Environmental risk factors of Wilms tumour: A systematic review and meta-analysis

Abstract

Wilms tumour (WT) is the fourth leading cause of cancer death in children. Elucidating modifiable risk factors is crucial in identifying venues for primary prevention of the disease. This study aimed to review literature and synthesize environmental risk factors for WT. We conducted a systematic review and meta-analysis of epidemiological studies using PubMed, Web of Science, and Embase databases. Studies were included if they were case-control or cohort studies of children under the age of 20 years at diagnosis and reported Relative Risks (RRs) with 95 % confidence intervals (CIs). Pooled effect sizes (ES) and 95 % CIs for risk factors associated with WT were estimated using random-effects models. We included 58 eligible studies from Asia, Europe, Latin and North America, and Oceania totalling approximately10000 cases of WT diagnosed between 1953 and 2019. We confirmed an association between high birthweight ((>4000 g) ES 1.54, CI 1.20–1.97) and WT. Similarly, consistent associations were suggested for Caesarean section (ES 1.23, CI 1.07–1.42), gestational age <37 weeks (ES 1.45, CI 1.21–1.74), and large-for-gestational age (ES 1.52, CI 1.09–2.12). Parental occupational exposure to pesticides during preconception / pregnancy also showed increased risks of WT (maternal ES 1.28, CI 1.02–1.60, paternal ES 1.48, CI 0.98–2.24). There were inverse associations for breastfeeding (ever breastfed = ES 0.71, CI 0.56–0.89; < 6 months ES 0.67, CI 0.49–0.91; and ≥6 months ES 0.75, CI 0.59–0.97), and maternal intake of vitamins (unspecified) and folic acid during pregnancy (ES 0.78, CI 0.69–0.89). Among factors showing no associations were low birthweight (<2500 g), small-for-gestational age, assisted reproductive technology, parental age, and smoking or alcohol consumption during preconception / pregnancy, paternal occupational extremely low frequency magnetic fields (ELF-MF) exposures, and maternal X-ray exposure during pregnancy. Our findings suggest that modifiable risk factors of WT are parental occupational exposure to pesticides, breastfeeding (beneficial), and intake of folic acid during preconception / pregnancy (beneficial), but all associations were rather modest in strength.

Highlights

• •Parental occupational exposure to pesticides showed increased risks for Wilms tumour.
• •Breastfeeding and intake of folic acid is protective against Wilms tumour.
• •Association between high birthweight and Wilms tumour was confirmed.
• •Associations suggested for C-section, gestational age <37 weeks and LGA.

1. Introduction

Wilms tumours (WT), also known as nephroblastoma, are solid malignant tumours that originate from the primitive embryonic tissue of the kidney [1]. Approximately 90 % of renal cancers in children are WT with its peak incidence observed between the ages of 1 and 4 years, other types of renal childhood cancers account for the remaining 10 % of cases [1], [2], [3]. On the other hand, 98–99 % of WT occur under the age of 16 years with the majority under the age of 8 years. It is the fourth leading cause of cancer death in children [4], [5]. They develop slightly more often in girls than boys [6]. Incidence rates of WT are higher in low- and middle-income countries (LMICs) [7], at the same time having lower survival rates than observed in high income countries (HICs) [8], [9]. This disparity is attributed to the delay in diagnosing WT, limited access to treatment, and inadequate follow-up [9]. Despite being rare with an estimated global total of annually 14,600 new patients, WT still contributes to the overall childhood cancer burden [7].

The aetiology of WT is largely unknown. There are known susceptibility genes, paediatric overgrowth disorders such as Beckwith-Wiedemann syndrome which causes an association between high birth weight and WT, and other congenital malformation syndromes, which have been suggested to account for about 10 % of WT cases [3], [8], [10]. Several large, population-based studies have reported on WT in children with birth defects [11], [12], [13], [14], but these studies are not within the scope of the present systematic review and meta-analysis.

Rare solid tumours such as WT have received less research attention compared to the more common childhood cancers leukaemia and brain tumours [15]. Hence, only few risk factors, such as genetic predispositions and family history of the diseases are established, while several others are hypothesized. Original studies are mostly inconsistent and inconclusive for most potential risk factors and generally show risk estimates with low precision due to small numbers [16], [17], [18], [19], [20].

So far, only few systematic reviews and meta-analyses have evaluated risk factors for WT [15], [21]. Chu et al. [15] assessed hypertension, history of miscarriage, birth and parental characteristics, pesticides and hydrocarbon exposures in a review published in 2010. However, they combined all exposure windows in their meta-analyses. Doganis et al. [21] in 2020 explored the associations between WT and maternal exposures during pregnancy, including intake of vitamins, cigarette smoking, and alcohol consumption. Our present systematic review and meta-analysis updates previous reviews and includes the following potential risk factors: birth and parental characteristics (gestational age and weight, breastfeeding, birth order, mode of delivery, and parental age), parental smoking, maternal consumption of alcohol, coffee, and intake of vitamins and folic acid during preconception / pregnancy, parental exposure to pesticides, paternal occupational exposure to extremely low frequency magnetic fields (ELF-MF), wood and paper dust, as well as maternal exposure to X-ray/CT scans and diesel engine exhaust fumes. Articles were published from 1987 to 2022 with cases diagnosed between 1953 and 2019.

