Six polymorphisms in the lncRNA H19 gene and the risk of a systematic review and meta-analysis

Authors: Maoquan Yang, Mingwei Zhang, Qiong Wang, Xiaojing Guo, Peizhen Geng, Jinhua Gu, Wansheng Ji, Li Zhang

Abstract

Background

Numerous studies have demonstrated long noncoding RNA (lncRNA) play an important role in the occurrence and progression of cancer, and single nucleotide polymorphisms (SNPs) located in lncRNA are considered to affect cancer suspensibility. Herein, a meta-analysis was carried out to better assess the relationship of H19 polymorphisms and cancer susceptibility.

Methods

A literature search was conducted through using PubMed, EMBASE, and Web of Science databases to obtain relevant publications before Aug 23, 2022. The reference lists of the retrieved studies were also investigated to identify additional relevant articles. The pooled odds ratios (ORs) with 95% confidence intervals (CIs) were calculated to appraise the risk of various cancers.

Results

There appeared to be a remarkable correlation between the rs2107425 variation and decreased cancer risk among Caucasians. Nevertheless, the rs217727 polymorphism was significantly associated with an increased risk of lung cancer, hepatocellular carcinoma and oral squamous cell carcinoma. Also, we found a significant correlation between the rs2839698 polymorphism and increased cancer risk among Asians, gastric cancer, hepatocellular carcinoma, hospital-based control and larger simple size subgroups, respectively. Similarly, the rs3741219 mutation was notably related to cancer risk in higher quality score. As for rs3024270 polymorphism, the homozygous model was markedly linked to cancer risk in overall analysis and population-based controls. There was no significant association between the rs3741216 polymorphism and cancer risk.

Conclusion

H19 rs2839698 and rs3024270 were closely associated with overall cancer risk. H19 rs2107425 was related to lower cancer risk among Caucasians, while the rs2839698 was related to increased cancer risk among Asians. Our results supported that H19 SNPs were significantly correlated with cancer risk.

Supplementary Information

The online version contains supplementary material available at 10.1186/s12885-023-11164-y.

Introduction

Cancer has been the second biggest cause of mortality worldwide, seriously endangering public health and increasing economic burden on society [1]. In 2023, 1,918,030 new cancer cases and 609,360 cancer mortalities are estimated to occur in the United States. And prostate, lung, and colorectal cancers account for 48% of all male incident cases, while 51% of all female incident cases are diagnosed with breast, lung, and colorectal cancers [2]. Although the specific pathological mechanism of tumorigenesis still remains unclear, cancer is considered as a complex and multifactorial disease that results from the interaction of environmental and genetic risk factors, such as high-calorie diet, smoking, excessive drinking, obesity, hypertension, diabetes [3–5]. Recent advances in cancer diagnosis and treatment, including multifunctional nanomaterials combined with imaging probes and drugs, nanomedicine products and therapeutic vaccines are improving options for cancer patients [6–8]. On the other hand, preference heterogeneity between patients indicates that tradeoffs between survival benefits and long-term physical, emotional, cognitive, and functional side effects should be carefully considered in treatment decision-making [9]. At present, genome-wide association studies (GWAS) have identified a strong association of several common single nucleotide polymorphisms (SNPs) with cancer risk [10, 11]. Certain genetic SNPs were found to be related to cancer risk, including miR-143/145, CASP9, CASP10 and IL-1β [12–14]. In addition, functional SNPs are present in lncRNA genes and influence gene expression and function through various means, and then result in the occurrence and progression of cancer [15].

Being widely transcribed in the human genome, long non-coding RNAs (lncRNAs) are defined as single stranded non-coding RNAs with a length of more than 200 base pairs and no open reading frames, thereby lacking of protein-coding function, although some of them may produce small functional peptides [16]. LncRNAs take part in numerous cellular processes by interacting with cellular molecules, such as DNAs, RNAs, or proteins [17]. At the levels of epigenetic, transcriptional, and post-transcriptional modifications, they can regulate gene expression via different mechanisms, including chromatin remodeling induction, alternative splicing, intranuclear transport, production of miRNA sponges, and transcriptional interference [18–22]. Interestingly, lncRNAs paly crucial regulatory roles in a variety of physiological and pathological processes and cancer biology, including cell proliferation, differentiation, apoptosis, and carcinogenesis progression [23–27]. It has been found that lncRNAs are dysregulated in various types of cancer, which contributes to tumorigenesis and development of tumors by affecting the expression of oncogenes or tumor suppressors. Generally, lncRNAs are thought to have prospective clinical implications and to be appraised as independent novel biomarkers for diagnosis and prognosis in human cancer treatment [28–30].

