Role of Maternal Diet in the Risk of Childhood Acute Leukemia: A Systematic Review and Meta-Analysis

Many studies have investigated the etiology of acute leukemia, one of the most common types of cancer in children; however, there is a lack of clarity regarding preventable risk factors. This systematic review and meta-analysis aimed to summarize the current evidence regarding the role of maternal dietary factors in the development of childhood leukemia. All epidemiological studies published until July 2022 that evaluated maternal dietary risk factors for childhood acute leukemia were identified in two electronic databases (PubMed and Web of Science) without limits of publication year or language. A total of 38 studies (1 prospective cohort study, 34 case-control studies and 3 studies with pooled analysis) were included. The published risk estimates were combined into a meta-analysis, using the Generic Inverse Variance method. The maternal consumption of fruits (two or more daily servings vs. less) was inversely associated with acute lymphoblastic leukemia (odds ratio = 0.71; 95% CI, 0.59–0.86), whereas maternal coffee intake (higher than two cups per day vs. no consumption) was associated with an increased risk of acute lymphoblastic leukemia (odds ratio = 1.45; 95% CI, 1.12–1.89). Despite these findings, more high-quality research from cohort studies and the identification of causal factors are needed to develop evidence-based and cost-effective prevention strategies applicable at the population level. Review Registration: PROSPERO registration no. CRD42019128937.


Introduction
Acute leukemia is one of the most frequently diagnosed cancers among children and adolescents. Its worldwide incidence is approximately 46 cases per million per year in children aged 0-14 years and approximately 29 cases per million in adolescents aged 15-19 years [1]. In many populations of children aged 0-14 years, almost 80% of cases have acute lymphoblastic leukemia (ALL), while 15% have acute myeloid leukemia (AML) [2].
The etiology of childhood acute leukemia is linked to X-ray radiation exposure, specific forms of chemotherapy and some genetic syndromes [3]. However, all these risk factors together may explain only 10% of cases [4]. Although acute leukemia is the most common cancer in children and adolescents, it is a rare disease; therefore, studying its etiology remains challenging. Nonetheless, a causal mechanism for the most common variant (ALL) has been proposed, which incorporates both antenatal and postnatal factors [5].
Considering the differential incidence of acute leukemia by geographic region worldwide, the potential roles of lifestyle and environmental factors in its etiology have been proposed [1]. A peak incidence occurs at approximately two-five years of age, and it is reasonable to assume that factors in the prenatal or perinatal period affect the pathogenesis of childhood leukemia [3,6]. Indeed, several studies have demonstrated an increased risk of leukemia due to maternal smoking and alcohol consumption during pregnancy [7,8]. However, little is known about the impact of other lifestyle factors such as the mother's overall diet during pregnancy on the risk of childhood leukemia.
The current systematic review and meta-analysis focused on the role of maternal diet in the risk of childhood acute leukemia. It provides an overview of existing data linking the influence of maternal diet, supplementation and the consumption of non-alcoholic beverages to the development of acute leukemia in children. We used combined evidence to assess the potential impact of maternal diet on the risk of childhood acute leukemia.

Definition of the Outcome
The main outcome of our systematic review is acute leukemia (AL) incidence in children aged 0-14 years and adolescents aged 15-19 years. The leukemia types considered included acute lymphoblastic leukemia (ALL) and acute non-lymphoblastic leukemia (ANLL), mainly acute myeloid leukemia (AML).

Definition of the Risk Factors
The exposures included maternal dietary and non-alcoholic drink intake and supplements consumed during the peri-pregnancy period.

Systematic Review Registration
Details of the protocol for this systematic review were registered on PROSPERO, registration number CRD42019128937.

Search Strategy
To identify the maternal dietary factors that have been studied in relation to childhood leukemia risk, we performed a multi-tiered electronic search of PubMed and Web of Science. First, an explorative search for studies on leukemia in infants, children and adolescents was conducted in December 2017; this was not restricted by language or publication date. A total of 66 studies were identified in this search. Based on the results of this first search, we were able to narrow down the keywords further and perform a more targeted search in July 2022, which focused on some identified maternal exposures such as dietary intake (food groups and micronutrients), consumption of non-alcoholic beverages (coffee, tea and soft drinks) and supplements used during pregnancy, and no limit of publication year or language restriction was used. For all constructs, database-specific thesauri were used to identify the relevant synonyms. The search strategy is illustrated in Figure 1.