2. Methods

The present study was carried out in compliance with the 2020 Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) checklist [22] (Supplemental material). The protocol used for data collection and analysis have been reported elsewhere [23].

2.1. Search strategy and study selection

A thorough search was conducted across the PubMed, Web of Science, and Embase databases, with no limitations on publication date or language. Retrieved peer-reviewed articles were imported and screened for duplicates in EndNote version X9.3.3. Authors (FMO and RD) independently assessed the titles, abstracts, and full text of the articles for eligibility (Supplemental material). In accordance with the Cochrane handbook for systematic reviews [24], discrepancies during the independent selection process were resolved through consensus. This was supplemented by exploring of lists of references. The search strategy included a list of key words and MeSH terms with filters (Supplemental material). The search was initially conducted in June 2022 and continued through November 2023. Extracted information was from original articles published in peer-reviewed journals spanning the period from 1987 to 2022. The studies were included if they were case-control or cohort studies of children under the age of 20 years, reported exposure time windows, and provided estimates of Relative Risks (RRs) such as Odds Ratio (OR), Hazard Ratio (HR), Standardized Mortality Ratio (SMR), Mortality Rate Ratio (MRR), Standard Incidence Ratio (SIR) or Incidence Rate Ratio (IRR) with 95 % confidence intervals (CIs). We checked publications from the same region for potential overlaps of their study populations. The inclusion and exclusion criteria were defined a priori (Supplemental material).

2.2. Data extraction

We extracted authors’ name, year of publication, study location, period and age range of diagnosis, exposures, exposure assessment methods, outcome ascertainment, number of WT cases or, if not available, the number of childhood cancer cases and controls or study population, follow-up duration, as well as risk estimates with their respective CIs. Information on study design (case-control and cohort or nested case-control) was also extracted. Nested case-control studies were combined with cohort studies, irrespective of whether they were nested in a cohort study or as registry-based study in the source population [23]. Case-control studies are therefore studies requiring active participation and exposure assessment. Risk factors extracted included birth and parental characteristics, pesticides and other chemicals, radiations, and lifestyle exposures. Exposure time periods such as preconceptional, prenatal and postanal were also considered.

2.3. Quality assessment of eligible articles

Included articles underwent an evaluation of their methodological quality using the critical appraisal tools by the Joanna Briggs Institute (JBI) for case-control and cohort studies [25]. The critical appraisal checklist has 10 criteria for case-control and 11 for cohort studies. Each question with “yes” score 1, “no” score 0 and “unclear” or “not applicable” score 0 (Supplemental material).

2.4. Statistical analyses

We conducted random-effects meta-analyses to estimate and report pooled effect sizes (ES) along with their corresponding 95 % CIs. Funnel plots and Egger's test were used to evaluate potential publication bias [26]. The I² statistic was calculated to quantify the heterogeneity of the results between studies. I² values of 0 % were considered to represent “no heterogeneity”, from 1 % to 35 % “low heterogeneity”, from 36 % to 55 % as “moderate”, from 56 % to 70 % as “substantial” and above 71 % as “considerable” heterogeneity [27]. Analyses were conducted both combining case-control and cohort studies, and separately by study design to identify possible systematic differences in the results by study design (case-control vs cohort studies), and further sub-analyses were conducted by decade and continent. In addition, we conducted 3 categories of leave-one-out cross-validation meta-analysis for all risk factors with 5 or more studies, leaving out the highest or lowest risk estimate as well as the one from the oldest study. A nominal significance level of 0.05 was used for analyses. Analyses were conducted using STATA® software, version 15.1 (College Station, TX, USA).

3. Results

3.1. Study bias and quality assessment

Overall, the 58 articles critically appraised for quality using the JBI tools were of good quality. Case-control studies had a slightly higher average score (89 %) compared to cohort studies (80 %). Thus, all screened articles appraised were included in the final analysis (Supplementary Table 4 and 5). Prior to the critical appraisal of the 58 articles, we systematically checked the articles for overlaps in their study populations and by risk factors. We excluded 6 articles with substantial overlap, most of the excluded articles were from the Nordic countries (5 out of 6).

