The effects of genetic polymorphisms on benzene-exposed workers: A systematic review.

BACKGROUND AND AIMS: Benzene is a group I carcinogen, which has been associated with leukemia and myelodysplastic syndrome. Moreover, it has been proposed that polymorphisms in benzene metabolizing genes influence the outcomes of benzene exposure in the human body. This systematic review aims to elucidate the existent relationship between genetic polymorphisms and the risk of developing adverse health effects in benzene-exposed workers.
METHODS: Three databases were systematically searched until April 2020. The preferred reporting items for systematic reviews and meta-analyses method was used to select articles published between 2005 and 2020. Quality assessment and risk of bias were evaluated by the Newcastle-Ottawa scale.
RESULTS: After full-text evaluation, 36 articles remained out of 645 initially screened. The most studied health effects within the reviewed papers were chronic benzene poisoning, hematotoxicity, altered urinary biomarkers of exposure, micronucleus/chromosomal aberrations, and gene methylation. Furthermore, some polymorphisms on NQO1, GSTT1, GSTM1, MPO, and CYP2E1, among other genes, showed a statistically significant relationship with an increased risk of developing at least one of these effects on benzene-exposed workers. However, there was no consensus among the reviewed papers on which specific polymorphisms were the ones associated with the adverse health-related outcomes, except for the NQO1 rs1800566 and the GSTT1 null genotypes. Additionally, the smoking habit was identified as a confounder, demonstrating worse health outcomes in exposed workers that smoked.
CONCLUSION: Though there is a positive relationship between genetic polymorphisms and detrimental health outcomes for benzene-exposed workers, broader benzene-exposed cohorts that take into account the genetic diversity of the population are needed in order to determine which specific polymorphisms incur in health risks.

INTRODUCTION

Benzene is an important chemical and ubiquitous environmental pollutant usually used as a solvent in industrial environments (eg, petrochemical industry, steel plants, shoe manufacturing, etc.). Moreover, it is an important toxicant, given that it is the main component of cigarette smoke, gasoline, crude oil, and automotive emissions. , , , Benzene is classified by the International Agency for Research on Cancer (IARC) as a group I human carcinogen ; furthermore, it is the cause of several hematological disorders, such as anemia, leukopenia, thrombocytopenia, acute myeloid and lymphocytic leukemia, myelodysplastic syndrome, and non‐Hodgkin lymphoma.

The toxicity of benzene has been related to its metabolism, which is illustrated in Figure 1. After benzene inhalation, a number of reactions occur, which involve different enzymes such as NADPH quinone oxidoreductase‐1 (NQO1), myeloperoxidase (MPO), glutathione S‐transferases (GST), hydrolases and CYP enzymes (mainly CYP2E1). , , , These metabolic pathways produce metabolites that are excreted in the urine, for instance trans,trans‐muconic acid (t,t‐MA) and S‐phenylmercapturic acid (S‐PMA). Additionally, enzymes like NQO1 or GSTs catalyze detoxification reactions. ,

FIGURE 1

Metabolic pathways of benzene. ADH, alcohol dehydrogenase; ALDH, aldehyde dehydrogenase; CYP, cytochrome P‐450; DHDH, dihydrodiol dehydrogenase; EPHX1, microsomal epoxide hydrolase 1; GST, glutathione S‐transferase; MPO, myeloperoxidase; NQO1, NAD(P)H quinone dehydrogenase 1

Even though the mechanisms by which benzene exerts its genotoxic and hematotoxic effects have not yet been fully elucidated, it is widely accepted that benzene reactive intermediates can bind covalently to macromolecules including DNA, tubulin, histones, and topoisomerase II in the tissue. Furthermore, the resultant metabolites are produced in conjunction with reactive oxygen species, and therefore, cause oxidative stress and subsequent genotoxicity. This results in cell damage and DNA double‐strand breaks; thus, altering the normal cell cycle, generating carcinogenic effects on the bone marrow and the lympho‐hematopoietic system. It has also been proposed that this aromatic hydrocarbon can produce direct damage to hematopoietic progenitor cells, which could lead to apoptosis or altered responsiveness to cytokines and cellular adhesion molecules. , , , Moreover, benzene toxicity to mature blood cells or stromal cells could disrupt the regulation of hematopoiesis, including maturation, hematopoietic commitment, or mobilization, through the network of chemokines, adhesion molecules, and cytokines.

As mentioned above, industrial environments are an important source of benzene exposure, with workers in major industry sectors (such as petrochemical plants, petroleum refineries, coke and coal chemicals or tire manufacturers) exposed to ranges that vary from 0 to 0.325 mg/m3 to more than 32.5 mg/m3 of benzene, contrasting the environmental exposure of the general population that varies from 0.0028 to 0.04 mg/m3. , Consequently, international agencies have set occupational exposure limits in order to reduce the risk for adverse health outcomes in subjects exposed to this hydrocarbon at their workplace. , Nonetheless, uniformity between these guidelines when establishing occupational exposure limits is lacking, , , , especially considering that some individual factors such as genetic diversity predispose the population to benzene‐related adverse health effects, even at low levels of exposure.

For example, several studies have reported a relationship between polymorphisms of benzene‐metabolizing enzymes and higher susceptibility to benzene toxicity. , , , , Dougherty et al, De Palma et al, and Carbonari et al reviewed, in 2008, 2014, and 2016, respectively, the effect of genetic polymorphisms on biomarkers of exposure and biomonitoring, among benzene‐exposed workers. , , However, since then, new studies have surfaced, and a review that includes benzene health‐related effects other than biomarker excretion is in order. Consequently, in this systematic review, we aim to elucidate the existent relationship between genetic polymorphisms and the risk of developing adverse health outcomes in benzene‐exposed workers.

METHODS

Search strategy

A systematic search, based on preferred reporting items for systematic reviews and meta‐analyses (PRISMA) guidelines, was conducted on Scielo, Pubmed, and Medline databases using Boolean operators, Medical Subjects Heading (MeSH), and non‐MeSH terms: Benzene, occupational, mutations, and polymorphism. The full search strategy was adapted for each database and is listed on the Supporting Information.

Study eligibility criteria

We only included studies that evaluated the effect of at least one polymorphism in different variables, with human subjects older than 18, whose main source of benzene exposure was occupational. We also exclusively added papers written in English or Spanish. Additionally, we filtered the results by only using articles published from 2005 to April 2020. Papers that only focused on environmental exposure were rejected, as were in‐vitro studies. The accepted types of research were solely observational studies such as cross‐sectional and case‐control studies.

Study selection

Article selection was conducted independently by two reviewers (VR‐L and DU‐C), and this process is illustrated in Figure 2. The first search retrieved 645 results, and after the application of two filters (year‐of‐publication and not‐in‐vitro‐studies), followed by narrowing of the search strategy with the use of Boolean operators (see Supporting Information), 549 papers were excluded. We added 21 cross‐references to the remaining 96 articles, found either on the searched databases or on the remaining‐papers references. Those articles were then screened by their titles, and 35 duplicates as well as nine titles that fulfilled the exclusion criteria were excluded. The results of the search were imported to the Zotero software, which was used as a reference manager. Afterward, two reviewers analyzed whether or not the abstracts met the inclusion criteria previously established, and then the same procedure was conducted with the full‐text articles. If there was a disagreement, a third reviewer (TLP‐C) resolved it. After that, 36 papers were included. To avoid the omission of articles relevant to the research, the references included in the reviewed articles were compared and checked.

FIGURE 2

PRISMA search strategy flowchart

Data extraction

A table was created for summarizing the following characteristics from each paper: authors and year of publication, country of publication, sample size, age, gender, evaluated variable, evaluated genes and polymorphisms, quality assessment, and relevant results. The mean summary measures used in this review were odds ratios (OR), adjusted odds ratios (ORadj), P values (P), risk ratios (RR), and frequency ratios (FR).

Protocol and study quality assessment

This systematic review was indexed in the prospective register of systematic reviews (PROSPERO). To assess the quality of these studies, the Newcastle‐Ottawa Scale (NOS) was modified to fit each study type, as stated on the Supporting Information. For case‐control studies, three categories were evaluated: selection, comparability, and exposure; for cross‐sectional studies, the exposure category was replaced with “outcome.” Points were assigned according to the study's quality and bias risk, the maximum number of points each study could get was 9. The higher the number of points, the lower the bias risk was (see Supporting Information, Appxs. B‐E).

RESULTS

Characteristics of eligible studies

A total of 36 articles were retrieved from the conducted systematic search, six were cross‐sectional studies and 30 were case‐control studies. All of them assessed occupationally benzene‐exposed population and evaluated one or more of the following benzene effects: chronic benzene poisoning (CBP), hematotoxicity, altered urinary biomarkers, micronucleus/chromosomal aberrations (CA), and gene methylation. Regarding the NOS, both case‐control and cross‐sectional studies reached an average of 6 points out of 9, the former ranging from 5 to 8, and the latter from 5 to 7. These results are summarized in Table 1.