As a critical maternally imprinted gene, lncRNA H19 was initially discovered in the 1990s [31]. The H19 gene, possessing five exons and four introns, encodes a 2.3-kb long, capped, spliced, and polyadenylated noncoding RNA, of which the transcript is highly conserved at a cluster with the insulin-like growth factor 2 (IGF2) locus on human chromosome 11p15.5, and plays an essential role in embryonic development and growth control [32–35]. It has been reported that the aberrant expression of H19 was implicated in various types of cancer, including breast, lung, esophageal, gastric, pancreatic, colorectal, liver, bladder and cervical cancer. H19 acts as an oncogene or a suppressor gene, which may be attributed to the heterogeneity of different types of cancer [36–38]. Previous researches have shown that H19 gene polymorphisms are markedly associated with malignancies, however, the results were controversial and inconsistent. Therefore, the aim of this meta-analysis was to accurately examine the correlation between H19 polymorphisms and cancer susceptibility.

Materials and methods

Literature search strategy

Eligible studies were retrieved from the PubMed, EMBASE, and Web of Science electronic databases up to Aug 23, 2022. Our search strategy included the main terms (H19 or long Noncoding RNA H19 or lncRNA H19) and (polymorphism or genotype or SNP) and (carcinoma or neoplasm or cancer or tumor). At the same time, we manually screened out the relevant potential articles in the references extracted.

Selection and exclusion criteria

Inclusion criteria are as (1) case-control studies investigated the relationship between H19 polymorphisms and the risk of cancer; (2) the histopathological diagnosis of cancer patients was clearly defined; the control group did not have any history of cancer; (3) sufficient data on genotype distribution of H19 polymorphisms was applied to calculate the odds ratio (OR) and 95% confidence interval (CI).

The exclusion criteria were as (1) abstract, case reports, comment, editorials and review; (2) duplication of the previous reports; (3) lack of the full text or main genotyping data; (4) non-case-control or cohort design studies.

Data extraction

Two investigators separately conducted literature screening, data extraction, literature quality evaluation, and any disagreements that could be resolved through discussion or a third analyst. The relevant information independently extracted by two investigators included the following information from each first author, year of publication, country of the population, ethnicity, source of controls, genotyping methods, cancer types, sample size and P value of (HWE).

The Newcastle-Ottawa scale (NOS) was adopted to assess the process in terms of queue selection, comparability of queues, and evaluation of results [39]. A study with a score of at least six was considered as a high-quality literature. Higher NOS scores showed higher literature quality.

Statistical analysis

All data analysis was conducted using Stata16.0 software (Stata Corp LP, TX, USA). Odds ratio (OR) and 95% confidence intervals (CIs) were used to evaluate the association between lncRNA H19 polymorphisms and various cancers. After that, the heterogeneity test was conducted. When P ≥ 0.05 or I^2^ < 50% was performed, it indicated that there was no obvious heterogeneity, and the fixed-effect pattern should be applied for a merger. Otherwise, the random-effect model was used. Results were considered significant statistically when the p-value less than 0.05. Subgroup analysis was implemented to determine the source of heterogeneity. Additionally, sensitivity analysis was performed to assess the impact of each individual study on overall results. The Begg’s rank correlation test and Egger’s linear regression test were used to verify the publication bias among these studies. If P < 0.05 indicates obvious publication bias.

False-positive report probability (FPRP) analysis

The probability of meaningful relationships between H19 SNPs and cancer risk can be determined through carrying out the FPRP analysis [40]. In order to investigate the remarkable associations observed in the meta-analysis, we adopted prior probabilities of 0.25, 0.1, 0.01, 0.001, and 0.0001 and computed the FPRP values as described previously. The association that reached the FPRP threshold of < 0.2 was considered significant.

Results

Process of study selection and description of qualified studies

As shown in Fig. 1, the initial 472 studies were retrieved by databases of PubMed (n = 229), Embase (n = 76), Web of science (n = 166). After eliminating 152 duplicate articles, 191 additional publications were excluded by screening the abstract and title. Among these, 147 articles were reviews, letters, conference abstracts, meta-analysis, notes, editorials and short surveys, and 44 articles focused on animal or vitro experiment. After careful review of the full texts, 88 articles were further excluded due to the following 30 articles were involved with other genes or other SNPs of H19, 45 studies were not relevant to cancer and 13 studies had no available data. Finally, the remaining 40 eligible articles were included in this analysis [41–80].