3.2. Study characteristics

A total of 3763 nonduplicate records were retrieved and screened, resulting in the assessment of 71 full texts. Among these, 58 studies (23 case-control and 35 cohort studies (including nested case-control studies)) involving approximately 10000 children diagnosed with WT between 1953 and 2019 met the study inclusion criteria [2], [4], [5], [10], [12], [16], [17], [18], [19], [20], [28], [29], [30], [31], [32], [33], [34], [35], [36], [37], [38], [39], [40], [41], [42], [43], [44], [45], [46], [47], [48], [49], [50], [51], [52], [53], [54], [55], [56], [57], [58], [59], [60], [61], [62], [63], [64], [65], [66], [67], [68], [69], [70], [71], [72], [73], [74], [75] Table 1 lists the characteristics of all 58 studies included in the present systematic review and meta-analyses, the effect sizes for each exposure are reported in the forest plots and table shown in Fig. 2, Fig. 3 and Table 2 (Supplemental material).

Table 1.

Characteristics of the 58 studies included in the systematic review and meta-analysis, sorted by country.

First author	Country	Date of diagnoses	Age (Years)	Risk factor	Exposure assessment	Outcome ascertainment	Cases	Control
Latin America Case-control studies
Sharpe et al., 1995 [28]	Brazil	1987–1989	<19	Parental exposures to pesticides	Face-to-face interview	Hospital records	109	218
Sharpe et al., 1999 [29]	Brazil	1987–1989	< 19	Parental age	Face-to-face interview	Hospital records	109	218
Sharpe et al., 1996 [74]	Brazil	1987–1989	< 19	Maternal vitamin supplements	Face-to-face interview	Hospital records	109	218
Rios et al., 2018 [36]	Peru	2012–2015	<18	Parental age	Self-administered questionnaire and hospital records	Hospital records	35	70
Europe
Rios et al., 2020 [31]	France	2010–2011	<11	Multiple risk factors	Telephone interview	Cancer registry	117	1100
Bauer et al., 2020 [12]	France	2010–2011	<11	Maternal and perinatal characteristics	Telephone interview	Cancer registry	117	1100
Hug et al., 201032	Germany	1992–1994	<15	Parental occupational exposure to ELF-MF	Self-administered questionnaire and telephone interview	Cancer registry	178	2382
Schüz et al., 2001 [33]	Germany	1988–2006	<10	Multiple risk factors	Self-administered questionnaire and telephone interview	Cancer registry	177	2006
Schüz and Forman, 2007 [34]	Germany	1992–1994	<14	Birth characteristics	Self-administered questionnaire and telephone interview	Cancer registry	147	2057
Schüz et al., 2007 [35]	Germany	1992–1994	<14	Medication use during pregnancy	Self-administered questionnaire and telephone interview	Cancer registry	147	2057
Smulevich et al., 1999 [37]	Russia - Moscow	1986–1988	<15	Multiple risk factors	Face-to-face interview	Cancer registry	48	1181
Sorahan et al., 1997 [38]	UK	1953–1955	<16	Parental smoking	Face-to-face interview	Birth registry	133	133
Pang et al., 2003 [39]	UK	1991–1994	<15	Parental smoking	Face-to-face interview	Cancer registry	182	7581
Smith et al., 2009 [40]	UK	1991–1996	<15	Birth weight	Face-to-face interview	Cancer registry	179	6337
Rajaraman et al., 2011 [45]	UK	1992–1996	<15	diagnostic radiation and ultrasound scans	Self-administered questionnaire	UKCCS, histopathology review, or consultant treating the child	135	258
North America
Olshan et al. 1993 [30]	USA and Canada	1984–1986	<15	Multiple risk factors	Self-administered questionnaire	National pathology centre	200	233
Saddlemire et al., 2006 [41]	USA and Canada	1999–2002	<15	Breastfeeding	Telephone interview	Children's cancer group	501	480
Cooney et al., 2007 [4]	USA and Canada	1999–2002	<16	Household pesticides	Telephone interview	Children's cancer group	523	517
Goel et al., 2009 [42]	USA and Canada	1999–2002	<16	Maternal exposure to medical radiation	Telephone interview	Children's cancer group	512	509
Bunin et al., 1987 [43]	USA	1970–1983	<15	Maternal Xray exposure	Telephone interview	Hospital records	88	88
Tsai et al., 2006 [44]	USA	1992–1995	<10	Multiple risk factors	Telephone interview	Cancer registry	303	575
Daniels et al. 2008 [5]	USA and Canada	1999–2002	<16	Birth characteristics	Self-administered questionnaire and telephone interview	Children's cancer group	521	517
Carozza et al., 2009 [46]	USA-Texas	1990–1998	<15	Pesticides	Digital orthophoto quadrangle (DOQ) data	Cancer registry	19	1802
Cohort studies (including nested / registry-based case-control studies)
First author	Country	Data collection year	Age (years)	Risk factor	Exposure assessment	Outcome ascertainment	Cases	Study population	Follow up (Years)
Asia
Cha et al., 2011 [57]	Korea	1995–2006	<12	Birth characteristics	Birth database	Death database	52	6479406	11