Selection of the Studies and Quality Assessment
Studies were eligible for inclusion if they reported effect estimates (relative risk (RR), odds ratio (OR) or hazard ratio (HR) and 95% confidence intervals (CI)) or if they provided sufficient data to calculate the estimates for different levels of exposure. Table 1 lists the inclusion and exclusion criteria used in the review. The PRISMA guidelines were followed [9] (Supplementary Table S1). All references found through the search questions ( Figure 1) were imported into the Endnote software program (Version x8, Clarivate Analytics, Philadelphia, PA, USA). Two authors independently searched each database (Pubmed and Web of Science) for publications. Any uncertainty regarding the selection of a given article was resolved after screening by a third reviewer (IH) and discussed in the working group, if needed. The final eligibility criteria for the included studies are listed according to the PICOS (Table 2), and the selected studies included for each extracted factor are summarized in Table 3.

Selection of the Studies and Quality Assessment
Studies were eligible for inclusion if they reported effect estimates (relative risk (RR), odds ratio (OR) or hazard ratio (HR) and 95% confidence intervals (CI)) or if they provided sufficient data to calculate the estimates for different levels of exposure. Table 1 lists the inclusion and exclusion criteria used in the review. The PRISMA guidelines were followed [9] (Supplementary Table S1). All references found through the

General inclusion criteria
All studies describing the maternal dietary factors and their influence on the incidence of acute leukemia in children and adolescents were included in the systematic review. The risk factors considered were maternal nutrition (food and beverages) and supplement use during pregnancy. Pooled analysis including new data was also included in the review.  Subsequently, quality control of the selected articles was performed using the checklist by Fowkes et al. to ensure the inclusion of high-quality studies [10]. This checklist includes questions on the study design and sample, control group characteristics, quality of measurements and outcomes, completeness and influences of distortions. A major flaw identified for a given criterion is flagged by the sign ++. The + sign was applied to a criterion with a minor flaw. In case of no flaws, nil (0) was applied. If a study scored ++ on more than one criterion of this checklist, it was excluded ( Figure 1). The quality control results of the selected articles are shown in Supplementary Table S2.

Quantitative Meta-Analysis
A quantitative analysis of the risk of AL, ALL, or AML in children was performed to determine the risk factors that met the eligibility criteria for the meta-analysis. For each included factor, the choice of performing an additional quantitative analysis was based on the number (at least two studies) and homogeneity of the included studies (e.g., results from studies restricted to infants or from univariate analysis with no adjustments were not pooled with the rest of the results) and overlapping study data were also excluded. We used the adjusted odds ratio (OR) reported in case-control studies and the relative risk (RR) reported in cohort studies to calculate the overall effect. The analyses were conducted using the Generic Inverse Variance method, in which the weight of each study was equivalent to the inverse of the variance of the effect estimate [47]. A random-effects model was used if there was evidence of heterogeneity. Some exposures were reported at different levels; consequently, we calculated an estimate for binary exposure status (exposed vs. not exposed) by pooling exposed cases and controls across all levels of exposure. I2 statistics were used to assess heterogeneity, where 0% indicated perfect homogeneity and 100% complete heterogeneity [48]. Data were analyzed using Review Manager (RevMan) V.5.4.1, Cochrane, London, UK [49]. The results are summarized in Table 11.                 For each included factor, quantitative analysis was performed with at least two studies; results from studies restricted to infants, univariate analysis with no adjustments and overlapping data were excluded.

Selected Studies
The results of the search strategy and process of selecting studies for review are illustrated in Figure 1. The preliminary search returned 7813 studies, which were screened based on the title. Following the assessment of titles and abstracts, 317 articles met the criteria for full-text review. The subsequent screening of the full reports of the 317 selected publications resulted in the selection of 38 eligible studies. Table 3 provides an overview of the number of studies included in each extracted exposure. The detailed characteristics (authorship, year of publication, study design, period and place of the study, number and age of participants, control source and extracted exposure variables) of each included study are shown in Table 4.