TABLE 1

Characteristics of the included studies

Authors and year of publication	Country	Sample size	Participants (age range or mean [in years], gender)	Mean benzene exposure	Evaluated variable*	Genes and polymorphisms studied		Results	Quality assessment score
Hosgood et al 26	China	250 workers exposed to benzene 140 unexposed controls	21.5‐39.03 y.o. 138 male 252 female	Exposed workers: 5.4 ppm (SD 12.1 ppm) Unexposed controls: <0.04 ppm	Total WBC count	VEGF rs3025030 rs833058 rs699946 ERCC3 rs4150441 rs6731176 Other genes: BLM, GPX3, IL8RB/IL8RA, RIPK2, IL6, IL6R, IL10/IL19, IL12RB1, WRN, IFNAR2		Increased WBC count was evidenced in exposed workers with: VEGF rs3025030 (variant allele C; P < 0.001) VEGF rs833058 (variant allele C; P = 0.0011) Homozygous variant alleles of rs3025030 ‐ rs833058 ERCC3 rs4150441 (variant allele T; P = 0.0086) ERCC3 rs6731176 (variant allele C; P = 0.0087)	8/9
Xiao et al 27	China	102 patients with CBP 204 patients without CBP	18‐63 y.o. 63 male 243 female	—	Risk of developing CBP	ERCC1 rs11615 rs3212986 ERCC2/ XPD rs13181 rs1799793 rs238406		Increased risk of CBP in non‐smokers with: ERCC1 rs11615 (TT genotype) [OR = 3.21 (95% CI 1.36‐7.60), P = 0.006]	6/9
Sun et al 11	China	303 benzene‐poisoned patients 295 workers occupationally exposed to benzene (controls)	18‐68 y.o. 379 male 219 female	—	Risk of developing CBP	GADD45A rs581000 rs11544978 rs532446 MDM2 Del1518 rs2279744	p14ARF rs3731217 rs3731245 rs3088440	Increased risk of CBP in individuals with: p14ARF TGA/TAG diplotype (P = 0.0006) Decreased risk of CBP in individuals with: p14ARF rs3731245 (GA + AA genotypes) [ORadj = 0.57 (95% CI 0.36‐0.89)] p14ARF TGG/TAA diplotype (P < 0.001) MDM2 Del1518 (WW genotype) plus p14ARF rs3731245 (GA + AA genotypes) [ORadj = 0.25 (95% CI 0.10‐0.62), P = 0.003]	6/9
Sun et al 28	China	345 benzene‐poisoned patients 336 (controls) 37 non‐exposed workers	18–68 y.o. 440 male 280 female	—	Risk of developing CBP	TP53 rs17878362 rs1042522 rs1625895 p21 rs1801270 rs1059234		Decreased risk of CBP in individuals with: p21 rs1801270 (CA + AA genotype) [OR = 0.51 (95% CI 0.32‐0.83)] rs1059234 (CT + TT genotype) [OR = 0.53 (95% CI 0.29‐0.95)]	5/9
Pesatori et al 29	Bulgaria	158 petrochemical workers exposed to benzene 50 unexposed subjects	30.5‐52.1 y.o. 171 male 32 female	1.71 ppm	Total blood cell count	NQO1 rs1800566 CYP2E1 rs2031920 (RsaI) rs6413432 (DraI)		None of the investigated polymorphisms was related with blood cell count	7/9
Torres et al 30	Colombia	30 directly exposed and 60 without occupational exposure	19‐56 y.o. 65 male 25 female	—	DNA damage and urinary biomarker PH	CYP2E1 rs3813867/ rs203192 (PstI/RsaI) rs6413432 (DraI) GSTT1 Null and no null GSTM1 Null and no null		No significant differences were found between DNA damage, urinary phenol level and the polymorphisms evaluated	6/9
Chanvaivit et al 31	Thailand	62 cases 34 controls	16‐60 y.o. 87 male 9 female	Laboratory workers: 24.4 ppb Gasoline service attendants: 112.41 ppb Controls: 1.39 ppb	DNA repair‐capacity, blood biomarkers (blood benzene levels) and urine biomarkers (t,t‐MA)	CYP2E1 CYP2E1 * 1/ * 5 CYP2E1 * 5/ * 5 CYP2E1 * 1/ * 1 NQO1 rs1800566 GSTT1 Null and no null XRCC1 rs25487		XRCC1 rs25487 (399Gln allele) had a lower DNA repair‐capacity than those with 399Arg/Arg genotype (P < 0.01, in laboratory workers only) CYP2E1 * 1/ * 5 and CYP2E1 * 5/ * 5 genotypes had lower benzene levels in blood than those with the CYP2E1 * 1/ * 1 genotype (in laboratory workers only) NQO1 and GSTT1 genotypes: no effects on t,t‐MA levels	5/9
Gu et al 32	China	152 benzene poisoning patients 152 control workers (occupationally exposed to benzene)	19‐61 y.o. 118 male 186 female	40 mg/m3: Cases: 18.4% Controls: 21.7% 41‐100 mg/m3: Cases: 61.2% Controls: 61.8% >100 mg/m3 Cases: 20.4% Controls: 16.5%	Risk of developing CBP	CYP1A1 rs4646903 CYP2D6 rs1065852 rs1135840 c. 212 G > A	UGT1A6 c.181 T > A UGT1A7 208Trp > Arg SULT1A1 c.638G > A	More susceptibility to CBP in subjects with: CYP1A1 rs4646903 (TT genotype) [ORadj = 1.21 (95% CI 1.03–1.42), P < 0.05] CYP2D6 rs1065852 (CC or CT genotypes) [ORadj = 2.11 (95% CI 1.22‐3.65), P < 0.05] CYP2D6 rs1135840 (CC genotype) [ORadj = 1.69 (95% CI 1.04‐2.74), P < 0.05]	6/9
Wu et al 33	China	152 benzene poisoning patients 152 control workers (occupationally exposed to benzene)	19–61 y.o. 118 male 186 female	40 mg/m3: Cases: 18.4% Controls: 21.7% 41‐100 mg/m3: Cases: 61.2% Controls: 61.8% >100 mg/m3 Cases:20.4% Controls: 16.5%	Risk of developing CBP	hMTH1 rs4866 (Val83Met) hOGG1 rs1052133 (Ser326Cys) hMYH rs3219489 (His324Gln)		Higher risk of CBP with: hMTH1 83Val/Met + Met/Met [ORadj = 2.47 (95% CI 1.03‐5.92)] compared to Val/Val (only in non‐smokers) hOGG1 326Cys/Cys [ORadj = 3.06 (95% CI (1.74‐5.35)] compared to Ser/Ser + Ser/Cys (only in non‐smokers) hMTH1 83Val/Val and hOGG1 326Cys/Cys at the same time [ORadj = 2.57 (95% CI 1.49‐4.42), P < 0.01] Lower risk of CBP with: hMYH 324His/Gln + Gln/Gln [ORadj = 0.15 (95% CI 0.03‐0.68)] (only in smokers)	6/9
Zhang et al 34	China	152 benzene poisoning patients 152 control workers (occupationally exposed to benzene)	19–61 y.o. 118 male 186 female	40 mg/m3: Cases:18.4% Controls: 21.7% 41‐100 mg/m3: Cases: 61.2% Controls: 61.8% >100 mg/m3 Cases: 20.4% Controls: 16.5%	Risk of developing CBP	XRCC1 rs1799782 (Arg194Trp) rs25489 (Arg280His) rs25487 (Arg399Gln) APE1 rs1130409 (Asp148Glu)	ADPRT rs1136410 (Val762Ala) XRCC2 rs3218536 (Arg188His) XRCC3 rs861539 (Thr241Met)	Higher risk of CBP with: ‐XRCC1 rs25489 *Arg/His [ORadj = 1.67 (95% CI 1.02‐2.74), P = 0.04] compared to Arg/Arg *Arg/His+His/His in non‐smokers [ORadj = 1.96 (95% CI 1.14‐3.38)] and non‐alcohol users [ORadj = 1.78 (95% CI 1.05‐3.03)] only Haplotypes of XRCC1 *194Arg, 280His and 399Arg [ORadj = 2.96 (95% CI 1.60‐5.49), P = 0.001] Lower risk of CBP with: XRCC1 rs1799782 *Arg/Trp + Trp/Trp [ORadj = 0.60 (95% CI 0.37‐0.98), P = 0.041] compared to Arg/Arg APE1 rs1130409 *Asp/Glu + Glu/Glu in alcohol users [ORadj = 0.11 (95% CI 0.02‐0.69)]	6/9
Xue et al 35	China	102 CBP patients 204 controls	18–63 y.o. 63 male 243 female	—	Risk of developing CBP	XRCC1 rs25487 rs25489 rs1799782 CD3EAP rs96759 PPP1R13L rs1005165	XPB/ERCC3 rs4150441 XPC rs2228001 rs227901 XPF rs4781560	Higher risk of CBP with: XRCC1 rs25487 (AA genotype) [ORadj = 14.89 (95% CI 6.54‐30.21), P < 0.001] XRCC1 rs1799782 (TT genotype): only in alcohol drinkers [OR = 8.0 (95% CI 1.32‐48.64), P = 0.022], males [OR = 9.33 (95% CI 1.59–54.67), P = 0.019] and < 12 year exposure [OR = 2.61 (95% CI 1.05‐6.51), P = 0.035] XPB/ERCC3 rs4150441: GA [OR = 1.73 (95% CI 1.00‐2.99), P = 0.049] and GA + AA genotypes [OR = 1.72 (95% CI 1.01‐2.9), P = 0.043] compared to GG genotype Lower risk of CBP in males with: PPP1R13L rs1005165 (CT, TT and CT + TT genotypes) (P < 0.05) CD3EAP rs967591 (GA and GA + AA genotypes) (P < 0.05)	6/9
Mansi et al 4	Italy	181 occupationally exposed petrochemical workers (cases) 134 administrative employees (controls)	23‐65 y.o. 309 male 6 female	0.0368 mg/m3 (0.01 ppm)	Urinary biomarkers: S‐PMA, t,t‐MA and t,t‐MA/S‐PMA ratio	GSTP1 rs1695 GSTM1 Null and no null GSTT1 Null and no null		Lower S‐PMA and higher t,t‐MA/S‐PMA ratio in subjects with GSTT1 null genotype (compared to no null genotype) (P < 0.001) Lower S‐PMA and higher t,t‐MA/S‐PMA ratio in subjects with GSTM1 null genotype compared to no null genotype (P < 0.001, only in smokers)	7/9