Australia /Oceania
Stavrou et al., 2009 [47]	Australia	1994–2005	<13	Maternal smoking	Midwives data collection	Cancer registry	74	798,987	12
Europe
Volk et al., 2020 [51]	Denmark	1968– 2016	<19	Parental occupational exposure	Employment history from civil registration system	Cancer Registry	64	801,867	18
Jepsen et al., 2004 [52]	Denmark	1973–1993	<15	Fetal growth	Birth register	Cancer registry	126	1260	14
Hall et al., 2020 [53]	Denmark	1968–2016	<20	Occupational livestock or animal dust	Birth registry and parental occupational history	Cancer registry	13	123,228	16
Heck et al., 2019 [54]	Denmark	1973–2004	<20	Multiple risk factors	Central population register	Cancer registry	217	4340	19
Sundh et al., 2014 [16]	Denmark	1968–2004	<15	assisted reproductive technology (ART)	Hospital records	Cancer registry	6	45	9.5 mean
Schüz et al., 2015 [56]	Denmark	1973–2010	<20	Birth order	Central Population Register	Cancer registry	130	2461,283	19
Momen et al., 2014 [55]	Denmark, Sweden, and Finland	1973–2007	<15	Delivery by caesarean section	Birth registers	Cancer registers	717	882 907	15
Seppälä et al. 2021 [20]	Finland	1996–2014	<20	Preterm birth	Medical records	Cancer registry	9	110	19
Meulepas et al., 2019 [19]	Netherlands	1979–2012	<19	Pediatric CT Scans	Hospital records	Cancer registry	6	168394	8
Schüz et al., 2011 [58]	Nordic countries	1985–2006	<15	Birth characteristics	Birth registry	Cancer registry	690	3450	14
Bjørge et al., 2013 [73]	Nordic Countries	1967–2010	<15	Foetal growth	Face-to-face interview	Cancer registries and direct reports from paediatric, oncology hospitals	961	172 422	14
Heuch et al., 1996 [59]	Norway	1967–1992	<15	Birth characteristics	Birth registry	Cancer registry	119	1 489 297	15
Mortensen et al., 2016 [60]	Norway	1999–2010	< 11	Maternal folic acid supplements	Birth registry	Cancer registry	105	687 406	10
Rodvall et al., 2003 [17]	Sweden	1965–1976	0–19	Pesticides	Death records	Cancer registry	<5	27, 329	18
Feychting et al., 2000 [61]	Sweden	1976, 1977, 1981, and 1982	<15	ELF-MF	Occupational history	Cancer registry	15	235635	14
Hardell and Dreifaldt, 2001 [62]	Sweden	1988–1991	<15	Breastfeeding	Medical records	Cancer registry	44	49	14
Rossides et al., 2022 [63]	Sweden	1960–2015	<20	Parental occupational exposure to hydrocarbon solvents and engine exhaust fumes	Employment history and job exposure matrix	Cancer registry	494	8927	19
Crump, 2014 [10]	Sweden	1973–2009	<10	Perinatal risk factors	Birth registry and national census data,	Cancer registry	443	3595055
Coste et al., 2020 [64]	Switzerland	1990–2015	<16	Parental occupational exposure to pesticides	Census records / CLIC-JEM	Cancer registry	51	2129958	15
Williams et al., 2018 [18]	UK	1992–2008	<15	Assisted reproductive technology (ART)	Human fertilization and embryology authority records	Cancer registry	< 5	12 137	8
Fear et al., 1998 [65]	UK	1959–1990	<15	Occupational exposure to pesticide	Death certificates	National statistics	42	167 703	14
Fear et al., 2009 [2]	UK	1962–1999	<15	Occupational exposure to dust	Birth /employment records and job exposure matrix	National registry of childhood tumours	2568	2568	14
Latin America
Rangel et al., 2010 [48]	Brazil	1984–2008	<18	Birth weight	Birth certificates	Cancer registry	115	1575	17
North America
Grupp et al., 2011 [49]	Canada - Ontario	1995–2006	<10	maternal folic acid intake	Hospital records	Cancer registry	270	9288014	9
Marcoux et al., 2022 [50]	Canada - Quebec	2006–2019	<14	Gestational diabetes	Hospital records	Cancer registry	12	1030537	14
Spector et al., 2019 [68]	USA	2004–2012	<10	In vitro fertilization	Hospital records	Cancer registry	28	275 686	8
Puumala et al., 2008 [66]	USA - Minnesota	1976–2004	<15	Birth characteristics	Birth registry	Cancer surveillance system	2188	8752	15
Williams et al., 2021a [69]	USA - Minnesota	1976–2014	<15	Caesarean section	Birth certificate	Cancer registry	a3166	20589	14
Williams et al., 2021b [67]	USA	1976–2014	<15	Birth characteristics	Birth certificate	Cancer registry	a16411	69816	14
Contreras et al., 2016 [70]	USA -California	1988–2013	<6	Gestational diabetes	Birth records	Cancer registry	1052	270147	5
Heck et al., 2016 [71]	USA -California	2007–2011	<6	Parental smoking	Birth certificates	Cancer registry	148	40356	5
Kumar et al., 2018 [72]	USA-Texas	2003–2009	<5	Maternal residential proximity to major roadways	Texas roadway network StratMap	Cancer registry	143	2855	4
Linabery et al., 2012 [75]	USA	1986–2008	<5	Maternal folic acid intake	National Cancer Institute’s Surveillance, Epidemiology, and End Results (SEER)	SEER	571	7089	4