Quality Assessment
The quality assessment [10] for the included studies is shown in Supplementary Table  S2. The most common weakness of many studies was the lack of adjustment for possible confounding factors such as maternal smoking status, maternal alcohol consumption and perinatal factors. The confounding factors used in the analysis for each included study are shown in Tables 5-10.
Two studies reported results for AL, one focused on cured meats but their results did not achieve statistical significance [16] and the other reported a statistically significant increased risk of leukemia in infants with the maternal daily consumption of fresh vegetables identified a priori as containing DNA topoisomerase II inhibitors [15].
Regarding ALL, three consecutive studies [11,14,17] conducted by the Northern California Childhood Leukemia Study (NCCLS) reported a statistically significant inverse association with a higher daily consumption of vegetables, fruits and proteins. Further analysis determined that these associations were statistically significant for carrots, cantaloupe, oranges, green beans, beans and beef. A case-control study [13] reported statistically significant inverse associations between the consumption of fruits, vegetables and fish during pregnancy and the risk of ALL in children. Conversely, a higher intake of sugar and syrup and meat and meat products had a statistically significantly increased risk of ALL. Two other studies [12,15] did not report statistically significant associations between the consumption of cured meats and ALL. A meta-analysis of two studies [13,17] was performed, in which a statistically significant inverse estimate for ALL was found for higher maternal consumption of fruits (OR = 0.71, 95% CI: 0.59-0.86), but not for vegetables (OR = 0.88, 95% CI: 0.69-1.11), and of two [12,14] studies for cured meats, which was not significant (OR = 1.01, 95% CI: 0.68-1.51) (Table 11).
A statistically significant increased risk for AML was reported in one case-control study related to the intake of certain fresh vegetables (e.g., vegetables such as onions, broccoli and asparagus that are rich in Quercetin, which induces human DNA topoisomerase II) [15], and another study reported an inverse association with fruit consumption [17].
A cohort study [27] (N = 141,216/AL cases = 96) reported no statistically significant association between coffee intake during pregnancy and the risk of AL. In the present meta-analysis, including five studies [22,[24][25][26][27]] that compared the highest vs. the lowest consumption reported for coffee intake during pregnancy, we did not observe a statistically significant association for AL (OR = 1.36, 95% CI: 0.97-1.91).
Another pooled study [18] of eight case-control studies reported an increased risk of AML for children whose mothers consumed more than one cup per day (OR = 1.40, 95% CI: 1.03-1.92), and the risk increased with increments of one cup per day (OR = 1.18, 95% CI: 1.01-1.39). A meta-analysis of four studies comparing the highest versus the lowest consumption for AML [20,[22][23][24] reported that the association was not statistically significant (OR = 1.18, 95% CI: 0.66-2.13) (
Overall, the case control studies [15,[20][21][22][23][24] did not report any statistically significant associations between maternal tea intake during pregnancy and the risk of childhood AL. Both pooled analyses looked for dose responses and interactions, but the findings were null [18,19]. A recent cohort study also added tea intake to their model and found no association (data not reported) [27]. A meta-analysis of two [22,24] studies found no statistically significant association estimate for AL with the highest vs. the lowest consumption of tea (OR = 0.97, 95% CI: 0.80-1.51) (Table 11).

C. Soft drinks
Three case-control studies [15,16,22] evaluated the association between maternal intake of soft drinks (mostly cola beverages) and AL risk (Table 8). Only one case-control study [22] reported an increased risk with the maternal intake of cola beverages and AL, Figure 2. Meta-analysis of the association between the consumption of coffee during pregnancy and (a) acute leukemia [22,[24][25][26][27], (b) acute lymphoblastic leukemia [20][21][22][23][24] and (c) acute myeloid leukemia [20][21][22]24]. Note: the individual estimate (OR) from the studies are represented by the red box, and the black diamond represents the estimate of the meta-analysis.
Overall, the case control studies [15,[20][21][22][23][24] did not report any statistically significant associations between maternal tea intake during pregnancy and the risk of childhood AL. Both pooled analyses looked for dose responses and interactions, but the findings were null [18,19]. A recent cohort study also added tea intake to their model and found no association (data not reported) [27]. A meta-analysis of two [22,24] studies found no statistically significant association estimate for AL with the highest vs. the lowest consumption of tea (OR = 0.97, 95% CI: 0.80-1.51) (Table 11).

C. Soft drinks
Three case-control studies [15,16,22] evaluated the association between maternal intake of soft drinks (mostly cola beverages) and AL risk (Table 8). Only one case-control study [22] reported an increased risk with the maternal intake of cola beverages and AL, especially for AML. A meta-analysis of the three studies found marginal association estimates for AL with the highest vs. the lowest consumption of cola beverages (OR = 1.23, 95% CI: 0.97-1.55) (Table 11).

Micronutrient Intake
Four case-control studies [11,14,28,29] evaluated the association between maternal micronutrient intake and the risk of childhood AL ( Table 9). Two of these studies from the NCCLS reported estimations with micronutrients from food only. One [11] found that the total glutathione level was inversely associated with ALL. Another study reported an inverse association between the daily vitamin B6 intake dose and the risk of AML [29].