Mitri et al 36	Brazil	114 gas‐station attendants: 72 with clinical findings (CF) 52 with no clinical findings (NCF)	19‐82 y.o. 87 male 27 female	—	Risk of developing CBP	CYP2E1 rs2031920 rs6413432 NQO1 rs1800566 MPO rs2333227	GSTM1 Null and no null GSTT1 Null and no null	GSTM1 null genotype was associated with changes related to CBP (ie, symptoms, altered MCV and neutrophil %) [OR = 5.13 (95% CI 1.13‐23.15)]	7/9
Xing et al 37	China	77 benzene‐exposed workers 25 unexposed controls	43‐67 y.o. 38 male 64 female	For 34 exposed workers: 324 ppm‐years For 43 exposed workers: >100 ppm‐years	Altered DNA methylation and total WBC count	CYP1A1 rs4646903 EPHX1 rs1051740 rs2234922 NQO1 rs1800566 Methylation levels of: BLM, CYP1A1, EPHX1, ERCC3, NQO1, NUDT1, p15, p16, RAD51, TP53, and WRAP53		ERCC3 showed an increased methylation level in exposed workers (P = 0.048) Increased number of C allele for EPHX1 rs1051740 was associated with decreased methylation level of the ERCC3 gene in exposed workers (P = 0.001) Reduced WBC count was associated with increasing number of G allele for EPHX1 rs2234922 in exposed workers (P = 0.044) Increased WBC count was related to increasing number of C allele for CPY1A1 rs4646903 in exposed workers (P = 0.001)	6/9
Fustinoni et al 2	Italy	308 cases (urban policemen, gas station attendants and bus drivers) 107 controls	28.9‐48.1 y.o. 352 male 63 female	Gas station attendants: 61 μg/m3 Urban policemen: 22 μg/m3 Bus drivers: 21 μg/m3 Controls: 7.5 μg/m3	Urinary biomarkers: S‐PMA, t,t‐M, U‐benzene and U‐cotinine	CYP2E1 rs2031920 (RsaI) rs6413432 (DraI) NQO1 * Polymorphism not specified		Higher t,t‐MA in exposed subjects with at least one variant allele in CYP2E1 rs6413432 (P = 0.03) Reduced U‐benzene excretion in subjects with at least one mutant allele of CYP2E1 rs2031920 (P < 0.01) All the biomarkers were influenced by smoking	7/9
Manini et al 5	Italy	239 workers (taxi drivers, traffic policemen and gasoline pump attendants)	27.7‐54.5 y.o. 170 male 69 female	38.3 μg/m3	Urinary biomarkers: S‐PMA, t,t‐MA and biomarkers of nucleic acid oxidation: 8‐oxodGuo, 8‐oxoGuo and 8‐oxoGua	NQO1 rs1800566 NQO1 * 1 * 1 NQO1 * 1 * 2 GSTM1 Null and no null GSTT1 Null and no null GSTA1 GSTA1 * A * A GSTA1 * A * B GSTA1 * B * B		Subjects bearing the NQO1 * 1 * 1 (wild‐type genotype) showed lower levels of oxidative damage to RNA compared to subjects with at least one variant allele (P < 0.05) Lower S‐PMA excretion with GSTM1 null (P = 0.01), GSTT1 null (P = 0.023) and GSTA1 * B * B (P = 0.048) genotypes compared to positive genotypes In subjects defective for one GST enzyme, the other one could effectively play a vicarious activity	7/9
Sun et al 38	China	268 benzene‐poisoned patients 268 workers occupationally exposed to benzene	17‐68 y.o. 342 male 194 female	40 mg/m3: Cases: 53.7% Controls: 55.6% 41‐100 mg/m3: Cases: 34.7% Controls: 35.1% >100 mg/m3 Cases: 11.6% Controls: 9.3%	Risk of developing CBP	CYP1A1 rs4646421 rs4646422 rs1048943 rs4646903 CYP1A2 rs2445618 rs762551 rs2472304 rs2470890 CYP1B1 rs1056836 ADH1B rs1229984 EPHX1 rs2854451 rs3738047 rs2234922 rs1051741 EPHX2 rs781141	NQO1 rs1800566 MPO rs7208693 GSTP1 rs1695 UGT1A6 rs6786892 rs1105879 rs4124874 rs3755319 rs887829 rs4148323 Haplotypes and diplotypes of CYP1A1 CYP1A2 EPHX1 UGT1A6	Higher risk of CBP in: ‐EPHX1 GGAC/GAGT (P = 0.00057) or AGAC/GAGT (P = 0.00086) diplotypes Decreased risk of CBP in GSTP1 rs1695 (AG + GG genotype) [OR = 0.44 (95% CI 0.24‐0.81), P = 0.007] only in non‐alcohol drinkers Higher risk of CBP in alcohol drinkers with: EPHX1 rs3738047 GA + AA genotype [OR = 5.0 (95% CI 0.89‐30.52), P = 0.073] compared to GG genotype Decreased risk of CBP in alcohol drinkers with: EPHX1 rs2234922 AG + GG compared to GG (P = 0.008) or rs1051741 CT + TT compared to CC (P = 0.043)	7/9
Chen et al 39	China	100 workers with CBP 90 controls	37 male 63 female	—	Risk of developing CBP	NQO1 rs1800566 MPO rs2333227 CYP2E1 rs2031920 rs6413432 GSTM1 Null and no null GSTT1 Null and no null		Higher risk of CBP in: NQO1 rs1800566 TT genotype [OR = 2.82 (95% CI 1.42‐5.58)] compared to CT + CC genotypes GSTT1 null genotype [ORadj = 1.91 (95% CI 1.05‐3.45)] NQO1 rs1800566 TT genotype plus GSTT1 null genotype [OR = 4.59 (95% CI 1.73‐12.20)] compared to CT + CC plus no null genotypes NQO1 rs1800566 TT genotype plus GSTT1 null genotype plus GSTM1 null [OR = 16.13 (95% CI 3.15‐83.33) NQO1 rs1800566 CT + CC genotype plus GSTT1 null genotype [OR = 2.21 (95% CI 1.09‐4.46)] compared to CT + CC genotype plus no null genotype	6/9
Lan et al 9	China	250 workers exposed to benzene 140 unexposed controls	21.5‐39.03 yo 138 male 252 female	Exposed workers: 5.4 ppm (SD 12.1 ppm) Unexposed controls: <0.04 ppm	Total WBC count	ICAM1 rs5491 VCAM1 rs1041163 rs3176879 CSF2 rs1469149 CSF3 rs1042658 IL‐1A rs1800587 IL‐1B rs16944 IL‐2 rs2069762 IL‐4 rs2243248 IL‐4R rs1805010 IL‐5 rs2069812	IL‐10 rs1800871 IL‐12A rs568408 IL‐12B rs3212227 IL‐13 rs20541 IL‐16 rs859 LTA rs909253 TNF rs1800629 CCR2 rs1799864 CCR5 rs2734648 IL‐8 rs4073	Decreased WBC count in exposed workers with: IL‐1A rs1800587 CT + TT genotype (P < 0.001) compared to CC IL‐4 rs2243248 TG genotype (P = 0.0046) compared to TT IL‐10 rs1800871 CC genotype (P = 0.0034) compared to TT IL‐12A rs568408 AA genotype (P < 0.001) compared to GG VCAM1 rs1041163 CC genotype (P = 0.0022) compared to TT Increased WBC count in exposed workers with: CSF3 rs1042658	8/9
Shen et al 40	China	250 workers exposed to benzene 140 unexposed controls	21.5–39.03 y.o. 138 male 252 female	Exposed workers: 5.4 ppm (SD 12.1 ppm) Unexposed controls: <0.04 ppm	Total WBC count	WRN rs4987236 rs2725349 rs1800392 rs2725362 rs4987036 rs1346044 TP53 rs1042522 NBS1 rs1805794 BRCA1 rs16940 rs799917 rs16941	BRCA2 rs1799943 rs1801406 rs543304 rs766173 rs144848 rs1799944 rs1799955 XRCC3 rs861539 XRCC4 rs3734091 rs1805377 rs1056503	Decreased WBC count in exposed workers with: WRN Homozygous variants in: rs4987236 (P = 0.0003), rs2725349 (P = 0.022), rs1800392 (P = 0.001), rs2725362 (P = 0.0003) TP53 rs1042522 (P = 0.001) BRCA2 Homozygous for uncommon allele of: rs1801406 (P = 0.045)	8/9
Shen et al 1	China	250 workers exposed to benzene 140 unexposed controls	21.5–39.03 y.o. 138 male 252 female	0.36 ± 0.31 ppm	Total WBC count	MBP rs470261 VCAM1 rs1041163 rs3176867 ALOX5 rs4948671 rs7099684 MPO rs2071409 RAC2 rs2239773 CRP rs180094		Decreased granulocyte, lymphocyte, and monocyte population counts in: VCAM1 rs3176867 (P < 0.0001) ALOX5 rs709968 (P = 0.0001) MPO rs2071409 (P = 0.0001)	8/9
Lan et al 41	China	250 workers exposed to benzene 140 unexposed controls	21.5–39.03 y.o. 138 male 252 female	Exposed workers: 5.4 ppm (SD 12.1 ppm) Unexposed controls: <0.04 ppm	Total WBC count	APOB rs3791981 IGF2R rs1570070 IL1A rs17561 GSK3B rs1719888 WRN rs2230009 rs2725362 TP53 rs12951053 GPX3 rs8177426	RXRA rs1805352 BLM rs2270132 CSF3 rs3917979 RAD51 rs4924496 EFNB3 rs3744262 IL10b rs1800871 MPO rs2071409 WDR79 rs17885803 rs2287499	Decreased WBC count in exposed subjects with: BLM rs2270132 (P = 0.00021), rs1694489 (P = 0.0038) and rs414634 (P = 0.0077) RAD51 rs4924496 (P = 0.00053) TP53 rs12951053 (P = 0.0011) WRN rs2725362 (P = 0.00029) and rs2230009 (P = 0.0002) WDR79 rs2287499 (P = 0.005)	8/9