Fig. 1Flow diagram of the eligible study selection process

Through literature search and selection, a total of 40 eligible articles embodying 95 studies were embodying in our study, which included 13 studies for rs2107425, 30 studies for rs217727, 26 studies for rs2839698, 10 studies for rs3741219, 12 studies for rs3024270, and 4 studies for rs3741216 polymorphisms. One article referred to two independent case-control studies, and thus the study was regarded as two separate estimates [44]. Among the included studies, 30 studies were from China, four studies from Iran, two studies form European countries, two studies from Egypt, two studies from the mixed countries, and one study from America. At the same time, 34 studies were conducted in the Asian descent, five studies were conducted in the Caucasian descent and two studies were conducted in the African descent. Thirteen of the studies focused on population-based controls and 27 on hospital-based controls. If the number of different cancer types is less than 1, the cancer type is classified into other cancer subgroup. The detailed characteristics of selected studies are illustrated in Table 1, such as cancer type, genotyping method, sample size, distributions of genotype frequency and Hardy-Weinberg equilibrium. The NOS score of all articles ranged from 6 to 8, implying that all included studies were of high quality.

Table 1Characteristics of all studies included in the meta-analysisFirst authorYearCountryEthnicityGenotypingSource ofGenotype distributionHWENOSCancermethodscontrolCaseControlP-valuetypes 13 Studies for H19 gene rs2107425 C/T polymorphism CCCTTTCCCTTTVerhaegh2008NetherlandsCaucasianPCR-RFLPPB9265208996190.34029BLCSong HL2009MixedCaucasianTaqManPB26192192555402936678420.85659OCQuaye2009MixedCaucasianTaqManPB767544149111810982470.34496OCBarnholtz2010USAAfricanIlluminaPB1613901861703391490.41997BCBarnholtz2010USACaucasianIlluminaPB6045161055214781190.54897BCButt S2012SwedenCaucasianMass ArrayPB361250686685731450.18168BCGong WJ2016ChinaAsianSequenomHB181235637996280.89206LCYin ZH2018ChinaAsianIlluminaHB161266129140185700.51298LCWu2019ChinaAsianTaqManHB134185404225602080.34518HCCHuang MC2019ChinaAsianPCRPB8810738109155480.55908CCYang PJ2019ChinaAsianRT-PCRHB15221366171190700.16367UCCGhapanchi2020IranAsianARMS-PCRPB79942774101250.29118OSCCKhalil2022EgyptAfricanQIAampHB25321320910.99196CRC 30 Studies for H19 gene rs217727 G/A polymorphism GGGAAAGGGAAAVerhaegh2008NetherlandsCaucasianPCR-RFLPPB1145941158090.28809BLCYang C2015NetherlandsCaucasianTaqManHB16025288193244630.29578GCLi SW2016ChinaAsianTaqManHB4805141534565701770.95859CRCHua QH2016ChinaAsianTaqManHB4314671485736651560.07407BLCXia Z2016ChinaAsianCRS-RFLPPB1601561481392121160.05219BCJin TB2016ChinaAsianMass ArrayPB1171032616999160.76519CCGuo QY2017ChinaAsianIlluminaHB101181802523481370.38408OSCCHassanzarei2017ChinaAsianPCR-RFLPHB711322712511320.0000^^7BCHe TD2017IranAsianTaqManHB7910212195165230.12076OsteosarcomaHu PH2017ChinaAsianTaqManHB13320083128196920.30228PCLin YX2017ChinaAsianG0104KHB4034711314654501050.80078BCLi LL2018ChinaAsianTaqManHB21025095246305670.05428LCYin ZH2018ChinaAsianIlluminaHB20426488165172580.23198LCYuan ZY2018ChinaAsianMass ArrayPB18619451488423730.15117OSCCCui P2018ChinaAsianTaqManPB6116921856857732170.96287BCGGGAAAGGGAAAAbdollahzadeh2018ChinaAsianRFLP-PCRHB116295861400.45168BCHu C2019IranAsianTaqManHB18616443382342860.46967NeuroblastomaLi Z2019ChinaAsianTaqManHB5114098490260.80618BLCMohammad2019ChinaAsianARMS-PCRHB793026454120.90037BCWang GZ2019IranAsianTaqManHB1622771254937512910.86766LCWu2019ChinaAsianTaqManHB154170354955391560.62658HCCWei MR2019ChinaAsianTaqManHB8872656344930.0000^^7GCHuang MC2019ChinaAsianPCRPB10210328135139390.72898CCYang PJ2019ChinaAsianRT-PCRHB18520244191188520.58457UCCCao