceptible to ADR owing to insufficient standardized information, unlicensed and off-label use, unavailability of pediatric formulations, and physiological peculiarities inherent to age (Ramos et al., 2021).
Different systematic reviews and meta-analysis including ADR observational studies have found an incidence of ADR in pediatric inpatients ranging 0.6-16.8%, from 1.8 to 2.09% leading to hospital admission and 1.0-1.46% in outpatient setting (Impicciatore et al., 2001;Clavenna and Bonati, 2009;Smyth et al., 2012). In addition, ADR prevention in outpatients remains a public health and a patient safety challenge (Lombardi et al., 2018). A systematic review including 102 articles assessed preventability in only 19, which ranged from 7 to 98%. This high variability was explained due to a high heterogeneity in study designs, methods and settings (Smyth et al., 2012).
Risk factors for ADR in children are poorly characterized Lombardi et al., 2018). Age on admission, number of drugs, off-label drug use, and oncology diagnosis and treatment have been described as ADR risk factors Thiesen et al., 2013). Moreover, one of these studies stated the risk in oncology patients and found an increased risk for ADR (OR 1.89 [95% CI 1.36-2.63]) .
Chemotherapy toxicity is a common cause of morbidity and mortality in most pediatric cancer patients, and a frequent cause of mid and long term sequel (Conyers et al., 2018). Even though drugs used in cancer diseases are described as a risk factor of ADR occurrence, and that ADR are frequent in oncology and hematology hospitalization wards, there are very few studies that have quantified or analyzed any of these aspects in pediatric population.
Oncohematological diseases have a high impact on children and their families, and on their quality of life. Improving the knowledge of ADR incidence, characteristics and preventability can be useful to compare results between studies and centers and to detect improvement areas, as a way to offer quality healthcare. Our aim was to perform a systematic review in order to describe the incidence and characteristics of ADR in pediatric oncology and hematology patients, to describe the methodology used in the included studies and, if possible, to identify preventive actions in order to minimize ADR occurrence.

Study Design
A systematic review of observational studies that evaluated the prevalence, incidence and/or characteristics of ADR in pediatric oncohematology was performed. This study was conducted in accordance with the recommendations of the Joanna Briggs Institute (Munn et al., 2015) for systematic reviews of observational epidemiologic studies that evaluate prevalence and incidence data, and the PRISMA recommendations for systematic reviews (Tricco et al., 2018). This study was registered (CRD42018096513) at PROSPERO systematic review database.

Systematic Literature Search
A systematic literature search was carried out in PubMed from inception to 31st December 2020, both using free search and the combination of different MeSH terms ("Pediatrics," "Neoplasms," "Hematology," "Antineoplastic agents," "Drugrelated side effects and adverse reactions," "Iatrogenic disease," "Prevention and control," "Medical oncology," and "Primary prevention"). References of the articles assessed for eligibility were also reviewed and included if considered relevant.

Inclusion and Exclusion Criteria
Studies that described the incidence and/or characteristics of ADR in pediatric oncohematology patients or in pediatric population with a differentiated oncohematology subgroup were included in this systematic review. Articles describing infectious outbreaks related to immunosuppression, data from national or international clinical databases of spontaneous pharmacovigilance reporting systems and pharmacovigilance studies including one single drug or specific ADR were excluded. No language or other search filters were applied.

Screening and Data Extraction
All articles were screened independently by two authors (KA-H, ID) to identify relevant studies based on titles and abstracts, and on full texts of potentially relevant papers if study relevance could not be determined from the titles and abstracts. For studies meeting inclusion criteria, data were extracted independently using a standardized data collection form defined and agreed previously. Data extracted included article identification, methodology characteristics (study design, setting, study aim, ADR definition and detection method, and causality, severity and preventability scales used), and relevant results (sample size, study duration, population characteristics, ADR frequency and description, severity and preventability). A third author (AA) participated in the review and in the data extraction in case of disagreement.