Ye et al 7	China	Cases: 385 exposed workers Controls: 220 healthy subjects	19‐57 y.o. 317male 288 female	6.4 mg/m3	Total WBC count	GSTT1 Null and no null GSTM1 Null and no null GSTP1 rs1695 CYP2E1 rs3813867 rs2031920 rs6413432 mEH rs1051740 rs2234922		Decreased WBC counts in exposed subjects with: GSTT null genotype (P = 0.045) compared to no null genotype GSTM1 null genotype (P = 0.03) compared to no null genotype CYP2E1 rs2031920 CT genotype compared to CC (P = 0.020) and rs3813867 GC genotype compared to GG (P = 0.014)	7/9
Kim et al 42	Korea	108 workers directly exposed to benzene 33 office workers	30–52 y.o.	0.51 ppm	MN and CA	NQO1 rs1800566 MPO rs2333227 XRCC1 rs25487		Exposed workers with NQO1 TT genotype had increased MN [RR = 1.9 (95% CI 1.5–2.3)] and CA [RR = 2.6 (95% CI 1.7‐3.9)] compared to those with CT or CC genotypes A rise in CA on subjects with MPO GG genotype [RR = 2.3 (95% CI 1.3‐4.0)] and XRCC1 AA genotype [RR = 2.2 (95% CI (1.5‐3.1)] compared to those with MPO GA or AA and XRCC1 GG or AG respectively	6/9
Fang et al 43	China	461 exposed workers 88 controls	25.1‐27.7 y.o. 484 male 65 female	Less than 0.6 mg/m3	MN	NQO1 rs1800566 CYP2E1 rs3813867		Lower MN frequencies in exposed subjects with NQO1 TT genotype [FR = 0.79 (95% CI 0.66‐0.95), P < 0.05] compared to the CC genotype	6/9
Zhang et al 44	China	294 benzene‐exposed participants 102 controls indoor workers	17‐71 y.o. 174 male 222 female	6.4 mg/m3	MN and methylation	XRCC1 rs25489 rs25487 APE1 rs1130409 XPA rs1800975	XPC rs2228000 rs2228002 ERCC2 rs13181 rs1799793 XPG rs17655 ERCC1 rs3212986	Higher MN frequency on workers with: ‐XRCC1 rs25487 AA genotype [FR = 1.50 (95% CI 1.16–1.9), P = 0.002] compared to GG; and GA genotype [FR = 1.20 (95% CI 1.06‐1.37), P = 0.006] ‐APE1: rs1130409 GG genotype [FR = 1.28 (95% CI 1.05‐1.55), P = 0.01] compared to TT; and GT genotype [FR = 1.20 (95% CI 1.04‐1.37), P = 0.012] ‐XPG rs17655 GC genotype [FR = 1.18 (95% CI 1.02‐1.38), P = 0.038] compared to GG ‐ERCC1: rs3212986 TT genotype [FR = 1.55 (95% CI 1.31‐1.83), P < 0.001] compared to GG Low global DNA methylation in subjects with APE1 rs1130409 GG + GT genotype (P = 0.045)	6/9
Nourozi et al 6	Iran	Cases: 124 petrochemical plant benzene‐exposed workers Controls: 184 subjects with a similar exposure scenario	27.62‐40.9 y.o. All male	Cases: 0.10 ± 0.195 ppm Controls: 0.12 ± 0.284 ppm	Total WBC count	GSTP1 rs1695 CYP2E1 rs3813867 GSTM1 null and no null GSTT1 null and no null		GSTT1 null was associated with lower platelet count (P = 0.015) and higher risk for hematological disorders [OR = 2.1 (95% CI 1.23‐3.56)] compared to GSTT1 positive Higher leukocyte counts with GSTM1 null compared to GSTM1 positive (P = 0.026)	8/9
Kim et al 3	China	250 benzene‐exposed workers 136 control workers	21‐43 y.o. 248 males 138 females	0.512 ppm	Urinary biomarkers: t,t‐MA, S‐PMA, PH, CAT, and HQ	CYP2E1 rs203192 NQO1 rs1800566 rs4986998 EPHX1 rs1051740 rs2234922 GSTT1 Null and no null GSTM1 Null and no null GSTP1 rs947894 MPO rs2333227		NQO1 rs1800566 lowered t,t‐MA, S‐PMA (P = 0.001), PH (P = 0.022), CAT (P = 0.036) and HQ (P = 0.036) CYP2E1 rs2031920 affected t,t‐MA (P < 0.001), PH (P < 0.001), HQ (P < 0.001) and S‐PMA EPHX1 rs1051740 or rs2234922 affected CAT and S‐PMA GSTT1 null and GSTM1 null lowered S‐PMA (P = 0.018) MPO rs2333227 showed no effect on urinary biomarker excretion	8/9
Carbonari et al 45	Italy	301 oil refinery workers in Italy	30.6‐53.4 y.o.	0.021 mg/m3	Urinary biomarkers: S‐PMA, t,t‐MA	GSTA1 rs3957356 GSTT1 Null and no null GSTM1 Null and no null EPHX1 rs67892231	NQO1 rs1800566 CYP2E1 rs2031920 CYP1A1 rs1048943 MPO rs2333227	Lower median S‐PMA urinary concentration and a consequently higher t,t‐MA/S‐PMA (R value) in smokers with GSTT1 null and GSTM1 null compared to no null genotypes (P < 0.05 for both genes) Higher R value in non‐smokers with GSTT1 null compared to no null (P < 0.05) Lower median R value (higher S‐PMA) in non‐smokers with: NQO1 rs1800566 wild‐type compared to heterozygous and mutant genotypes (P < 0.05) GSTA1 rs3957356 wild‐type compared to heterozygous and mutant genotypes (P < 0.05)	7/9
Zhang et al 46	China	410 benzene‐exposed shoe factory workers 102 control participants	236 male 276 female	6.4 mg/m3	MN and methylation	DNMT3A rs36012910 rs1550117 R882 DNMT3B rs1569686 rs2424909 rs2424913		Increased MN frequency in subjects with DNMT3A rs1550117 variant allele (AG + AA) [FR = 1.19 (95% CI 1.05‐1.36), P = 0.003] Lower global DNA methylation (P = 0.094) and higher MN frequency [FR = 1.18 (95% CI 0.99‐1.40), P = 0.054] in subjects with DNMT3A (R882) variant allele (R882C + R882H) compared to wild‐type genotype Decreased global DNA methylation in subjects with DNMT3B rs2424909 GG genotype (P = 0.031)	7/9
Lin et al 47	Taiwan	105 exposed workers from Taiwan	33‐57 y.o. all males	Groups: High benzene exposure (1 ppm; n = 33) 15 ± 19 ppm Low benzene exposure (<1 ppm; n = 37) 0.20 ± 0.22 ppm	Urinary biomarkers: S‐PMA, PH and t,t‐MA	GSTT1 Null and no null GSTM1 Null and no null GSTP1 rs1695		GSTT1 null is related to a reduced S‐PMA excretion (P = 0.041), compared to GSTT1 no null	5/9
Qu et al 48	China	130 exposed workers 51 unexposed workers	‐	Groups: GSTT1 null: 7.5 ± 9.1 ppm no null: 11.7 ± 20.6 ppm NQO1 rs1800566 Wild‐type variant: 12.1 ± 23.6 ppm Homozygous variant:8.4 ± 11.8 ppm Heterozygous variant:10.3 ± 11.8 ppm	Urinary biomarkers: S‐PMA, PH and t,t‐MA	CYP2E1 rs2031920 rs6413432 NQO1 rs1800566 GSTT1 null and no null MPO rs 2 333 227 (not analyzed)		GSTT1 null is related to a reduced S‐PMA excretion (P < 0.0001), compared to GSTT1 no null	8/9
Carrieri et al 49	Italy	28 petrochemical workers from Italy	33.3‐50.3 yo All males	34.5 μg/m3	Urinary biomarkers: S‐PMA and t,t‐MA	GSTT1 null and no null GSTM1 null and no null		GSTT1 null is related to a reduced S‐PMA excretion (P = 0.0098) compared to GSTT1 no null GSTM1 null did not influence biomarker excretion	6/9
Zhang et al 2014	China	Cases: 385 benzene‐exposed workers Controls: 197 non‐exposed workers	289 male 293 female	6.4 mg/m3	MN	GSTM1 null and no null GSTT1 null and no null GSTP1 rs1695 CYP2E1 rs3813867 rs2031920 rs6413432 mEH exon 3 rs1051740 mEH exon 4 rs2234922		Higher MN frequency in subjects with: CYP2E1 rs3813867 mutant allele (CC + GC) [FR = 1.15 (95% CI 1.02‐1.29), P = 0.02] and rs2031920 variant allele (CT + TT) [FR = 1.23 (95% CI 1.09‐1.37), P < 0.01] both SNPs compared with the wild type Higher MN frequency (adjusted for age, gender and cumulative exposure dose) in subjects with rs2031920 variant allele (CT + TT) [FR = 1.17 (95% CI 1.04‐1.31), P < 0.01], compared to the wild type	7/9
Wan et al 50	China	120 workers	46 male 74 female	—	Risk of developing CBP	GSTM1 Null and no null GSTT1 Null and no null NQO1 rs1800566 CYP2E1 rs3813867		Increased risk of CBP in exposed workers with GSTT1 null [OR = 4.45 (95% CI 1.13‐17.54)] NQO1 rs1800566 plus GSTT1 null at the same time [OR = 1.14 (95% CI 0.42‐3.05)]	5/9
Carrieri et al 51	Italy	146 workers employed at an oil refinery 25 non‐exposed participants as a control group	All males 20‐72 y.o.	32.6 ± 50.6 (μg/m3) for exposed workers 11.5 ± 3.2 (μg/m3) for controls	Urinary biomarkers: S‐PMA, urinary benzene and t,t‐MA	GSTT1 Null and no null GSTM1 Null and no null		GSTT1 no null significantly increases the urinary levels of S‐PMA (P < 0.0094), compared to GSTT1 null GSTM1 null and no null showed no effect on biomarker excretion	8/9