Q2020ChinaAsianRT-PCRHB3435502013504941830.70428RCCGhapanchi2020ChinaAsianARMS-PCRPB11075151336430.12598OSCCDeng YJ2020IranAsianMass ArrayHB254278735575911520.80187GliomaTan TB2020ChinaAsianTaqManHB12668194384101090.38118HepatoblastomaLi WY2021ChinaAsianTaqManPB177130484864691130.99257WilmsPei JS2021ChinaAsianRT-PCRPB11112035114120320.96107Leukemia 26 Studies for H19 gene rs2839698 G/A polymorphism GGGAAAGGGAAAVerhaegh2008NetherlandsCaucasianPCR-RFLPPB54744952109430.31259BLCYang C2015ChinaAsianTaqManHB25019555284178380.17548GCLi SW2016ChinaAsianTaqManHB583462102666462750.66659CRCHua QH2016ChinaAsianTaqManHB552418797295651030.65107BLCGong WJ2016ChinaAsianTaqManHB237220399980270.09826LCGuo QY2017ChinaAsianIlluminaHB133171582443771200.20218ORCHassanzarei2017ChinaAsianPCR-RFLPHB0641660182220.54617BCHe TD2017IranAsianTaqManHB839812178175300.14626OsteosarcomaLin YX2017ChinaAsianG0104KHB4524401134844321040.59988BCYang ML2018ChinaAsianKASPHB21521140245185320.71418HCCCui P2018ChinaAsianTaqManPB8015681228756731290.97937BCHu C2019ChinaAsianTaqManHB17917539365373720.08967NeuroblastomaMohammad2019ChinaAsianARMS-PCRHB1557395355220.24107BCWang GZ2019ChinaAsianTaqManHB277225617126451750.11736LCWu2019ChinaAsianTaqManHB140178415325241340.77188HCCWei MR2019IranAsianTaqManHB9068678878340.0248^^7GCHuang MC2019ChinaAsianPCRPB1159920154134300.91328CCGGGAAAGGGAAAYang PJ2019ChinaAsianRT-PCRHB20617055192184550.29737UCCCao Q2020IranAsianRT-PCRHB51643576615425540.07328RCCDeng YJ2020ChinaAsianPCRHB13414040154211720.05817GliomaYu BQ2020ChinaAsianMass ArrayHB311240546755041210.98477CRCZhang HB2020ChinaAsianMass ArrayHB7093389288160.42606OCTan TB2020ChinaAsianTaqManPB1027833439424940.56798HepatoblastomaLi WY2021ChinaAsianTaqManHB174127544884801000.24537WilmsPei JS2021ChinaAsianRT-PCRPB9113144119117300.87817LeukemiaZhang JZ2021IranAsianPCR-RFLPHB192244137351248890.0000^^8Lymphoma 10 Studies for H19 gene rs3741219 A/G polymorphism AAAGGGAAAGGGYang C2015ChinaAsianTaqManHB26018753268189430.24468GCHassanzarei2017IranAsianPCR-RFLPHB6312642109102290.49797BCCui P2018ChinaAsianTaqManPB7825821278327061390.52917BCAbdollahzadeh2018IranAsianRFLP-PCRHB119247801730.09938BCWu2019ChinaAsianTaqManHB129182485175361370.91408HCCYang PJ2019ChinaAsianRT-PCRHB19218158185190560.51217UCCHuang MC2019ChinaAsianPCRPB11210021152130320.59068CCCao Q2020ChinaAsianRT-PCRHB567416111552389860.14008RCCDeng YJ2020ChinaAsianMass ArrayHB439107596515201290.09567GliomaZhang HB2020ChinaAsianMass ArrayHB190120186090.0000^^6OC 12 Studies for H19 gene rs3024270 C/G polymorphism CCCGGGCCCGGGHua QH2016ChinaAsianTaqManHB1745273462606884470.86867BLCLi SW2016ChinaAsianTaqManHB3855272354205822010.97949CRCGuo QY2017ChinaAsianIlluminaHB751831041453502450.32138OSCCHe TD2017ChinaAsianTaqManHB179185311791730.10146OsteosarcomaYang ML2018ChinaAsianKASPHB95225151812151700.36098HCCHu C2019ChinaAsianTaqManHB99203912134241730.15917NeuroblastomaLi Z2019ChinaAsianTaqManHB16101832297810.37718BLCHuang MC2019ChinaAsianPCRPB511206071150950.42258CCWu2019ChinaAsianTaqManHB87187853345932630.99458HCCYang PJ2019ChinaAsianRT-PCRHB1142101071202081030.48947UCCTan TB2020ChinaAsianTaqManPB5087762644892040.42168HepatoblastomaCCCGGGCCCGGGLi WY2021ChinaAsianTaqManHB120141942905562220.13767Wilms 4 Studies for H19 gene rs3741216 A/T polymorphism AAATTTAAATTTYang C2015ChinaAsianTaqManHB38010218379109120.22108GCHassanzarei2017IranAsianPCR-RFLPHB026204065260.0150^^7BCWei MR2019ChinaAsianTaqManHB7991557078520.0025^^7GCCao Q2020ChinaAsianRT-PCRHB79125548718264350.08348RCCBC: breast cancer; LC: lung cancer; BLC: bladder cancer; GC: gastric cancer; CRC: colorectal cancer; PC: pancreatic cancer; OC: ovarian cancer; CC: cervical cancer; OSCC: oral squamous cell carcinoma; UCC: urothelial cell carcinoma; RCC: renal cell carcinoma. ^^P < 0.05