^aTotal cases in some publications were summed-up with other types of childhood cancer but the actual numbers for WT were not made available despite separating the risk estimates for WT.

Of all eligible studies, 57 % (n = 33) were conducted in Europe. North America followed with 31 % (n = 18), while Latin America had 8 % (n = 5), Asia and Oceania had 2 % (n = 1) each. Notably, there were no studies conducted in Africa.

Fig. 1.

Preferred Reporting Items for Systematic Reviews and Meta-Analyses Flow Diagram Outlining the Study Selection.

3.3. Birth and parental characteristics

Association was suggested for Caesarean section (C-section) and the risk of WT (ES 1.23, CI 1.07–1.42), based on 2 cohort studies (ES 1.26, CI 1.05–1.50) and a slightly more statistical uncertainty but comparable magnitude of association, 3 case-control studies (ES 1.19, CI 0.95–1.50). There was no heterogeneity across studies, with Egger’s P-value of 0.61. Similarly, we observed an increased risk of WT for gestational age <37 weeks (ES 1.45, CI 1.21–1.74), based on 6 cohort studies (ES 1.43, CI 1.18–1.74) and a comparable strength of association, 2 case-control studies (ES 1.60, 0.96–2.66). No heterogeneity was observed across studies, Egger’s P-value was 0.65. LGA showed an increased risk of WT (ES 1.52, CI 1.09–2.12), based on 2 cohort studies (ES 1.32, CI 0.95–1.84) and 1 case-control study (RR 2.10, CI 1.20–3.60). There was low heterogeneity across studies, and Egger’s P-value 0.08. However, no association was found for children born at gestational age > 40 weeks (ES 0.93, CI 0.81–1.08) or for SGA (ES 0.83, CI 0.49–1.42) after combining the respective study designs (Fig. 2 and Supplement Fig. 1, pg 15–19).

Fig. 2.

Meta-analysis of pooled effect sizes (ES) of exposure to birth characteristics (Gestation age < 37 weeks, > 40 weeks, SGA, LGA, ART and Caesarean section) for the risk of WT and heterogeneity, by study design. † Risk estimate not ES, as only one study was identified.

The well-known association between high birthweight (>4000 g) and WT was confirmed (ES 1.54, 1.20–1.97), based on case-control (ES 1.25, CI 0.84–1.87, 3 studies) and cohort studies (ES 1.69, CI 1.24–2.30, 4 studies). There was a moderate heterogeneity across studies, with Egger’s P-value as 0.69. Birthweight < 2500 g, ART, birth order 2 and ≥ 3 did not show association with WT (Fig. 3 and Supplement Fig. 1, pg 20–21).

Fig. 3.

Meta-analysis of pooled effect sizes (ES) of exposure to birth characteristics (Birthweight, Breastfeeding and Birth order 2 and ≥ 3) for the risk of WT and heterogeneity, by study design. † Risk estimate not ES, as only one study was identified.

Ever breastfeeding the child was protective (ES 0.71, CI, 0.56–0.89) against WT, based on 3 case-control studies (ES 0.73, CI 0.57–0.93), and with wider CI in 1 cohort study (ES 0.45, CI 0.17–1.14). In similar manner, the protective effect for breastfeeding was seen in both shorter and longer duration, with no apparent association with duration of breastfeeding (< 6 months ES 0.67, CI 0.49–0.91, 3 studies; ≥ 6 months ES 0.75, CI 0.59–0.97, 4 studies), consistent across study designs. Children whose mothers’ age at birth was <25 years old was not associated with WT (ES 1.02, CI 0.77–1.34, 7 studies), as well as mothers’ age >35 years old (ES 0.80, CI 0.65–1.00, 5 studies), this was also consistent across study design. Fathers’ age at childbirth was ES 1.12, CI 0.94–1.34 for younger fathers (< 25 years old), 3 studies, and ES 1.01, CI 0.87–1.16, 6 studies for older fathers (>35 years old) (Fig. 3 and Supplement Fig. 1, pg 22–28).

