Genetic susceptibility in childhood acute leukaemias: a systematic review

Acute leukaemias (AL) correspond to 25–35% of all cancer cases in children. The aetiology is still sheltered, although several factors are implicated in causality of AL subtypes. Childhood acute leukaemias are associated with genetic syndromes (5%) and ionising radiation as risk factors. Somatic genomic alterations occur during fetal life and are initiating events to childhood leukaemia. Genetic susceptibility has been explored as a risk factor, since environmental exposure of the child to xenobiotics, direct or indirectly, can contribute to the accumulation of somatic mutations. Hence, a systematic review was conducted in order to understand the association between gene polymorphisms and childhood leukaemia risk. The search was performed in the electronic databases PubMed, Lilacs, and Scielo, selecting articles published between 1995 and 2013. This review included 90 case-control publications, which were classified into four groups: xenobiotic system (n = 50), DNA repair (n = 16), regulatory genes (n = 15), and genome wide association studies (GWAS) (n = 9). We observed that the most frequently investigated genes were: NQO1, GSTM1, GSTT1, GSTP1, CYP1A1, NAT2, CYP2D6, CYP2E1, MDR1 (ABCB1), XRCC1, ARID5B, and IKZF1. The collected evidence suggests that genetic polymorphisms in CYP2E1, GSTM1, NQO1, NAT2, MDR1, and XRCC1 are capable of modulating leukaemia risk, mainly when associated with environmental exposures, such as domestic pesticides and insecticides, smoking, trihalomethanes, alcohol consumption, and x-rays. More recently, genome wide association studies identified significant associations between genetic polymorphisms in ARID5B e IKZF1 and acute lymphoblastic leukaemia, but only a few studies have replicated these results until now. In conclusion, genetic susceptibility contributes to the risk of childhood leukaemia through the effects of gene–gene and gene–environment interactions.


Childhood acute leukaemia
Acute leukaemias (AL) are the highest incidence malignancy in children and adolescents (≤19 years of age) and as a whole, the aetiology has not yet been unveiled. There are two major groups of AL, acute lymphoblastic leukaemia (ALL) and acute myeloid leukaemia (AML), classified according to characteristics presented by leukaemic cells, such as morphological features, surface antigens, chromosomal and molecular abnormalities [1], and gene expression profile [2].
Observational epidemiology has demonstrated that about 5 to 10% of AL are associated with ionising radiation exposure and congenital genetic syndromes (Down, neurofibromatosis, Fanconi anaemia, and Bloom syndrome) [3], that are associated with specific leukaemia subtypes. For the remaining 90% of AL, the aetiopathology is postulated to be a multistep process and somatic mutations are the start point of the pathway. When initiated during fetal life, the majority of them require postnatal events that contribute to accumulation of secondary mutations and proliferative advantage [3][4][5]. In this regard, the initiating event originating in cells would take advantage of the genetic predisposition conferred, in part, by genetic susceptibility and damage from exogenous exposures [6].

Inclusion and exclusion criteria
The inclusion criteria for all publications were: 1) case-control genotyping studies published between 1995 and 2013 that tested the risk of genetic polymorphisms with childhood ALL and/or AML (ages ≤ 21 years); 2) studies that provided sufficient data for estimating the risk association with odds ratio (OR), relative risk (RR) or interaction odds ratio (IOR); and 3) full text available in English, Spanish, and Portuguese. The exclusion criteria were: 1) studies with a different theme from what was proposed for this review; 2) publications in different languages, otherwise the ones specified; 3) studies that specifically include leukaemia cases related to genetic syndromes-Down, neurofibromatosis, Fanconi anaemia, Bloom syndrome and ataxia-telangiectasia-or secondary leukaemia; 4) articles that include other malignancies besides leukaemia in the same cohort, avoiding extrapolation of results exclusively for leukaemias; 5) articles about family gene transmissions; 6) articles about genes and prognosis; 7) articles about folate genes and immune system; and 8) comments and editorials.

Data extraction
Information was extracted from each eligible article supervised by two investigators (LRA and MSPO), according to the inclusion criteria listed above. The following variables were collected from eligible studies: geographical origin, first author's name, year of publication, leukaemia subtype, number of cases and controls, age, candidate genes investigated, and significant genotyping results. www.ecancer.org ecancer 2015, 9:539

Statistical analysis
The strength of association between different genetic polymorphisms through the case-control method was evaluated by analysis of OR, RR, or IOR, with 95% confidence interval (95% CI), that were collected from the studies. Risk associations were considered significant when the P-value was ≤0.05.