Data Analysis and Quality Assessment
\This review focuses on both the incidence of ADR in a high-risk population and on the methodological characteristics of the studies included. Quality assessment was performed independently by two authors (KA-H, ID), using a scale designed and previously published (Laatikainen et al., 2017), available in the Supplementary Material. The scale includes six questions related to study design, study population, ADR definition and identification, causality assessment and result description. Each question can be evaluated as 0 or 1, where 0 indicates the poor quality of the study regarding that item. A third author (AA) participated in the critical appraisal in case of disagreement.

Methodological Results
Eight observational studies collected data prospectively, whereas six were performed retrospectively. Seven studies focused on hospitalized patients, four included admissions related to ADR and three analyzed both settings. No studies assessing outpatient setting were found. Twelve studies evaluated ADR, predominantly using WHO or Edward and Aronson definitions, and two studies used adverse drug events (ADE). Causality was estimated in nine studies, using mainly Naranjo and WHO-UMC scales. Severity was assessed in 11 studies, mostly using Hartwig et al. scale and NCI CTCAE criteria. Finally, preventability was only evaluated in five studies out of 14, using Shumock and Thorton in two of them. Ten studies used a single ADR detection method, and four studies used a combination of them: intensive monitoring chart review method was used in seven studies, chart review was used in four studies, and three studies based their results in triggers.
Critical appraisal is summarized in the supplementary material. Most of the studies defined adequately the study population and stated the causality assessment tool used (questions 2 [Q2] and 5 [Q5]). In contrast, results were considered not clearly described in half of the studies (Q6), as the information provided by the original articles on number of patients or ADR was missing. In addition, study design (Q1), ADR definition (Q3), and ADR detection method (Q4) were not clearly mentioned in three studies.

Clinical Results
Sample size varied from 52 to 10,297 patients, as well as study duration, which ranged from 30 days up to 12 years. Age was expressed in means in five studies, as median in four or with percentage of patients in an age range (2-12 years old) in three; age values can be found in Tables