3.4. Lifestyle

Maternal smoking during pregnancy was not associated with WT (ES 0.90, CI 0.78–1.04) and the results were consistent across study designs (ES 0.88, 0.75–1.03, 6 case-control studies; ES 0.99, CI 0.69–1.43, 3 cohort studies). No heterogeneity was observed across studies, Eggers P-values was 0.08. Paternal smoking during preconception / pregnancy of the index child did not show an association with WT (ES 1.01, 0.87–1.18) which was reported in 5 case-control studies. Maternal consumption of alcohol during preconception / pregnancy did not show any association with WT (ES 1.18, 0.72–1.92, 3 studies) but there was modestly elevated risk estimate with wide confidence interval for maternal consumption of 1–3 cups of decaffeinated coffee per day during pregnancy compared with no consumption (ES 1.34, CI 0.79–2.28, 2 studies), both risk factors where only reported in case-control studies. On the other hand, maternal intake of vitamins and folic acid during preconception / pregnancy showed a protective effect against WT (ES 0.78, 0.69–0.89), based on 2 case-control (ES 0.81, CI 0.47–1.39) and 3 cohort studies (ES 0.79, CI 0.69–0.90). Maternal gestational diabetes was ES 1.15, CI 0.83–1.59, based on 2 case-control (ES 1.07, CI 0.70–1.63) and 2 cohort studies (1.27, CI 0.77–2.11) (Table 2 and supplement Fig. 2, pg 29–31).

Table 2.

Meta-analysis of pooled effect sizes (ES) of exposure to parental characteristics, lifestyle, pesticides, and other chemicals for the risk of WT and heterogeneity (I²) between studies, by study design.

		Case-control						Cohort						Combined
Risk factor	Window period	N	†ES	LCI	UCI	I2(%)	Egger’s p value	N	†ES	LCI	UCI	I2(%)	Egger’s p value	N	†ES	LCI	UCI	I2(%)	Egger’s p value
Parental characteristics
Maternal age <25		5	1.29	0.80	2.07	60.4	0.30	2	0.84	0.70	1.00	20.6	-	7	1.02	0.77	1.34	66.4	0.20
Maternal age >35		3	0.76	0.29	1.18	0.0	0.55	2	0.82	0.64	1.05	0.0	-	5	0.80	0.65	1.00	0.0	0.88
Paternal age <25		1	3.62	0.94	13.90	-	-	2	1.10	0.91	1.32	0.0	0.97	3	1.12	0.94	1.34	0.0	0.51
Paternal age >35		4	1.31	0.82	2.08	0.0	0.35	2	0.98	0.84	1.14	0.0	-	6	1.01	0.87	1.16	0.0	0.55
Gestational diabetes		2	1.07	0.70	1.63	0.0	-	2	1.27	0.77	2.11	0.0	-	4	1.15	0.83	1.59	0.0	0.06
Lifestyle
Maternal smoking	Prenatal	6	0.88	0.75	1.03	0.0	0.35	3	0.99	0.69	1.43	0.0	0.18	9	0.90	0.78	1.04	0.0	0.08
Paternal smoking	Preconception/prenatal	5	1.01	0.87	1.18	0.0	0.27
Maternal coffee 1–3 cups/day	Prenatal	2	1.34	0.79	2.28	41.9	-
Maternal alcohol	Preconception/prenatal	3	1.18	0.72	1.92	80.7	0.57
Maternal vitamins and folic acid intake	Prenatal	2	0.81	0.47	1.39	-	-	3	0.79	0.69	0.90	0.0	-	5	0.78	0.69	0.89	0.0	0.63
Pesticides
Child's exposure to general pesticides	Postnatal	2	0.81	0.54	1.20	2	-
Maternal occupational exposure to general pesticides	Preconception/prenatal	4	1.26	1.00	1.58	0.0	0.75	1	2.19	0.68	7.05	-	-	5	1.28	1.02	1.60	0.0	0.87
Maternal use of any herbicide	Preconception/prenatal	2	1.03	0.74	1.43	0.0	-
Maternal use of any Fungicide	Preconception/prenatal	2	1.32	0.65	2.66	48.3	-
Maternal use of any Insecticide	Preconception/prenatal	2	1.49	1.17	1.90	0.0	-
Paternal occupational exposure to general pesticides	Preconception/prenatal	2	1.73	0.53	5.65	67.3	-	3	1.48	1.02	2.13	0.98	0.74	5	1.48	0.98	2.24	23.5	0.85
Radiation
Maternal exposure to X-rays	Preconception/prenatal	4	0.96	0.76	1.21	0.0	0.04
Paternal exposure to ELF-MF >0.1–0.2μT	Preconception/prenatal	2	1.29	0.93	1.80	0.0	-
Paternal exposure to ELF-MF >0.2μT	Preconception/prenatal	2	0.92	0.65	1.30	0.0	-
Dust
Paternal Occupational exposure to wood dust	Preconception/prenatal	2	1.03	0.83	1.21	0.0	-
Paternal Occupational exposure to paper dust	Preconception/prenatal	2	1.88	0.83	4.22	0.0	-

† Where only one study was identified, it is referred to as RR and not ES.

