Association between Birthweight and Acute Lymphoblastic Leukemia in Children , a Systematic Review

BackgroundBirthweight is normally determined by a range of genetic traits and exposures occurring within the intra-uterine environment. Some epidemiological studies have reported high birthweight as a risk factor of Acute lymphoblastic Leukemia (ALL). Other studies have however not demonstrated this relationship.ObjectivesThe objective of this review is to assess the association between birthweight and Acute Lymphoblastic Leukemia in children. Search methods We searched observational studies from Cochrane, MEDLINE, EMBASE, ISI Web of Science, BIOSIS, the allied and Complementary Medicine Database and National Research Register, ClinicalTrial.gov, WHO International Trials Registry Platform. Selection criteriaWe included case control and cohort studies assessing the association between birthweight and ALL in children. All particpants below the age of 18 years (children) with ALL were included in the analysis. The independent variable in this review was birthweight. Birthweight was categorised into two (2). Birthweight >4kg (experimental arm) and ≤ (control arm).Data collection and analysisTwo reviewers independently assessed identified studies through two stages of screening. First, titles and abstracts of all references identified through searches were screened and irrelevant studies were excluded. Also, full texts of potentially eligible studies were further assessed according to previously defined inclusion criteria. Studies that did not meet inclusion criteria were excluded and reasons for their exclusion were stated. All studies that met the inclusion criteria were included. Areas of disagreement were resolved by a third-party review. Two review authors double checked the studies independently.Main resultsOut of the 348 studies screened, 16 of them met the inclusion criteria. A total of 3650728 participant provided data for analysis in this review. These studies were published between 1987 and 2018. The age span of studies was similar across studies (roughly (0-18 years). The vast majority of ALL was diagnosed before 15 years. 14 of the included studies were case control studies and 2 of them were cohort studies. Figure 1 presents odd ratio estimates for effect of birthweight on ALL (≤ 4000g vs. > 4000g). There was a statistically significant positive relationship between high birthweight (birthweight > 4000g) and risk of ALL (16 studies, OR 0.81, 95% CI 0.77, 0.85). Authors' conclusionsOur study revealed a significant positive relationship between high birthweight and ALL. Several studies have demonstrated an association between factors such as: high pre-pregnancy weight and height; gestational age greater than 42 weeks; parity greater than and high birthweight. Therefore, public health programs and interventions aimed at reducing the incidence of these maternal factors can reduce the risk of high birthweight and lower the incidence of ALL.