Abbreviations: CA, chromosomal aberrations; CAT, catechol; CBP, chronic benzene poisoning; FR, frequency ratio; HQ, hydroquinone; MN, micronucleus; OR, odds ratio; ORadj, adjusted odds ratio; PH, phenol; RR, risk ratio; S‐PMA, S‐phenylmercapturic acid; t,t‐MA, trans,trans‐muconic acid; WBC, white blood cell; y.o., years old.

The evaluated variables were changed in risk of developing chronic benzene poisoning, excretion of urinary biomarkers, blood cell count or hematotoxicity; the presence of micronucleus, chromosomal aberrations, and methylation.

rs3791981

rs1570070

rs1719888

rs2230009

rs12951053

rs8177426

rs1805352

rs2270132

rs3917979

rs4924496

rs3744262

rs17885803

rs2228000 rs2228002

Effects of polymorphisms on susceptibility to CBP

There were 10 studies that researched the relationship between polymorphisms and CBP (see Table 2).

TABLE 2

Effect of different polymorphisms on the development of CBP

Group/gene	Genes and polymorphisms	Effect on CBP	Risk	References
NQO	NQO1	Possible a		38
	rs1800566		No change	39
	rs1800566 (T/T genotype)		Increased	50
	rs1800566 (combined with null		Increased
	GSTT1)
MPO	MPO rs7208693 rs2333227	No b	No change No change	38 39
CYP	CYP1A1 rs4646421 rs4646422 rs1048943 rs4646903 rs4646903 (T/T genotype)	Conflicting c	No change No change No change No change Increased	38 38 38 38 32
	CYP1A2 rs2445618 rs762551 rs2472304 rs2470890	No b	No change No change No change No change	38 38 38 38
	CYP2D6 rs1065852 (C/C + C/T genotype) rs1135840 (C/C genotype)	Yes d	Increased Increased	32 32
	CYP1B1 rs1056836	No b	No change	38
	CYP2E1 rs2031920	No b	No change	39
GST	GSTT1 non‐null null	Yes d	No change Increased	39, 50 39, 50
	GSTM1 null (in combination with NQO1 rs1800566 variation [T/T], GSTT1 null) null and non‐null	Conflicting c	Increased No change	39 50
	GSTP1 rs1695 (AA genotype, non‐alcohol drinkers)	Yes d	Increased	38
XRCC	XRCC1 rs25487 (AA genotype) rs1799782 (TT genotype) rs25489 (Arg/His+His/His genotype combination) rs1799782(Arg/Trp + Trp/Trp genotype combination)	Yes d	Increased Increased Increased Decreased	35 35 34 34
	XRCC2 ** rs3218536	‐	‐	34
	XRCC3 rs861539	No b	No change	34
ERCC	ERCC1 rs11615 rs3212986	Yes d	Increased No change	27 27 27 27 35
	ERCC2 rs13181 rs1799793	No b	No change No change
	ERCC3 rs4150441 (GA and GA + AA genotypes)	Yes d	Increased
CDKN2A	CDKN2A rs3731245 (GA + AA genotypes in combination with MDM2 rs3730485 WW)	Yes d	Decreased	11
CDKN1A	CDKN1A rs1801270 (CA + AA genotype) rs1059234 (CT + TT genotypes)	Yes d	Decreased Decreased	28 28
POLR1G *	POLR1G rs967591 (GA and GA + AA genotypes)	Yes d	Decreased	35
PPP1R13L *	PPP1R13L rs1005165 (T genotype)	Yes d	Decreased	35
hMTH	hMTH rs4866	Yes d	Increased	33
OGG1	OGG1 rs1052133	Yes d	Increased	33
MUTYH	MUTYH rs3219489	No b	No change	33
TP53	TP53 rs17878362 rs1042522 rs1625895	No b No b No b	No change No change No change	28 28 28
UGT	UGT1A6 rs2070959	No b	No change	32
	UGT1A7 rs11692021	No b	No change	32
SULT1A1	SULT1A1 rs9282861	No b	No change	32
ADH1B	ADH1B rs1229984	No b	No change	38
EPH	EPHX1 rs3738047 (GA + AA genotypes) rs2854451 rs2234922 rs1051741	Yes d	Increased No change No change No change	38 38 38 38
	EPHX2 rs781141	No b	No change	38
UGT1A6	UGT1A6 rs6786892 rs1105879 rs4124874 rs3755319 rs887829 rs4148323	No b	No change No change No change No change No change No change	38 38 38 38 38
GADD45A	GADD45A rs581000 rs532446 rs11544978	Yes d	Decreased Decreased No change	11 11 11
MDM2	MDM2 rs3730485 (in combination with CDKN2A rs3731245) rs2279744	Yes d	Decreased No change	11 11
APE1	APE1 rs1130409	No b	No change	34
ADPRT	ADPRT rs1136410	No b	No change	34
XPB	XPB rs4150441 (GA and GA + AA genotypes)	Yes d	Increase	35
XPC	XPC rs2279017 rs2228001	No b	No change No change	35
				35
XPF	XPF rs4781560	No b	No change	35

Possible: More than half of all the studies that researched that polymorphism has encountered a relationship between it and the development of CBP.

No: None of the studies that researched the polymorphism encountered a relationship between it and CBP.

Conflicting: Half of the studies that researched said polymorphism found a relationship between it and CBP, yet the other half did not.

Yes: All of the studies that researched the polymorphism found a relationship between it and a higher risk of developing CBP.

This effect was exclusively observed in males.

The study did not detect any subjects with the desired allele.