Results and discussion
After screening of the retrieved titles, 312 publications were identified as potentially relevant to this review ( Figure 1). After application of exclusion criteria, 103 publications were pre-selected, of which 22 were reviews, and 81 were case-control studies. After checking for bibliographies of pre-selected publications, nine papers were added, giving a total of 90 case-control publications included in this review. The majority of publications addressed polymorphisms in genes related to xenobiotic system (n = 50), followed by DNA repair genes (n = 16), regulatory genes (n = 15), and genome wide association studies (GWAS) (n = 9). The most frequently analysed gene polymorphisms, presented by at least three papers (Figure 2), were located in genes CYP1A1, CYP2D6, CYP2E1, CYP3A5, EPHX1, GSTM1, GSTT1, GSTP1, MPO, NAT2, NQO1, MDR1 (ABCB1), XRCC1, ERCC2, NBN, ARID5B, and IKZF1. Figure 2 also shows the proportion of publications that showed statistically significant associations for each gene, among the total. Considering only statistically significant results, median ORs for increased risk or protective associations for those genes were calculated, and are demonstrated in Figure 3.

Xenobiotic system
Genetic susceptibility studies related to xenobiotic system are presented in Table 1. The main investigated gene polymorphisms comprised  the genes CYP1A1, CYP2D6, CYP2E1, CYP3A4, CYP3A5, EPHX1, GSTM1, GSTP1, GSTT1, MDR1, MPO, NAT1, NAT2, and NQO1. Most publications are from Asia (39.6%), followed by North America (25%), Europe (20.8%), and South America (14.6%). The great disparity between ethnic groups is remarkable, since the Asian continent includes a wide range of people with distinct genetic backgrounds, such as Caucasians, Turkish, Indians, Japanese, Chinese, and the Korean population, likewise, Americans have diverse ancestries, mainly Caucasian, Hispanic, and African. The vast majority of publications (75%) investigated the genetic susceptibility in ALL only, and 25% in AML + ALL; no one has investigated AML solely.
Few publications [22,24,35,49] have explored polymorphisms of EPHX1*2 (T28C-Tyr113His) and EPHX1*3 (A52G-His139Arg); two out of four showed divergent results: variant alleles *2 and *3 were associated with protection for ALL in Brazilians [35], whereas the *2 allele was associated with increased risk for ALL in Turkish [49]. Given the duality of functions that enzyme EPHX1 performs, its interconnection with CYP450 family, the diversity of xenobiotics presented in the environment, and the differences in allele frequencies among populations, gene polymorphisms of EPHX1 may contribute in an unpredictable way in activation or detoxification of xenobiotics.
Finally, three publications [24,33,35] demonstrated that the variant alleles of MPO, mainly MPO*2 (G-463A) have a protective effect when combined to other gene polymorphisms [33,35]. No study has ever demonstrated an independent effect of MPO*2 in childhood leukaemia susceptibility.

DNA repair
The data from genotyping studies in genes related to DNA repair system are summarised in Table 2. The main investigated gene polymorphisms comprised the genes ERCC2, MLH1, MSH3, NBN, and XRCC1. Six publications (37.5%) were performed in Europeans, five (31.25%) in Asians, four (25%) in north-Americans, and one (6.25%) in Brazilians. The majority of them focused only in ALL.
No publication showed association of 280His allele with childhood ALL. XRCC1 399Gln allele was related to increased risk for ALL among Indians [71], Thai [73], Turkish [75], and Poles [74]. However, when combined with XRCC1 194Arg wild-type allele and the variant alleles ERCC2 751Gln and TYMS 3R, the XRCC1 399Gln allele was related to protection for ALL among Brazilians [68].
The ERCC2 gene encodes a DNA helicase involved in NER pathway, and its polymorphisms (Asp312Asn and Lys751Gln), were explored in five case-control publications [67][68][69][70]73]. No one was able to show an independent association with leukaemia risk. ERCC2 751Gln allele was related to protection for ALL among Brazilians when combined with XRCC1 399Gln-194Trp and TYMS 3R [68]. The haplotype GAA (rs3916874, rs238416, rs171140) was also associated with protection for ALL among north-Americans [70].
Genetic polymorphisms of NBN, which is involved in DNA repair by HR, were reported by three publications [70,76,77]. Two of them showed that 657del5 mutation was related to increased risk for ALL among Poles [76,77]. Recently, five SNPs of NBN gene (rs12680687, rs6470522, rs7840099, rs1805812, rs709816) were associated with protection for a subset of Bcp-ALL in north-Americans [70], reinforcing that the interaction of multiple polymorphisms can influence paediatric leukaemia risk.

Regulatory genes and GWAS
To date, the majority of publications regarding regulatory genes and childhood leukaemia were performed to validate GWAS results. The main investigated gene polymorphisms are summarised in Table 3. The majority of publications focused only in ALL (87.5%) and is from North America, followed by Asians (25%), and Europeans (25%).