3.5. Chemicals and radiations

We observed an association between maternal occupational exposure to pesticides and the risk of WT (ES 1.28, CI 1.02–1.60) based on 4 case-control studies (ES 1.26, 1.00–1.58) and one cohort study with wide CI (RR 2.19, CI 0.68–7.05). The association was also observed in the subtype of pesticides, insecticide (ES 1.49, CI 1.17–1.90) when 2 case-control studies were combined. There was no heterogeneity across studies, Eggers P-values was 0.87 for maternal exposure to pesticides. For paternal occupational exposure to pesticides, we found an indication of borderline association (ES 1.48, CI 0.98–2–24), but consistent between 3 cohort studies (ES 1.48, CI 1.02–2.13) and 2 case-control studies with a higher magnitude of association but wider CI (ES 1.73, CI 0.53–5.65). Conversely, no association was observed for childhood exposure to pesticides in 2 case-control studies (ES 0.81, CI 0.54–1.20) (Table 2 and Supplement Fig. 3, pg 31–33 1).

Children whose mothers were exposed to X-ray during preconception / pregnancy were not associated with WT based on 4 case-control studies (ES 0.96, 0.76–1.21). No heterogeneity across studies and P-value was 0.04 for Eggers test. Likewise, there were no associations observed between maternal ultrasound, residential proximity to major roads, occupational exposure to diesel engine exhaust fumes, paternal occupational exposure to ELF-MF (> 0.1–0.2 μT and > 0.2 μT), wood and paper dust, and the risk of WT (Table 2 and Supplement Fig. 3, Pg 34–35).

3.6. Subgroup analyses

3.6.1. Association between risk factors and WT tumour by time and continent

C-section showed an increase in strength of association over the decades (ES 1.16 for 2000–2009, 1.25 for 2010–2019, and 1.29 for 2020 to date) with no significant heterogeneity across studies in all decades combined (P value = 0.60). Similarly, this was also observed for gestational age <37 weeks (ES 0.66 for < year 2000, 1.30 for 2000–2009, 1.36 for 2010–2019, and 1.62 for 2020 to date), with no heterogeneity across studies (P value = 0.64).

We found an association for C-section in Europe and North America where the reported studies were all published. Parental smoking and maternal exposure to pesticides were exclusively reported in Europe and North America as well, with maternal exposure to pesticides showing high magnitude of association in both continents. There was no heterogeneity across studies (P value = 0.60) (Supplement Table 6). Association was also observed for gestational age <37 weeks with a higher magnitude of strength in North America (1.63) compared to Europe (1.27). The P value for heterogeneity across studies was 0.64.

3.7. Leave-one-out cross-validation meta-analysis

Leaving out the highest or lowest risk estimate as well as from the oldest study in our analyses did not reveal different results than those from the main analysis for all risks factors with 5 or more studies (Supplementary Table 7).

4. Discussion

In the present systematic review and meta-analysis comprising 58 articles with approximately 10000 children diagnosed with WT between 1953 and 2019. We confirmed an association between high birthweight (>4000 g) and WT. Associations were suggested for C-section, gestational age < 37 weeks, and large-for-gestational age. Paternal and maternal exposure to pesticides during preconception / pregnancy also showed increased risks of WT. On the other hand, breastfeeding, and maternal intake of vitamins and folic acid were inversely associated with WT. Low birthweight (<2500 g), SGA, ART, parental age, and maternal smoking during pregnancy were not associated with WT.

The association we confirmed between high birthweight (>4000 g) and WT in our systematic review and meta-analysis can also be initiated by overgrowth disorders such as Beckwith-Wiedemann syndrome. Our results is consistent with the findings of Chu et al. [15] in a systematic review and meta-analysis on WT. High birthweight (>4000 g) could be an early marker of childhood cancer, as the association has been previously reported in systematic reviews of other types of childhood cancers, such as brain [23] and leukemia [76]. In our study we noted consistent homogeneity in strength of association in high birthweight (>4000 g), as all studies had risk estimates greater than 1.25 except for Smith et al. [40], with an overall good quality in their methodology. In our sub-analysis, the association persisted across continents, indicating a concern for high birthweight globally. The cut-off of > 4000 g for defining high birthweight has been consistently used across countries and was therefore the only definition we could use in our literature-based review, but it may not be an appropriate definition across countries and continents. The prevalence of high birthweight (>4000 g) varies greatly by continents and influenced by various factors including genetic predisposition to a larger extent, and maternal health, nutrition, and high socioeconomic status in various degrees of influence [66], [77]. About 30 % of WT have been attributed to underlying genetic and epigenetic mechanisms [78]. It has been reported in association with several abnormal constitutional karyotypes. Wilms tumour 1 (WT1) associated syndromes, familial Wilms tumour, and certain overgrowth conditions such as Beckwith-Wiedemann syndrome [79]. Understanding the various causes of high birthweight is crucial for developing targeted interventions and policies for WT and childhood cancer in general.