No

No

No

and

Three publications evaluated the difference in susceptibility of developing CBP among patients with polymorphisms in the NQO1 and MPO genes; however, none of these studies found any relationship between the latter gene and the outcome. , , Conversely, only one article found no association between NQO1 polymorphisms and the risk of developing CBP, while the other two found to some degree a greater risk of benzene poisoning on individuals with a NQO1 polymorphism. Chen et al found that the NQO1 rs1800566 TT homozygous genotype was associated with an increased risk of CBP [OR = 2.82 (95% CI 1.42‐5.58)]. Wan et al found that the increase in the risk of CBP was only significant when the NQO1 rs1800566 genotype was present simultaneously as the null GSTT1 gene [OR = 1.14 (95% CI 0.42‐3.05)].

Cytochrome P450 encoding polymorphisms

There were three articles that studied the different CYP gene polymorphisms. Two of them researched CYP1A1, one of which found no relation between the polymorphisms and the risk of CBP, while the other found that the exposed workers with polymorphisms in CYP1A1 rs4646903 are at a greater risk of CBP [ORadj = 1.21 (95% CI 1.03‐1.42)]. Gu et al also discovered that people with CYP2D6 polymorphisms are more susceptible to CBP: [ORadj = 2.11 (95% CI 1.22‐3.65)] for rs1065852 (CC + CT genotype) and [ORadj = 1.69 (95% CI 1.04‐2.74)] for rs1135840 (CC genotype). Nevertheless, none of the articles found any correlation between the possibility of developing CBP and the CYP1A2, CYP1B1, and CYP2E1 polymorphisms. , ,

Three studies examined this correlation. Mitri et al found a relationship between the GSTM1 null genotype and a higher risk of developing CBP [OR = 5.13 (95% CI 1.13‐23.15)] while Chen et al only found it when said polymorphism was combined with the NQO1 rs1800566 TT homozygous genotype and the GSTT1 null [OR = 16.13 (95% CI 3.15‐83.33)]. Two of the papers found that the GSTT1 null genotype was related to a higher CBP risk with an [ORadj = 1.91 (95% CI 1.05‐3.45)] for Chen et al and an [OR = 4.45 (95% CI 1.13‐17.54)] for Wan et al. ,

XRCC1, XRCC2, and XRCC3

Two papers studied this relationship; however, XRCC2 could not be evaluated because the selected variant genotype was not detected. Additionally, they did not find any correlation between the XRCC3 rs861539 variant and variation in CBP risk. , Regarding XRCC1, Zhang et al detected that individuals carrying XRCC1 rs1799782 and rs25489 alleles had a decreased [ORadj = 0.60 (95% CI 0.37‐0.98)] and an increased [ORadj = 1.67 (95% CI 1.02‐2.74)] risk of CBP, respectively. According to Xue et al, the workers who had the XRCC1 rs25487 AA [ORadj = 14.898 (95% CI 6.55‐30.21)] and the rs1799782 TT genotypes also had an increased risk of developing CBP; it is important to mention that the increased risk with rs1799782 was exclusive to male [OR = 9.33 (95% CI 1.59‐54.67)], alcohol drinkers [OR = 8.0 (95% CI 1.32‐48.65), with an exposure lesser than 12 years [OR = 2.61 (95% CI 1.05‐6.51)].

One of the studies evaluated the effect of ERCC1 and ERCC2 and did not find any association between the latter gene and the risk of CBP; nonetheless, it found that individuals carrying the ERCC1 rs11615 TT genotype had an increased risk of benzene poisoning, compared to those carrying the CC genotype [OR = 3.21 (95% CI 1.36‐7.60), P = 0.006].

Other genes

More information about other genes can be found in Table 2. , , , , , ,

Susceptibility to hematotoxicity and changes in blood cell count

Polymorphisms on certain genes could increase susceptibility to hematotoxicity, which could be reflected with an altered blood cell count. We found eight studies that researched this correlation (Table 3).

TABLE 3

Effect of different polymorphisms on the development of hematological changes

Gene/Group	Polymorphisms and/or genotypes	Hematological effect	Effects on blood cell count	References
NQO1	rs1800566	No a	—	29
MPO	rs2071409	Yes b	Decreased WBC count	1
CYP2E1	rs2031920 CT genotype	Yes b	Decreased WBC count	7
	rs3813867	Conflicting c	Decreased WBC count	6, 7
	rs2031920 and rs6413432	No a	‐	29
GST	GSTP1 rs1695	No a	‐	7
	GSTM1 Null genotype	Conflicting c	Decreased WBC count	6, 7
	GSTT1 Null genotype	Conflicting c	Decreased WBC count	6, 7
IL‐1A	rs1800587	Yes b	Decreased WBC count	9
IL‐4	rs22432484	Yes b	Decreased WBC count	9
IL‐10	rs1800871	Yes b	Decreased WBC count	9, 41
IL‐12A	rs568408	Yes b	Decreased WBC count	9
VCAM1	rs1041163	Yes b	Decreased WBC count and CFU‐GEMM	9
	rs3176867	Yes b	Decreased WBC count	1
CSF3	rs1042658	Yes b	Augmented CFU‐GEMM and WBC count	9
ALOX5	rs7099684	Yes b	Decreased WBC count	1
WRN	rs4987236	Yes b	Decreased WBC count	40
	rs2725349	Yes b
	rs1800392	Yes b
	rs2725362	Yes b		40, 41
	rs2230009	Yes b		41
TP53	rs1042522	Yes b	Decreased WBC count	40, 41
	rs12951053	Yes b		41
BRCA2	rs1801406	Yes b	Decreased WBC count	40
BLM	rs2270132	Yes b	Decreased WBC count	41
	rs414634	Yes b
	rs16944894	Yes b
RAD51	rs4924496	Yes b	Decreased WBC count	41
WRAP53	rs2287499	Yes b	Decreased WBC count	41
ERCC3	rs4150441	Yes b	Increased WBC count	26
	rs6731176	Yes b
VEGF	rs3025030	Yes b	Increased WBC count	26
	rs833058

No: None of the studies investigated that the polymorphism encountered a relationship between it and hematological changes.

Yes: All of the studies investigated that the polymorphism encountered a relationship between it and hematological changes.

Conflicting: Half of the studies researched said that polymorphism found a relationship between it and CBP, yet the other half did not.

Two papers researched these two genes. One of them demonstrated that the MPO rs2071409 polymorphism decreases the white blood cell (WBC) count in exposed subjects, and possibly affects WBC subtypes (P < 0.001). NQO1 rs1800566 polymorphism was studied by Pesatori et al, and they found no association between this SNP and blood cell count.

CYP2E1

There were three studies that reported about CYP2E1. A research carried out by Ye et al found that WBC count was lower for individuals who possessed the CT genotype of the CYP2E1 rs2031920 polymorphism compared to the CC genotype (P = 0.02). The GC genotype of the CYP2E1 rs3813867 polymorphism was associated with a significantly lower WBC count when compared to the GG genotype (P = 0.02). rs3813867 polymorphism was also researched by Nourozi et al; however, they did not find a statistically significant relationship between this CYP2E1 SNP and altered blood analysis values. Both CYP2E1 rs2031920 and rs6413432 polymorphisms were evaluated by Pesatori et al, again no significant relationship was found between those SNPs and blood cell count.

GST enzymes ( , , and )

Two of the papers analyzed all three enzymes, , evaluating the GSTM1 null genotype, GSTP1 rs1695 polymorphism, and GSTT1 null genotype. None of them found a correlation between the GSTP1 polymorphism and anomalous hematological indices. However, regarding GSTM1 and GSTT1, the results disagreed: one of the studies found that WBC count in GSTT1 null (P = 0.045) and GSTM1 null (P = 0.03) genotypes decreased compared to the GSTT1/GSTM1 present group, while the other study found that individuals with GSTM1 null genotype had a significantly higher mean value of leukocytes (P = 0.026), and subjects with GSTT1 null genotype presented a lower platelet count (P = 0.015). Nonetheless, this same study observed that subjects with GSTT1 null genotype had a higher risk for hematological disorders compared to those with positive genotype [OR = 2.1 (95% CI 1.23‐3.56)].

More information about other genes can be found in Table 3. , , , ,

Effect on urinary biomarker

Eleven studies researched the influence that several polymorphisms have on the production of different urinary excreted metabolites produced in the metabolism of benzene, commonly used as biomarkers of exposure.