Gene-environment interaction
Few publications addressed the interaction between gene polymorphisms, environmental exposures, and childhood leukaemia. Studies performed in children exposed directly and/or indirectly throughout maternal exposures are scarce. To date, environmental exposures that have been explored were smoking (tobacco exposure), pesticides, insecticides, trihalomethanes (chlorination by-products of drinking water) alcohol consumption, paint use and x-rays. The absence of CYP1A1 CGACC haplotype, consisting of five SNPs (-T1761C, -G9893A, Ex7+A131G, C1188T, C11599G), was associated with increased risk for ALL among children with father's smoking or at least one smoker at home [28]. Also, haplotypes of CYP2C8 and MDR1 (ABCB1) were related to increased risk for ALL when associated with paint use and indoor insecticides, respectively among Hispanics living in the US [24]. Another finding was that children exposed to indoor insecticides carrying MDR1 haplotype CGC (C1236T, G2677T/A, C3435T) presented a lower risk for ALL [64], indicating that pesticides' toxic effects may be influenced by efflux through P-glycoprotein complex.
Six publications [102][103][104][105][106][107] estimated risk associations by calculating IOR in a case-only cohort. It was observed that CYP1A1*2A/*2B increased the risk for ALL by five-fold among children exposed to pesticides during maternal pregnancy and childhood, while CYP1A1*2B was related to a protective effect among children with mothers who had smoked 1-20 cigarettes during the first trimester of pregnancy (IOR = 0.1; IC95% 0.01-0.9) and with fathers who had smoked >20 cigarettes between birth and diagnosis (IOR 0.2; IC95% 0.04-0.9) [102,103]. Also, it was observed that among children exposed to higher levels of trihalomethanes in drinking water, the risk for ALL was increased in the presence of the polymorphic variant CYP2E1*5 (IOR 9.75; IC95% 1.10-86.01), and GSTT1 deletion (IOR 9.13; IC95% 1.44-57.82), in pre and post-natal periods, respectively [104]. GSTM1-null genotype and CYP2E1*5 variant were related to increased risk for ALL among children with mothers who had consumed alcoholic beverages during the third trimester of pregnancy (IOR 2.4; IC95% 1.1-5.4) and nursing period (IOR 4.9; IC95% 1.4-16.6), respectively [105]. Thus, it is notable that variations in xenobiotic metabolism resulted from genetic polymorphisms can modulate childhood leukaemia risk.

Conclusions
Great scientific advances in the understanding of paediatric leukaemia have been made. Unlike the adult, who usually develops cancer because of the cumulative effect of environmental exposures during his life, the child, which manifests leukaemia with a short latency period, does not have enough exposure time to allow the initiation of a long carcinogenic process. Thus, genetic susceptibility may play an important role in modulating environmental exposures' effects.
This systematic review gathered publications up to 2013 and was an attempt to overview the risk associations between several gene polymorphisms and paediatric AL. It was possible to collect from the selected studies significant amount of data, which is considered to be a fair representation of international scientific literature on this subject. The vast majority of studies so far focused on evaluating the magnitude of risk of genetic polymorphisms in ALL, mainly because Bcp-ALL is the most frequent type of leukaemia in children. In this context, we also realised that there is still a great need for further investigations on the risk factors for paediatric AML. Regarding the xenobiotic system, gene polymorphisms of CYP2E1, GSTM1, NQO1, NAT2, and ABCB1 (MDR1) were more frequently associated with childhood leukaemia risk, which also showed interaction effect with environmental exposures such as paints, household pesticides, insecticides, smoking, alcohol, and trihalomethanes. Gene polymorphisms related to DNA repair have been little investigated in paediatric leukaemia, maybe because of its association with genetic diseases. However, it was noticed that XRCC1 polymorphisms play an important role in the development of ALL, and postnatal exposure to x-rays can modulate leukaemia risk in the presence of APEX1, MLH1, and XRCC4 gene variants. While interpreting these results, one has to consider that fetuses and infants are naturally more affected than adults by a variety of environmental toxicants, mainly because of differential exposure and physiologic immaturity, which makes them more susceptible to suffer from DNA damage and less capable of detoxifying carcinogenic compounds [108]. And so, genetic polymorphisms involving xenobiotic and DNA repair systems have a major role in modulating the effects of environmental agents in children.
Some limitations were observed in the studies that might make the consolidation of scientific evidence difficult, such as: 1) relatively small number of cases, making it difficult to obtain statistically significant results; 2) ethnic and racial differences between populations, which are reflected in distinct polymorphic allele frequencies and patterns of exposure to environmental agents; 3) little information regarding the effect of gene polymorphisms on the encoded protein; 4) and few considerations about gene-environment interactions. Also, we could observe that the majority of associations provided low risk estimates (OR <2.0), which showed that gene polymorphisms are of low penetrance, and conceptually, are minor parts of multifactorial pathways to childhood leukaemia [6].
More recently, GWAS have identified new gene polymorphisms potentially related to paediatric ALL, particularly involving IKZF1 and ARID5B, which were subsequently replicated in independent studies. As a result, we see genetic susceptibility clearly contributes to childhood leukaemia risk, mainly through gene-gene and gene-environment interactions. Further studies are still needed to confirm the observed associations in different populations and to characterise environmental agents as risk factors for childhood leukaemia.