DISCUSSION
Even though ADR are an important cause of morbidity and mortality, are frequent in oncology and hematology, and chemotherapy is described as a risk factor, only 14 studies that assessed ADR were found. Incidence rates ranged from 14.4 to 61.3% in hospitalized patients and 19.6-68.1% in ADR leading to admission. A high heterogeneity in methodological aspects reviewed was also described and has likely influenced on the observed results. To our knowledge, this is the first systematic review on pharmacovigilance regarding pediatric oncology and hematology.
As mentioned previously, a high variability regarding methodology was found in almost every aspect of study design: sample size, study duration, study setting, population of interest, ADR detection method, the assessment of severity and preventability, and the different scales used. These findings could be explained by the different aim of each study, the effort to adapt the study to each local environment and available resources, research experience of the team and the moment in which they were designed and carried out, since methodology has evolved over time. These methodological differences have probably influenced on the clinical results found. A systematic review on ADR detection methods in hospitalized children was carried out (Ramos et al., 2021) and found that methods such as intensive monitoring chart review or trigger tools are more effective but time consuming, whereas spontaneous notification showed the lowest rate of detection. They concluded that most of the studies used a combination of methods, which might indicate a growing concern in ADR care in hospitalized children. This improvement in combined methods for ADR detection was previously suggested (Gonzalez-Gonzalez et al., 2013).
To our knowledge, there is no reference quality assessment tool for observational studies with other designs than cohort or casecontrol studies. A systematic review (Katrak et al., 2004) pointed out the variability in 121 published critical appraisal tools, regarding its intent, components and construction; this finding was later confirmed in another systematic review (Page et al., 2018), which concluded that there are several limitations of existing tools for assessing risk of reporting biases. STROBE statement (von Elm et al., 2007) or Johanna Briggs Institute (Munn et al., 2015) critical appraisal checklists are the most known tools, but their application was complex and troublesome. Therefore, the choice of the checklist used in this systematic review (Laatikainen et al., 2017) was agreed by the research team due to the lack of a standardized tool, its suitability to the type of studies included in the systematic review and to the aim of the critical appraisal analysis, and its easy application. The main area of improvement was the presentation of results, as results were insufficient or missing in half of the studies, and therefore it was considered to be the aspect most susceptible to introduce bias. Moreover, an adequate study design statement, ADR definition and identification clearly mentioned would likely reduce the risk of bias and improve study quality. Ten studies were published after the STROBE statement, which suggests a need to reinforce the use of these tools both during study design and manuscript drafting. Incidence described in oncohematology pediatric patients was higher, in contrast with studies in pediatrics, which described an overall rate of ADR of 9.53 and 2.09% (hospitalized and admission, respectively) (Impicciatore et al., 2001). This finding is expectable and consistent with chemotherapy safety profile and ADR risk factors, such as cancer diagnosis or number of concomitant drugs. Moreover, it is likely that the use of different scales in causality and severity assessment has influenced on the results observed too.
Hematological, gastrointestinal and skin toxicities are the most frequently described ADR in the articles included, which are in tune with the expected safety profile of conventional chemotherapy. No studies with novel drugs such as monoclonal antibodies or tyrosine kinase inhibitors were found up to 2020. A recently published study (Amaro-Hosey et al., 2021) prospectively assessed drug safety with some specific therapies, including novel drugs and conventional chemotherapy. The most frequent ADR were hematological, infections and gastrointestinal. Incidence using days at risk was calculated CT and antimicrobials were the drugs most frequently related to ADR. Gastrointestinal and hematologic ADR were the most frequently described during hospitalization. ADR most frequent during admission were nausea and vomiting with cyclophosphamide (9), cytosine arabinoside (6)   regarding novel therapies: 1.1 and 5.3 ADR/100 days at risk for blood disorders and 0.8 and four ADR/100 days at risk for infections, related to pegaspargase and thioguanine respectively; and 0.6 ADR/ 100 days at risk for infections attributed to rituximab. Only four out of 14 studies included in the systematic review reported fatal cases, and the global incidence of fatal cases could not determined because the total population was not specified in two studies (Le et al., 2006;Morales-Ríos et al., 2020). This finding has been previously described and could either suggest that fatal ADR are very rare in children or are frequently underreported or not suspected (Bouvy et al., 2015). ADR preventability is a key aspect to analyze, in order to identify areas of improvement to reduce ADR occurrence and improve patients' life quality. A systematic review (Smyth et al., 2012) identified that preventability was only assessed in 19 out of 102 studies, and ranged from 7 to 98%. This finding is similar to result obtained in the current systematic review, which evidences that it's an aspect poorly evaluated in pharmacovigilance studies and therefore should be encouraged.
This systematic review tries to add some evidence on an important health problem insufficiently studied that affects a fragile population. Summarized data on characteristics and incidence of ADR in this population is provided, as well as a methodological description in order to find areas of improvement. Defined inclusion and exclusion criteria, the selection of studies in pediatrics with an oncohematology subgroup, the lack of non-standardized critical appraisal tool that fitted the study characteristics and the use of a selected/concrete critical appraisal tool may have introduced bias, but was agreed and considered appropriate to enrich the results and the discussion. Great heterogeneity makes it difficult to compare results, but can also be interpreted as a need to establish methodology standards or to reinforce their use during study design and manuscript drafting, such as STROBE statement. Ultimately, our aim should be to provide a high quality research and healthcare to our patients and to improve their quality of life, regarding drug efficacy and safety.
In conclusion, ADR in oncohematology pediatric patients are more frequent than in general pediatric population, as expected. A high variability in study design and results has been found, which indicates a need to reinforce the use of methodological standards both in study design and manuscript drafting, in order to allow comparability between studies and to identify areas of prevention and improvement. Preventability assessment should be strongly encouraged in order to provide a high quality healthcare and to improve patients' quality of life.

DATA AVAILABILITY STATEMENT
The original contributions presented in the study are included in the article/Supplementary Material, further inquiries can be directed to the corresponding author.

AUTHOR CONTRIBUTIONS
KA-H and ID contributed to the conception and design of the study, and contributed to the recording of the data. Substantial contribution to the analysis or interpretation of data for the work was made by KA-H, ID and AA. KA-H wrote the first draft of the manuscript. All authors substantially contributed to the manuscript revision, read, and approved the submitted version.