Correlation between rs2107425 C/T polymorphism and cancer risk

Thirteen relevant studies with 11,972 cancer patients and 17,128 controls were examined for the association between the rs2107425 polymorphism and cancer risk. Compared with the wild-type CC homozygote, the genotypes of rs2107425 were not linked to cancer risk in overall analyses (T vs. C: OR = 0.98, 95%CI = 0.91–1.06, P = 0.595; TT vs. CC: OR = 1.01, 95%CI = 0.88–1.17, P = 0.846; TC vs. CC: OR = 0.96, 95%CI = 0.85–1.07, P = 0.438). Similarly, no relationships were detected in the dominant and recessive models (TT + TC vs. CC: OR = 0.97, 95%CI = 0.87–1.08, P = 0.543; TT vs. TC + CC: OR = 0.98, 95%CI = 091-1.06, P = 0.651; Table 2; Fig. 2). Stratification analysis by ethnicity showed the rs2107425 variation significantly reduced cancer risk among Caucasians (T vs. C: OR = 0.91, 95% CI = 0.85 − 0.7, P = 0.006; CT vs. CC: OR = 0.83, 95% CI = 0.73–0.94, P = 0.003; OR = 0.85, 95% CI = 0.76–0.94, P = 0.003), which might be a protective factor in the Caucasian population. Also, we found a significant association of rs2107425 variant with cancer risk under the heterozygote models in hospital-based subgroup (CT vs. CC: OR = 1.18, 95% CI = 1.00-1.39, P = 0.049) and population-based source of controls (CT vs. CC: OR = 0.87, 95% CI = 0.78–0.97, P = 0.016, Table 2). There was significant association between the rs2107425 variant and elevated risk of CRC (T vs. C: OR = 3.15, 95%CI = 1.51–6.57, P = 0.002; TT vs. CC: OR = 10.40, 95%CI = 0.1.25–86.4, P = 0.030; TC vs. CC: OR = 2.84, 95%CI = 1.11–7.32, P = 0.030; TT + TC vs. CC: OR = 3.60, 95%CI = 1.46–8.88, P = 0.005). The heterozygote and recessive models of rs2107425 notably decreased the risk of hepatocellular carcinoma (TT vs. CC: OR = 0.61, 95%CI = 0.41–0.90, P = 0.012; TT vs. CC + TC: OR = 0.59, 95%CI = 0.41–0.85, P = 0.004, Table 2). Heterogeneity test results suggested that heterogeneity existed in all five genetic models of overall analyses. Heterogeneity was not observed under the allelic, homozygote, and recessive models in Caucasians subgroup.