Ten studies analyzed the relationship between the polymorphisms of GST enzymes and the urinary excretion of benzene metabolites. Four of them studied both the enzyme's GSTM1 null and no null genotypes, and they found no correlation between the genotypes and the biomarkers of benzene exposure. , , , Conversely, four other studies found a significant correlation: both Mansi et al and Manini et al found that an expression of the GSTM1 null polymorphism was involved in a lower urinary excretion of S‐PMA (P < 0.001 and P = 0.010 respectively); furthermore, Carbonari et al (P < 0.05) and Kim et al (P = 0.018) discovered similar results. , , , Eight studies established an inverse relationship between the GSTT1 null polymorphism and the quantity of the S‐PMA marker excreted, both in smokers and non‐smokers (P values on Table 1). , , , , , , , Also, according to Chanvaivit et al and these eight studies, there was no association between GSTT1 null and the t,t‐MA metabolite. , , , , , , , , Three studies screened the influence of the GSTP1 polymorphism on urinary biomarkers; however, none found any interaction between these two factors. , ,

Six papers studied the effect of this polymorphism; four of them did not find any correlation. , , , The other two found conflicting results: Kim et al concluded that the workers with an homozygous variant genotype for the CYP2E1 rs2031920 SNP, produced significantly lower levels of t,t‐MA (P < 0.001), phenol (PH) (P < 0.001), and hydroquinone (HQ) (P < 0.001) than workers who had the wild‐type variant allele. Fustinoni et al found a higher t,t‐MA and a lower U‐benzene on subjects with at least one variant allele in CYP2E1 rs6413432 (P = 0.03) and rs2031920 (P < 0.01), respectively.

NQO1

Two out of four studies researched the influence of the NQO1 rs1800566 polymorphism and the biomarkers excretion that did not find any significant relationship between these two variants. , Instead, one found that patients with at least one variant allele of NQO1 rs1800566 affected five metabolites: t,t‐MA, S‐PMA (P = 0.001), PH (P = 0.022), catechol (CAT) (P = 0.036) and HQ (P = 0.036), as they found lower levels of them in these participants. The other study found that the NQO1 rs1800566 wild‐type polymorphism decreased the t,t‐MA/S‐PMA fraction in non‐smokers (P = 0.04).

Micronucleus and CAs

Four studies reported the existing relationship between polymorphisms on certain genes and the expression of cytokinesis‐block micronucleus (MN) and/or the frequency of CA in benzene‐exposed workers and non‐exposed controls.

Two papers studied either or both of these enzymes. , One of them showed that exposed workers with NQO1 rs1800566 polymorphism (TT genotype) had significant increases in MN [RR = 1.9 (95% CI 1.5‐2.3)] and CA [RR = 2.6 (95% CI 1.7‐3.9)] frequencies when compared to controls with CC and CT genotypes; moreover, it suggested that the benzene‐exposed population with the MPO rs2333227 polymorphism (GG wild‐type genotype) had a significant rise in CA frequency [RR = 2.3 (95% CI 1.3‐4.0)] compared to non‐exposed population with GA or AA genotypes. In contrast, the other paper evidenced that mutated homozygous genotype of NQO1 rs1800566 polymorphism (TT genotype) was related with lower MN frequencies [FR = 0.79 (95% CI 0.66‐0.95)] when compared to the homozygous wild‐type genotype (CC genotype).

DNA repair genes

One study analyzed the relationship between polymorphisms on genes involved in the DNA repairing process and the frequency of MN. Both the base excision (XRCC1 and APE1) and nucleotide excision repair pathway genes (XPA, XPC, XPG, ERCC1, and ERCC2) were studied. They found that MN frequencies were higher in XRCC1 rs25487 GA [FR = 1.20 (95% CI 1.06‐1.37), P = 0.006] and AA [FR = 1.50 (95% CI 1.16‐1.90), P = 0.002] alleles, APE1 rs1130409 GT [FR = 1.20 (95% CI 1.04‐1.37), P = 0.012] and GG [FR = 1.28 (95% CI 1.05‐1.55), P = 0.01], XPG rs17655 GC [FR = 1.18 (95% CI 1.02‐1.38), P = 0.038] and ERCC1 rs3212986 TT [FR = 1.55 (95% CI 1.31‐1.83), P < 0.001] with a directly proportional relationship between the number of present mutant alleles of these polymorphisms and MN frequency. Kim et al also studied XRCC1 rs25487 polymorphism, finding that, among exposed workers, subjects with AA variant type displayed a significantly higher CA frequency compared to its wild‐type controls [RR = 2.2 (95% CI 1.5‐3.1)].

One case‐control study carried out by Zhang et al, showed significantly increased MN frequency for carriers of CYP2E1 rs3813867 (CC + GC genotypes) [FR = 1.15 (95% CI 1.02‐1.29), P = 0.02] and rs2031920 (CT + TT genotypes) [FR = 1.23 (95% CI 1.09‐1.37), P < 0.01]; while the opposite was found with the CYP2E1 rs6413432 polymorphism. Another paper also studied the relationship between rs3813867 polymorphism and MN expression in benzene‐exposed workers without a statistically significant increase in MN frequencies for individuals carrying this SNP.

Methylation

Two of the reviewed studies explored the association between genetic polymorphisms and DNA methylation, and whether this methylation was related to benzene exposure. One of them genotyped four commonly studied SNPs on three metabolic enzymes: CYP1A1 (rs4646903), EPHX1 (rs1051740 and rs2234922), and NQO1 (rs1800566); they also analyzed DNA methylation on 11 genes associated with benzene‐induced hematotoxicity (BLM, CY1A1, EPHX1, ERCC3, NQO1, NUDT1, p15, p16, RAD51, TP53, and WRAP53). The authors found that ERCC3 methylation was higher on exposed individuals. Furthermore, they established that a larger number of C alleles on EPHX1 rs1051740 polymorphism was related to a reduction of ERCC3 methylation (P = 0.001), concluding that this SNP may be protective against benzene‐induced hypermethylation. On the contrary, Zhang et al demonstrated that benzene‐exposed workers experienced significant global DNA hypomethylation compared to non‐exposed subjects. As factors that influenced this process, DNMT3A (R882) variant allele (R882C + R882H) (P = 0.094) and DNMT3B rs2424909 polymorphism (GG genotype) (P = 0.031) showed an association with decreased global DNA methylation.

Results adjustment to smoking status

thirty‐one out of 36 included studies incorporated in their analysis a multivariate adjustment for the population that smoked, some demonstrating worse outcomes for smokers compared to non‐smokers. , , , , , , , , , , , , , , , , , , , , , , , , , , , , , , For instance, seven papers found that smoking was an important confounder for benzene biomarkers, as smokers excreted higher concentrations of benzene metabolites than non‐smokers. , , , , , , In five studies, the health outcomes of benzene exposure were only statistically significant when they stratified the population in smokers and non‐smokers. , , , , Moreover, two articles found evidence that smoking affects the prognosis of benzene poisoning and lowers the WBC count in exposed workers. , On the other hand, six papers did not find a statistically significant association between the smoking habit and the researched health outcome. , , , , , Furthermore, in two out of five studies that did not adjust for smoking habits, all of the participants were non‐smokers. ,

DISCUSSION

In this review, we aimed to evaluate the existent relationship between genetic polymorphisms and the risk of developing adverse health effects in benzene‐exposed workers. Among the assessed studies, we encountered that the most researched outcomes of benzene exposure were the development of CBP, the increase or decrease on the excretion of urinary biomarkers and hematotoxic effects. The genes that showed some consistent associations in the effects of their polymorphisms in the human body were NQO1, GSTT1, GSTM1, XRCC1, MPO, and CYP2E1.

NQO1 is a key enzyme involved in benzene metabolism because it reduces benzoquinones to HQ and CAT, resulting in the detoxification of those metabolites. It has been theorized that polymorphisms that cause a decrease in this enzyme's activity probably increase the risk of bone marrow toxicity and other adverse effects. In this review, regarding the polymorphisms on the NQO1 encoding gene, we found that they have a significant effect on the risk of developing CBP, , on MN frequencies and urinary biomarker excretion, , further validating this hypothesis. Two of the evaluated studies found an increased frequency of CBP in individuals with NQO1 rs1800566. , Those results are consistent with a modification in NQO1's detoxifying properties; thus, making the individual's organism more permissive to long‐term toxic effects.

On the other hand, only one study found no relationship between CBP and NQO1 polymorphisms, but it also stated that the sample of exposed workers with the studied polymorphism was probably not big enough to establish a statistically significant relationship in this variable. According to Pesatori et al, changes in the expression of NQO1 in combination with a MPO polymorphism did not show a correlation with altered WBC count ; however, this study did not have enough study subjects to be statistically significant; making it clear that more papers are necessary to reinforce these results.

Regarding biomarkers of exposure, theoretically, if you pare NQO1 activity, fewer benzoquinones will be reduced, subsequently producing less urinary biomarkers. Two studies found that the patients who had the variant NQO1 rs1800566 (C → T) polymorphism (which decreases NQO1's activity) showed a lower excretion of biomarkers, which produced a lower t,t‐MA/S‐PMA fraction. , Conversely, Chanvaivit et al and Qu et al did not find any significant change. , This discrepancy is likely caused by the median level of benzene exposure, which was lower in the subjects of the studies that did not find any correlation between NQO1 polymorphisms and the excretion of urinary biomarkers, compared to the ones that did.

Both Kim et al and Fang et al studied NQO1 rs1800566 involvement in MN frequency and CA; however, their results were contradictory. , This disagreement can be explained by the difference in the population size, as it was bigger in Fang et al's study, which established that the NQO1 CC genotype had a higher MN frequency than the TT genotype. Nonetheless, there are few studies that explore this subject and research with a bigger population sample is needed to understand this phenomenon better.