The choice for C-section over vaginal delivery is rising in developed countries, partly due to high socioeconomic status, attitudes towards childbirth and pain management as well as improved surgical procedures and advancement in medical science [80]. Our findings showed an increase in magnitude of association between C-section and WT across decades. C-section is well captured in hospital records and thereby cohort and registry-based case-control studies. It is unlikely to be affected by recall bias in traditional case-control studies. During C-section delivery, the newborn does not come in contact with mother's vaginal or fecal microbiome. This has been proven to have an impact on the composition of gut bacteria. The most notable change is the reduced presence of bifidobacteria, the near absence of bacteroides, and an increase in pathogenic bacteria [81]. Low birthweight (<2500 g) and high birthweight (>4000 g) are highly correlated to unplanned and emergency C-section compared to average birthweight (>2500 - <4000 g) [82], elective C-section delivery borne out of non-medical reasons should be practiced with caution. To our knowledge, this is the first systematic review and meta-analysis to evaluate and report association between C-section and WT, where previous reviews have reported associations on childhood leukemia [76], [83] and brain tumours [23] only.

Albeit of the protective association with ever breastfeeding, there was no dose-response relationship with duration of breastfeeding as children breastfed for less than six months and those greater than six months showed similar strength of inverse association. However, breastfeeding children has been reported to have a lower risk of developing several other types of childhood cancers [84], [85]. Breast milk is not only an essential nutritional source but also capable of protecting infants against various immunological diseases and disorders [84], [85]. Similarly, maternal intake of vitamins and folic acid which was also protective in our study, maybe attributed to the essential roles of vitamins in promoting healthy immune system, support normal cell growth and development [86]. However, our findings should be interpreted with caution as there were no specific information regarding the type of vitamins in the primary studies except for folic acid.

Maternal consumption of alcohol during preconception / pregnancy was not associated with WT. The primary studies that produced the effect sizes were all case-control studies with “considerable” heterogeneity of 81 % [31], [33], [44] which are prone to recall bias. On the other hand, the assessment of exposure to alcohol consumption is difficult in cohort studies and often bias at best can be assessed by a doctor at the registration of the pregnancy (often only as yes/no).

Our findings of association in children whose parents were occupationally exposed to pesticides is consistent with previous systematic review and meta-analysis [15]. The studies that produced our pooled estimates were from North America and Europe showing greater magnitude of association from the latter with Coste et al. [64] and Schüz et al. [33] from Switzerland and Germany reporting estimates of 2 fold associations for maternal exposures, respectively.

5. Limitations

Our review has several limitations, including potential information and selection biases inherent in the observational studies we extracted information from. Crude exposure assessment methods (ever vs. never) and exposure misclassification, most likely non-differential, may also have influenced the results. The number of cases for risk factors were not reported in some articles, as such we could not account for the number of cases by risk factors but only by studies. Our findings are based mostly on analyses of few studies, with majority of the studies published in Europe and North America. Finally, the sub- analysis was not analysed by study design due to limited number of eligible articles, hence we combined case-control and cohort studies. We could not address the question of the critical time window of exposure, especially looking at preconceptional exposures and exposures during pregnancy, because of small numbers of studies and many exposures being highly correlated across time periods.

6. Conclusions

Our systematic review and meta-analysis suggest that modifiable risk factors of WT are parental occupational exposure to pesticides, breastfeeding (beneficial), and intake of vitamins (unspecified) and folic acid during preconception / pregnancy (beneficial), but all associations were rather modest. Further studies are necessary with better and more precise exposure assessment methods to disentangle the various chemical constituents in risk factors like pesticides and specific type of vitamins.

Funding

The study was funded by a grant from the French National Cancer Institute (INCa:15670; PEDIAC consortium).

Disclaimer

Where authors are identified as personnel of the International Agency for Research on Cancer/World Health Organization, the authors alone are responsible for the views expressed in this article and they do not necessarily represent the decisions, policy or views of the International Agency for Research on Cancer/World Health Organization.

Role of the Funder

The funder had no role in the design of the study; extraction, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

CRediT authorship contribution statement

Felix M. Onyije: Methodology, Data curation, Software, Formal analysis, Visualization, Writing – original draft, Writing – review & editing. Roya Dolatkhah: Methodology, Data curation, Software, Visualization, Writing – original draft, Writing – review & editing. Ann Olsson: Conceptualization, Methodology, Visualization, Supervision, Writing – review & editing. Liacine Bouaoun: Methodology, Software, Formal analysis, Visualization, Writing – review & editing. Joachim Schüz: Conceptualization, Funding acquisition, Methodology, Supervision, Writing – review & editing.

Declaration of Competing Interest

The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Appendix A. Supplementary material

Supplementary material