Background: Birthweight is normally determined by a range of genetic traits and exposures occurring within the intra-uterine environment. Some epidemiological studies have reported high birthweight as a risk factor of Acute lymphoblastic Leukemia (ALL). Other studies have however not demonstrated this relationship. Objectives: The objective of this review is to assess the association between birthweight and Acute Lymphoblastic Leukemia in children. Search methods: We searched observational studies from Cochrane, MEDLINE, EMBASE, ISI Web of Science, BIOSIS, the allied and Complementary Medicine Database and National Research Register, ClinicalTrial.gov, WHO International Trials Registry Platform. Selection criteria: We included case control and cohort studies assessing the association between birthweight and ALL in children. All participants below the age of 18 years (children) with Acute Lymphoblastic Leukemia (ALL) were included in the analysis. The independent variable in this review was birth weight. Birthweight was categorized into two (2): Birthweight >4kg (experimental arm) and ≤4kg (control arm). Data collection and analysis: Two reviewers independently assessed identified studies through two stages of screening. First, titles and abstracts of all references identified through searches were screened and irrelevant studies were excluded. Also, full texts of potentially eligible studies were further assessed according to previously defined inclusion criteria were excluded and reasons for their exclusion were stated. All studies that met Introduction Leukemia is a heterogeneous group of blood-forming cancers that comprise several biologically distinct sub-groups. It accounts for 2.6% of all cancers worldwide with 437,033 new cases diagnosed in 2018. (American Institute for Cancer Research, 2019) An estimated 176,200 people in the United States were expected to be diagnosed with Leukemia, lymphoma and myeloma in 2019, which would account for 10% of new cases of cancer (American Cancer Society, 2019). Leukemias, the most common cancer in children, account for almost 1 out of 3 cancers worldwide (Gatta et al., 2014). About 3 out of 4 leukemias in children and teens are Acute Lymphocytic Leukemia (ALL; Gatta et al., 2014).
Several genetic and environmental risk factors have been implicated in the pathogenesis of ALL (Petridou et al., 2002). Leukemia development begins with cellular mutation, usually chromosomal translocation, during embryogenesis (Greaves 2002;Rossig & Juergens, 2008). Cells with certain chromosomal translocation such as t(12;21), (4;11) or t(8;21) are frequently found at birth in children who later develop leukemia (Greaves, 2005;WIemels et al. 2002). Moreover, there is some evidence that perinatal exposures may pose risk for ALL in children. A recent meta-analysis of original studies from the Childhood the inclusion criteria were included. Areas of disagreement were resolved by a third-party review. Two review authors double checked the studies independently. Main results:Out of the 348 studies screened, 16 of them met the inclusion criteria. A total of 3,650,728 participants provided data for analysis in this review. These studies were published between 1987 and 2018. The age span of studies was similar across studies (roughly (0-18 years). The vast majority of ALL was diagnosed before 15 years. 14 of the included studies were case control studies and 2 of them were cohort studies. Figure 1 presents odd ratio estimates for effect of birthweight on ALL (≤ 4000g vs. > 4000g). There was a statistically significant positive relationship between high birthweight (birthweight > 4000g) and risk of ALL (16 studies, OR 0.81, 95% CI 0.77, 0.85). Authors' conclusions:Our study revealed a significant positive relationship between high birthweight and ALL. Several studies have demonstrated an association between factors such as: high pre-pregnancy weight and height; gestational age greater than 42 weeks; parity greater than and high birthweight. Therefore, public health programs and interventions aimed at reducing the incidence of these maternal factors can reduce the risk of high birthweight and lower the incidence of ALL.
A commonly investigated risk factor of ALL is birthweight. The observation that heavy babies (variously defined as weighing more than 3500g, 4000g and 4500g) are at risk of developing ALL has received a lot of attention (Johnson, Soler, Puumala, Ross & Spector, 2008;Caughey & Michels, 2009;Smith, Lightfoot, Simpson, & Roman, 2009). Higher birthweight is associated with maternal and fetal hyperglycemia as well as elevated blood insulin levels -all of which are factors that are linked with a higher risk of ALL (McLaughlin, Baptiste, Schymura, Nasca, & Zdeb, 2006;Caughey & Michels, 2009).
Some studies have estimated that the risk of ALL increases by 25% for children with birthweight of more than 4000g (Hjalgrim et al, 2003). Other studies have also reported the following odds ratios per kilogram increase in birthweight: 1.45 (Westergaard et al. 1997);1.29 (Petridou et al. 1997); 1.03 (Shuz et al 1999) and 1.09 (Shu et al. 2002). The evidence on birthweight and ALL has however been inconsistent, with other studies failing to establish significant relationship between birthweight and ALL (Reynolds, Behren & Elkin, 2002;Mckinney et al. 1999;Roman Ansell & Bull, 1997).
Pooled evidence from various meta-analyses are suggestive of low to moderate positive association between ALL and birthweight (Hjalgrim et al, 2003). However, the most recent meta-analysis we found was conducted more than a decade ago (Caughey & Michels, 2009). In light of recent evidence from empirical studies, it is important that this association between birthweight and ALL is reexamined We therefore, conducted a systematic review of available clinical evidence to date to assess the influence of normal and high birthweight on the risk of ALL.

Inclusion and exclusion criteria
We included case control and cohort studies that a) assessed the association between birthweight (independent variable) and ALL (outcome variable) b) used participants aged below 18 years c) reported the number of ALL cases using two categories of birthweight, i.e., >4kg (high) and ≤4kg (normal) based on the World Health Organization's (WHO) classification of birthweight d) were written in English. Studies with no abstract were excluded. We also excluded studies that reported birthweight as a continuous variable only and not as our predefined categories of high and normal birthweight.

Search methods for identification of studies
An electronic search was conducted on 6th October 2019 on the following databases: Cochrane, MEDLINE, EMBASE, ISI Web of Science, BIOSIS, the allied and Complementary Medicine Database and National Research Register, ClinicalTrial.gov, and WHO International Trials Registry Platform. Search terms included variants of "acute lymphocytic leukemia OR ALL" AND "birthweight". Full details of search terms can be found in, Table 2. The search strategy was adapted for each electronic database by using database-specific index terms. We also hand-searched reference lists from published studies, bibliographies of relevant systematic reviews and grey literature sources such as dissertations, theses and google scholar.

Selection of studies
The results of the electronic search were imported into Covidence (https://www.covidence.org). Two reviewers (SA and LA) independently assessed identified studies through two stages of screening. Firstly, titles and abstracts of all references were screened, and irrelevant studies were excluded. Secondly, full texts of tentatively eligible studies were further assessed against the pre-defined inclusion criteria. Areas of disagreement were resolved through discussions between the two reviewers and the reaching of consensus. The use of two independent reviewers ensured objectivity and transparency in the selection and synthesis of evidence.