Considering that GSTs help in the benzene oxide (BO) detoxification process and, by extension, reduce the carcinogenic potential of benzene, the two most studied enzymes of this family within the papers that we reviewed were GSTT1 and GSTM1. All of them considered the null and no null genotypes of these genes as modifying factors of biomarker excretion, CBP, and hematological changes. Regarding urinary biomarker excretion, almost all of the analyzed papers concluded that GSTT1 null genotype was related to lower excretion of S‐PMA, , , , , , , , while the results were very conflicting for GSTM1 null genotype, with four of the articles finding no correlation between this genotype and S‐PMA excretion. , , , However, this is consistent with in vitro studies, which have identified that GSTT1 is more important in the BO detoxification process than GSTM1 because the latter is affected by competing non‐enzymatic product formation and lower enzymatic activity.

Regarding CBP, the importance of GSTT1 was once again demonstrated as a toxicity‐protector enzyme. Two studies associated the GSTT1 null genotype to an increased risk of benzene poisoning , ; moreover, it was found that GSTM1 null genotype has a strong relationship with CBP. The effects of GST enzymes on hematological abnormalities are related to their protective function against benzene, with the reviewed papers showing that GSTT1 null genotype is correlated with lower WBC and platelet count. , It has been recently reported that GST appears to defend against benzene‐induced DNA damage; therefore, with the loss of GSTT1 its DNA‐defensive characteristic is also gone.

CYP2E1 is a phase I enzyme, which plays a key role on the metabolic pathway of benzene, given that it is responsible for the first step of benzene breakdown, producing BO and then intermediate metabolites, which accumulate in the bone marrow and undergo autoxidation or activation by peroxidases to yield the corresponding quinones, which are believed to be among the ultimate toxic metabolites of benzene. Consequently, some of the articles we reviewed determined a relationship between CYP2E1 polymorphisms and effects on hematological abnormalities and biomarker excretion. Concerning hematological abnormalities, the rs2031920 and rs3813867 were two CYP2E1 of the polymorphisms that showed a statistically significant association with an altered WBC count.

As for biomarker excretion, two studies reported a relationship between some of the CYP2E1 polymorphisms and different biomarkers levels. , In accordance with the CYP2E1 function on benzene metabolism, one study showed that the rs2031920 polymorphism was related to lower levels of t,t‐MA, PH, and HQ. Another study demonstrated a relationship between rs2031920 and rs6413432 variant allele polymorphisms with lower U‐benzene and higher t,t‐MA, respectively. Nonetheless, four of the reviewed works did not find a correspondence between CYP2E1 polymorphisms and biomarker excretion changes. , , , This lack of consistency with the results among papers may be a consequence of the diversity of populations in the studies, as the family of cytochrome P450 (CYP450) enzymes might present several SNPs on different ethnical groups, which determines the toxicity of and response to a number of substrates, benzene included.

Another relevant enzyme is MPO, which converts CAT, HQ and 1,2,4‐benzenetriol to highly reactive intermediates: 1,2‐benzoquinone, 1,4‐benzoquinone, and 1,2,4‐benzoquinone. Few studies correlated the MPO encoding gene polymorphisms and human physiological changes, and only one of them found statistically relevant results regarding the rs2071409 polymorphism and hematological changes. Another one suggested that the rs2333227 polymorphism had a significant rise in CA frequency, compared to the non‐exposed population with the GA or AA genotype. All of this can be explained by CAT's increased toxic effect in progenitor cells, which is caused by a decreased MPO metabolic activity. ,

Though not directly involved in the benzene metabolic pathway, the polymorphisms in XRCC1 have shown consistent relationship with worsening adverse effects secondary to benzene exposition. Specifically, the rs25487 polymorphism was found to be associated with higher MN and CA frequencies, , which are indicators of the extent of chromosomal damage in human populations exposed to genotoxic agents, such as benzene, and some studies have found a link between chromosomal damage and an increased cancer risk. Furthermore, rs25487, rs1799782, and rs25489 polymorphisms were found to have an increased risk of developing CBP. , XRCC1 plays an important role in single‐strand break repair and base‐excision repair, acting as a scaffolding protein for other repair factors, including DNA ligase IIIα, DNA polymerase β or APE1. If this repairing function was impaired (which happens with the aforementioned polymorphisms), DNA lesions would accumulate; thus, configuring a threat to genetic stability and cell survival, accelerating mutation rates and increasing CA levels.

Concerning the relationship of the smoking habit and benzene health effects, several authors have found that it is an important source of environmental benzene contamination, and it is directly related to some adverse health outcomes. , In this review, most studies predicted that smoking was a confounding factor and therefore adjusted their analysis to have more reliable results. For instance, some of the reviewed papers found that the smoking habit correlates with worse health outcomes and suggested that future research should take into account this factor while studying occupational exposure. , , , , , , , , , , , , , Conversely, a minority of the included articles did not find a statistically significant interaction between those two variables; however, these results may be caused by the scarce quantity of smokers compared to non‐smokers both in the group with exposed workers and the controls in most of these studies. , , , , ,

These statistically relevant outcomes have established the link between genetic polymorphisms and the risk of developing adverse health effects in benzene‐exposed workers with a different genetic background. These findings should enable occupational medicine specialists, local governments and policy makers to create and improve new evidence‐based guidelines for benzene exposure limits that take into account the genetic diversity of the workforce. Those improved regulations will help workers to avoid health risks, thus lowering public health costs and overall making the population healthier while providing insight for future research.

Strengths and limitations

By using the PRISMA guidelines and the Newcastle‐Ottawa quality assessment score, this review captures a significant number of studies, anticipating and working around bias; nevertheless, weak selection bias could be induced by limiting the language of the included studies to English and Spanish. In addition, publishing bias should not be ignored, because papers that found a correlation between polymorphisms and different benzene‐exposure outcomes are more likely to be published than those with no significant findings. Additionally, some papers used the same study population, which can lead to more bias. Moreover, some polymorphisms did not have the same quantity of evidence as others, which may affect the results.

CONCLUSION

Overall, this review highlights the detrimental effects of occupational exposure to benzene. It also establishes a clear relationship between some polymorphisms and the extent of the consequences that come with the occupational exposure to this toxicant. While there are several studies investigating this topic, there are not enough papers to establish a consensus with statistically relevant results regarding some of the polymorphisms. Future research should focus on gathering broader cohorts with the desired polymorphism, given that the expression of genetic variants was not present in all of the participants, even when the cohort had a higher population. In conclusion, benzene is an important threat to occupational health worldwide; therefore, regulations should be adjusted to protect all the exposed workers, especially those with high‐risk genetic variants.

FUNDING

Publication costs were supported by Ministerio de Ciencia, Tecnología e Innovación (Grant number: COL126780763345, contract RC.847‐2019). However, the funding source did not have any involvement in the study design; collection, analysis, and interpretation of data; writing of the report or the decision to submit the report for publication.

CONFLICT OF INTEREST

The authors declare no conflicts of interest.

AUTHOR CONTRIBUTIONS

Conceptualization: Verónica Ramírez‐Lopera, Daniel Uribe‐Castro, Tania Liseth Pérez‐Cala

Formal Analysis: Verónica Ramírez‐Lopera, Daniel Uribe‐Castro, Tania Liseth Pérez‐Cala, Henry Bautista‐Amorocho, Jorge Alexander Silva‐Sayago, Enrique Mateus‐Sánchez, Wilman Yesid Ardila‐Barbosa

Funding Acquisition: Tania Liseth Pérez‐Cala and Henry Bautista‐Amorocho

Investigation: Verónica Ramírez‐Lopera, Daniel Uribe‐Castro, Tania Liseth Pérez‐Cala

Methodology: Verónica Ramírez‐Lopera, Daniel Uribe‐Castro, Tania Liseth Pérez‐Cala

Project Administration: Verónica Ramírez‐Lopera, Jorge Alexander Silva‐Sayago

Supervision: Enrique Mateus‐Sánchez, Wilman Yesid Ardila‐Barbosa, Tania Liseth Pérez‐Cala

Validation: Henry Bautista‐Amorocho, Jorge Alexander Silva‐Sayago, Enrique Mateus‐Sánchez, Wilman Yesid Ardila‐Barbosa

Writing – Original Draft Preparation: Verónica Ramírez‐Lopera, Daniel Uribe‐Castro

Writing – Review & Editing: Verónica Ramírez‐Lopera, Daniel Uribe‐Castro, Tania Liseth Pérez‐Cala, Henry Bautista‐Amorocho, Jorge Alexander Silva‐Sayago, Enrique Mateus‐Sánchez, Wilman Yesid Ardila‐Barbosa

All authors have read and approved the final version of the manuscript.

Verónica Ramírez‐Lopera had full access to all of the data in this study and takes complete responsibility for the integrity of the data and the accuracy of the data analysis.

TRANSPARENCY STATEMENT

The corresponding author confirms that the manuscript is an honest, accurate, and transparent account of the study being reported; that no important aspects of the study have been omitted; and that any discrepancies from the study as planned (and, if relevant, registered) have been explained.

Appendix S1. Supporting Information

Click here for additional data file.