Table 2Summary ORs and 95% CIs of H19 SNPs and risk of cancerLocusNo.AlleleHomozygoteHeterozygoteDominantRecessiveOR (95%CI) PI^2^ (%)OR (95%CI) PI^2^ (%)OR (95%CI) PI^2^ (%)OR (95%CI) PI^2^ (%)OR (95%CI) PI^2^ (%)rs2107425C/TTotal130.98 (0.91, 1.06) 0.59568.01.01 (0.88, 1.17) 0.84653.90.96 (0.85, 1.07) 0.43866.80.97 (0.87, 1.08) 0.54369.20.98 (0.91, 1.06) 0.65143.2EthnicityCaucasian50.91 (0.85, 0.97) 0.006^^46.20.94 (0.85, 1.04) 0.2266.40.83 (0.73, 0.94) 0.003^^67.00.85 (0.76, 0.94) 0.003^^62.31.01 (0.93, 1.11) 0.8135.1Asian61.02 (0.89, 1.16) 0.79954.91.00 (0.75, 1.35) 0.97860.61.08 (0.95, 1.23) 0.2490.01.06 (0.92, 1.22) 0.45324.60.96 (0.82, 1.13) 0.64960.1African21.77 (0.66, 4.76) 0.25585.82.82 (0.39, 20.15) 0.30272.51.64 (0.74, 3.63) 0.22465.51.93 (0.69, 5.38) 0.20879.71.21 (0.95, 1.54) 0.13262.8Source of controlPB80.94 (0.87, 1.00) 0.06750.10.97 (0.87, 1.08) 0.52916.80.87 (0.78, 0.97) 0.016^^61.60.89 (0.80, 0.98) 0.02460.51.03 (0.95, 1.12) 0.5190.0HB51.13 (0.90, 1.52) 0.31379.31.11 (0.69, 1.77) 0.66476.81.18 (1.00, 1.39) 0.049^^18.41.20 (0.94, 1.53) 0.14363.30.96 (0.80, 1.15) 0.47574.1NOS scoresN161.02 (0.88, 1.20) 0.76178.91.00 (0.79, 1.28) 0.97759.61.06 (0.83, 1.35) 0.65881.31.07 (0.84, 1.35) 0.59382.31.02 (0.90, 1.16) 0.7787.3N270.96 (0.87, 1.06) 0.41455.60.98 (0.80, 1.20) 0.83955.40.93 (0.82, 1.03) 0.12936.10.93 (0.83, 1.04) 0.19743.00.97 (0.89, 1.07) 0.56862.3Sample sizeS171.08 (0.92, 1.26) 0.34460.11.17 (0.91, 1.49) 0.22128.61.04 (0.84, 1.31) 0.70856.11.08 (0.86, 1.36) 0.49262.41.15 (0.96, 1.37) 0.2153.3S260.93 (0.85, 1.06) 0.06167.50.91 (0.76, 1.07) 0.24963.70.91 (0.80, 1.03) 0.11571.80.90 (0.80, 1.01) 0.07069.70.99 (0.91, 1.08) 0.83062.2Cancer typeOC20.92 (0.80, 1.05) 0.19682.50.98 (0.86, 1.11) 0.69818.40.82 (0.65, 1.04) 0.10289.20.85 (0.68, 1.05) 0.12788.61.05 (0.95, 1.16) 0.3710.0BC30.96 (0.82, 1.13) 0.62474.60.95 (0.69, 1.32) 0.77471.60.96 (0.78, 1.17) 0.66166.90.96 (0.77, 1.19) 0.68373.00.97 (0.83, 1.13) 0.68151.3LC21.15 (0.92, 1.44) 0.22656.01.30 (0.81, 2.09) 0.28065.11.17 (0.94, 1.47) 0.1680.01.21 (0.95, 1.54) 0.11921.01.24 (0.95, 1.63) 0.11443.7rs217727G/ATotal301.06 (0.99, 1.14) 0.09777.61.12 (0.97, 1.30) 0.10970.21.07 (0.97, 1.17) 0.18270.21.08 (0.98, 1.19) 0.10974.41.09 (0.96, 1.24) 0.20170.1EthnicityCaucasian10.74 (0.52, 1.05) 0.0890.49 (0.13, 1.50) 0.1920.74 (0.49, 1.14) 0.1720.71 (0.47, 1.08) 0.1110.50 (0.15, 1.66) 0.257Asian291.07 (1.00, 1.15) 0.06077.61.14 (0.98, 1.31) 0.08270.61.08 (0.98, 1.18) 0.13070.51.09 (0.99, 1.20) 0.07074.51.10 (0.96, 1.025) 0.16570.6Source of controlPB91.06 (0.91, 1.24) 0.46079.21.18 (0.87, 1.06) 0.28670.60.96 (0.79, 1.17) 0.68973.91.01 (0.83, 1.24) 0.90377.41.21 (0.94, 1.57) 0.14363.9.HB211.07 (0.98, 1.16) 0.14477.91.11 (0.94, 1.31) 0.23671.51.11 (1.00, 1.23) 0.05568.71.11 (1.00, 1.24) 0.06373.71.05 (0.90, 1.23) 0.56773.0NOS scoresN1111.03 (0.89, 1.19) 0.06884.71.12 (0.85, 1.47) 0.44177.91.05 (0.90, 1.22) 0.55870.71.04 (0.88, 1.23) 0.06378.81.10 (0.94, 1.27) 0.52978.1N2191.08 (0.99, 1.17) 0.71771.41.13 (0.96, 1.34) 0.14565.41.08 (0.95, 1.22) 0.22971.31.10 (0.98, 1.24) 0.10372.51.09 (0.84, 1.41) 0.24064.7Sample sizeS1141.08 (0.90, 1.29) 0.43284.31.11 (0.76, 1.60) 0.59576.51.15 (0.92, 1.44) 0.21577.91.14 (0.91, 1.42) 0.27281.11.03 (0.68, 1.40) 0.87379.2S2161.06 (0.99, 1.13) 0.09467.81.17 (1.02, 1.33) 0.022^^61.91.02 (0.94, 1.11) 0.63657.71.05 (0.96, 1.15) 0.29065.11.14 (1.03, 1.28) 0.015^^52.0Cancer typeBLC31.01 (0.82, 1.25) 0.92356.80.80 (0.40, 1.62) 0.538641.20 (0.64, 2.23) 0.57490.11.13 (0.68, 1.88) 0.62985.90.63 (0.22, 1.80) 0.39085.1GC20.88 (0.42, 1.84) 0.73895.10.92 (0.28, 3.04) 0.89693.81.23 (0.97, 1.56) 0.0930.01.00 (0.54, 1.84) 0.99384.70.84 (0.27, 2.59) 0.75694.5BC61.13 (0.87, 1.46) 0.35189.51.33 (0.79, 2.25) 0.28483.81.02 (0.74, 1.40) 0.90885.51.09 (0.79, 1.52) 0.59488.11.33 (0.85, 2.08) 0.21180.8CC21.22 (0.78, 1.90) 0.37982.41.46 (0.60, 3.55) 0.39976.31.21 (0.80, 1.84) 0.36462.51.26 (0.76, 2.07) 0.37176.31.34 (0.66, 2.72) 0.41665.8OSCC31.31 (1.14, 1.50) 0.000^^24.81.89 (1.18, 3.00) 0.008^^57.71.27 (1.07, 1.50) 0.006^^0.01.36 (1.16, 1.60) 0.000^^0.01.67 (1.04, 2.68) 0.035^^64.6Cancer typeLC31.16 (1.06, 1.27) 0.002^^0.01.38 (1.14, 1.67) 0.001^^0.01.09 (0.95, 1.26) 0.2190.01.16 (1.01, 1.33) 0.0310.01.31 (1.03, 1.66) 0.028^^44.7HCC20.79 (0.60, 1.05) 0.10071.70.68 (0.49, 0.93) 0.017^^0.00.77 (0.44, 1.34) 0.35986.30.75 (0.47, 1.21) 0.23783.80.73 (0.54, 1.00) 0.048^^0.0rs2839698G/ATotal261.10 (1.01, 1.20) 0.039^^82.81.29 (1.09, 1.52) 0.003^^74.71.06 (0.97,1.17) 0.21568.51.11 (1.01, 1.23) 0.036^^75.41.18 (1.01, 1.39) 0.042^^76.6EthnicityCaucasian11.03 (0.78, 1.37) 0.8271.10 (0.63, 1.92) 0.7450.65 (0.40, 2.06) 0.0840.78 (0.50, 1.22) 0.2761.43 (0.90, 2.30) 0.134Asian251.10 (1.00, 1.21) 