Data extraction and risk of bias assessment
Data on study design, sample characteristics, exposure and outcomes were extracted independently by the two reviewers using the default data extraction form in Covidence. The extracted data were then transferred from Covidence to Review Manager 5 software for meta-analysis. We also assessed the methodological quality of included studies using the Critical Appraisal Skills Program (CASP) for case control and cohort studies as appropriate. The CASP checklist for case control and cohort studies are 11-and 12-items checklists that enable the reviewer to assess the methodological rigor and validity and reliability of evidence from case-control and cohort studies respectively.

Data synthesis
Since all our outcomes were dichotomous, we used odds ratios (OR) as our effect measure. Narrative synthesis was used to summarize the key findings of included studies. Outcome data from included studies were pooled for metaanalysis using the fixed effects model. We used the inverse variance method to estimate the combined effect size for the outcomes in the meta-analysis. We evaluated the statistical heterogeneity in the data using the I2 statistic (Higgins, 2003). We also evaluated clinical heterogeneity of included studies by assessment of study population characteristics. Funnel plots was used to assess reporting (publication) bias. We assessed funnel plots visually and used formal tests for funnel plot asymmetry.

Results of the search
The search on all electronic databases produced a total of 348 records, of which 187 were duplicates. The remaining 169 studies were screened for titles and abstracts and 129 records were found ineligible for the review. Sixteen other studies were excluded after full text screening. The remaining 16 studies that met the eligibility criteria were included in the current review (Figure 3) Characteristics of included studies A total of 3,650,728 participants provided data for analysis in this review. The studies originated from the USA (n=9), Australia, Germany, Azerbaijan, Denmark, Mexico, Sweden and Brazil. The studies were published between 1987 and 2018. The age span of studies was similar across studies (roughly 0-18 years). The vast majority of ALL was diagnosed before 15 years. Thirteen of the included studies were case control studies and 3 of them were cohort studies.

Synthesis of evidence
Figure 1 presents odds ratio estimates for effect of birthweight on ALL (≤ 4000g vrs > 4000g). All but six of the included studies individually showed a statistically significant positive relationship between high birthweight (birthweight > 4000g) and ALL. Evidence from the fixed effect meta-analysis model suggests 24% higher odds of ALL with children whose birthweight exceeds 4000g (N=16, OR 1.24, 95% CI 1.18, 1.34). The quality of evidence was rated moderate as the average of total for all domains of included studies was 13.5 (Table 1).

Discussion
We analyzed 16 epidemiologic studies of the association between birthweight and Acute Lymphoblastic Leukemia, including information on about 17155 children with Leukemia. The analysis demonstrated a significant increase of ALL in children with high birthweight (≤4000g vrs >4000g), corresponding to an odd ratio of 0.81 (CI: 0.77-0.85). The association between birthweight and risk of Leukemia was consistently observed in the studies conducted over a period of more than 30 years.
A previous meta-analysis conducted by Hjalgrim et al. (2003) also found a similar positive association for high birthweight and Acute Lymphoblastic Leukemia, reporting (>4000g vs. ≤ 4000g), odd ratio of 1.26 (CI: 1.17-1.37). Also, our result is consistent with the findings of Caughey & Michels, (2009) who found a non-significant positive association between high birthweight and ALL in children, (OR, 1.23 (95% CI: 1.15, 1.32). The article by Gholami, 2003 identified a relatively strong association between birth weight and ALL risk (corresponding to an OR, 0.45, (95% CI: 0.24, 0.86).
Various studies have shown that genetic events such as chromosomal translocation initiated in-utero can lead to acute leukemias (Ford at al., 1993;Gill et al. 1994;Wiemels, Ford, Van & Postma, 1999). Ultimately this process leads to the uncontrolled proliferation and accumulation of a single clone of immature lymphoblasts (Greaves, 1999) Birth weight has been associated with several growth factors such as insulin-like growth factor I (IGF I), insulin-like growth factor II (IGF-II) and sex-steroid hormones (Michels & Xue, 2006). Growth factors may increase the number of stem cells in utero, which increases the total number of replicating cells at risk for conversion into tumor cells, leading to leukemias (Trichopoulos & Lipworth, .1995). In addition, high birthweight resulting from high levels of growth factors in-utero may increase the risk of ALL by inducing proliferative stress on the bone marrow (Ross et al. 1996;Albanes & Winick, 1988).
In summary, this study confirms previously reported evidence of a positive association between high birth weight and overall leukemia risk. This result emphasizes the need for studies to clarify the biologic mechanisms underlying the association between birthweight and ALL.

Conclusion
Our study revealed a significant positive relationship between high birthweight and ALL. Several studies have demonstrated an association between factors such as: high pre-pregnancy weight and height (Janne, Tine, Ulrik & Niels, 2003); gestational age greater than 42 weeks (Ceska, 2005); parity greater than 2 (Wikstrom, Axelsson , & Bergstrom, 1991) and high birthweight. Therefore, public health programs and interventions aimed at reducing the incidence of these maternal factors will help reduce the incidence of ALL.