Summary of the Main Findings
The main findings of the systematic review suggest that maternal dietary intake rich in vegetables, fruits and protein sources (beans and beef) and supplementation with folic acid and multivitamins during pregnancy could have a protective effect against childhood acute leukemia. The increased intake of cured meats, coffee and/or caffeinated beverages could increase the risk of childhood acute leukemia. Some of these findings were confirmed by our meta-analysis; a higher maternal consumption of fruits was significantly inversely associated with ALL (OR = 0.71, 95% CI: 0.59-0.86). This association was recently found in a previous meta-analysis [50] in which fruit (OR = 0.81, 95% CI: 0.67-0.99), vegetables (OR: 0.51, 95% CI: 0.28,0.94) and legumes (OR = 0.76, 95% CI: 0.62-0.94) were inversely associated with ALL.
Regarding maternal tea consumption during pregnancy, our analysis did not find a statistically significant association. These findings are consistent with those of a previous meta-analysis [52].
Regarding soft drink intake during pregnancy, the analysis suggested a possible increased risk for AL, but this finding was not significant. A previous meta-analysis [52] reported an increased risk between low and moderate consumption for AL (OR = 1.21, 95% CI: 1.03-1.44) and ALL (OR = 1.31, 95% CI: 1.09-1.59).
Folic acid supplementation during pregnancy suggested a modest inverse association with ALL (OR = 0.77; 95% CI: 0.59-1.01). A recent systematic review reported that the use of maternal folic acid supplementation before pregnancy was associated with a 30% decreased risk of ALL (OR = 0.69, 95% CI: 0.50-0.95) and the use of vitamin supplements before pregnancy was associated with a 20% decreased risk of ALL (OR = 0.81, 95% CI: 0.74-0.88) [50]. A more recent meta-analysis reported an inverse association between folic acid supplementation and ALL (OR = 0.75, 95% CI: 0.66-0.86) [54].

Potential Mechanisms
The inverse association between maternal fruit and vegetable intake and the risk of AL could be explained mainly by the high content of protective bioactive components such as micronutrients and fiber. Some fruits and vegetables that were found to be protective in some studies [11,14] are rich in carotenoids, flavonoids and vitamins A and C, which are important for DNA repair [55,56]. Also, the presence of the antioxidant tripeptide glutathione, present in meat and vegetables, plays a critical role in many cellular processes such as cell differentiation, proliferation and apoptosis [57].
Some studies suggest the association between the maternal intake of cured meats and the risk of childhood AL. This is because of the presence of N-nitroso precursors in these meats that can be converted to carcinogenic components [58] by acids in the stomach and then transported through the placenta to the developing child. This might increase the risk of malignancy as a result [59].
The decreased risk of ALL associated with the maternal use of vitamin formulations as supplements can be explained by the capacities of the components, including folate and iron, to protect against oxidative damage to lipids, lipoproteins and DNA [56]. On the other hand, folic acid and its derivatives, known as folates, are chemo-protective micronutrients with an essential role in the maintenance of health and genomic integrity [60].
The increased risk of AL associated with the maternal consumption of coffee and other caffeinated drinks could potentially be explained by the effects of their components (caffeine and chlorogenic acid) in inhibiting Topoisomerase I and Topoisomerase II [61,62], which are nuclear enzymes that solve the topological problems associated with DNA replication, transcription, recombination and chromatin remodeling [63]. Tea contains lower levels of caffeine, folate and polyphenols, which might explain the divergent results of coffee consumption.

Strengths and Limitations
The present systematic review and meta-analysis integrated the available published information regarding maternal dietary intake and the risk of childhood AL. Unfortunately, most of the evidence comes from case-control studies (which are prone to recall bias) that were conducted more than a decade ago and restricted to high-income countries. It is also important to note that the accurate evaluation of dietary intake is challenging. We also noticed that some studies restricted their search to specific food items [18] and, given the heterogeneity of how results were reported (above all the studies), this makes it difficult to compare results.
However, there are new generation studies (biomarker-based, which are not the focus of this review) that can support some of these findings. Studies have been conducted to assess the presence of carcinogens (such as acrylamide, nitrosamines and others) in the maternal diet, which can be found in the placenta, linked to fetal hemoglobin, and can also be found in newborn lymphocytes. Since these biomarkers were also found in healthy children, further studies focusing on genetic variants found some polymorphisms that could affect the expression of enzymes such as folate hydrolase 1. With lower detoxifying activity, DNA damage or the higher activity of carcinogens may occur [64].
Regardless of the limitations of these studies, professionals dedicated to treating childhood cancer have demonstrated a great capacity to collaborate and collect high-quality data and biospecimens. This could lead to improvements in the quantity and quality of future research, for more prospective cohort studies and a more comprehensive and accurate assessment of the role of nutrition in carcinogenesis are needed.

Conclusions
This systematic review of the available literature indicates that a maternal diet rich in vegetables, fruits and supplementation with folic acid and multivitamins during pregnancy could be inversely associated with the risk of childhood AL, while a higher intake of coffee and/or caffeinated beverages could increase the risk. Only the results regarding maternal diet rich in fruits and the consumption of coffee in relation with childhood AL were confirmed by the meta-analysis performed on the available published data. Although this review sheds light on the risk factors related to maternal diets and the prevention of cancers in children and adolescents, more studies are needed to confirm causality and the potential mechanisms involved.
Supplementary Materials: The following supporting information can be downloaded at: https://www. mdpi.com/article/10.3390/ijerph20075428/s1, Table S1: PRISMA 2020 checklist Maternal nutrition and Childhood Acute Leukemia, which refers to Page et al. [65]; Table S2: Quality assessment of the studies included in the systematic review evaluated by the checklist proposed by Fowkes et al. [10].