0.041^^83.51.30 (1.09, 1.54) 0.003^^75.71.07 (0.98, 1.18) 0.13868.21.12 (1.02, 1.24) 0.024^^75.91.17 (0.99, 1.39) 0.06074.4Source of controlPB51.06 (0.94, 1.21) 0.34445.61.22 (0.95, 1.57) 0.12234.40.94 (0.76,1.16) 0.56055.21.00 (1.025,1.21) 0.98351.31.28 (1.02, 1.59) 0.032^^27.9HB211.10 (0.99, 1.23) 0.07285.31.30 (1.07, 1.59) 0.009^^78.51.09 (0.98,1.21) 0.10669.51.14 (1.02, 1.28) 0.025^^77.61.16 (0.95, 1.04) 0.14280.2NOS scoresN1141.01 (0.90, 1.15) 0.83076.61.14 (0.91, 1.43) 0.25173.20.99 (0.88,1.11) 0.84360.31.02 (0.91, 1.16) 0.70466.41.02 (0.80, 1.31) 0.88181.1N2121.20 (1.08, 1.34) 0.001^^82.01.46 (1.17, 1.81) 0.001^^69.41.14 (1.00, 1.29) 0.05265.71.20 (1.05, 1.38) 0.010^^73.41.39 (1.17, 1.65) 0.000^^54.8Sample sizeS1101.07 (0.87, 1.32) 0.51086.51.33 (0.92, 1.92) 0.13479.61.11 (0.91,1.35) 0.31767.91.17 (0.95,1.45) 0.13875.31.08 (0.74, 1.57) 0.69284.5S2141.11 (1.01, 1.21) 0.030^^79.71.28 (1.07, 1.53) 0.006^^71.61.04 (0.94, 1.16) 0.42591.01.09 (0.97, 1.22) 0.13777.21.25 (1.09, 1.45) 0.002^^60.7Cancer typeBLC21.00 (0.89, 1.12) 0.9920.01.03 (0.79, 1.36) 0.8190.00.85 (0.59, 1.24) 0.40558.00.96 (0.82, 1.13) 0.5700.01.15 (0.84, 1.58) 0.37928.0GC21.33 (1.13, 1.56) 0.000^^0.01.76 (1.26, 2.46) 0.001^^0.01.07 (0.75, 1.54) 0.69952.31.27 (1.03, 1.57) 0.024^^0.01.74 (1.27, 2.40) 0.001^^0.0HCC31.15 (1.03, 1.29) 0.014^^0.01.33 (1.03, 1.72) 0.027^^0.01.12 (0.83, 1.51) 0.29973.11.17 (0.95, 1.44) 0.13645.41.27 (0.94, 1.73) 0.11734.5CRC20.98 (0.65, 1.48) 0.92790.91.01 (0.42, 2.42) 0.98190.00.96 (0.64, 1.42) 0.81779.60.95 (0.59, 1.55) 0.84287.81.06 (0.54, 2.07) 0.86985.0LC20.92 (0.81, 1.04) 0.1780.00.78 (0.54, 1.13) 0.19233.40.97 (0.78, 1.23) 0.82231.00.93 (0.78, 1.09) 0.3590.00.76 (0.47, 2.25) 0.28363.4BC40.96 (0.63, 1.45) 0.83994.31.72 (0.88, 3.34) 0.11189.31.28 (0.86, 1.90) 0.22087.31.40 (0.91, 1.11) 0.13290.70.91 (0.45, 1.83) 0.78992.7rs3741219 A/GTotal101.07 (0.88, 1.30) 0.50789.41.18 (0.94, 1.48) 0.15461.80.97 (0.71, 1.33) 0.84291.81.56 (0.79, 1.41) 0.70991.81.14 (1.01, 1.29) 0.037^^0.0EthnicityAsian101.07 (0.88, 1.30) 0.50789.41.18 (0.94, 1.48) 0.15461.80.97 (0.71, 1.33) 0.84291.81.56 (0.79, 1.41) 0.70991.81.14 (1.01, 1.29) 0.037^^0.0Source of controlPB20.94 (0.85, 1.04) 0.2530.00.96 (0.76, 1.22) 0.7300.00.90 (0.79, 1.03) 0.1260.00.91 (0.80, 1.04) 0.1490.01.00 (0.80, 1.26) 0.9780.0HB81.11 (0.86, 1.45) 0.42491.61.28 (0.96, 1.71) 0.10067.60.98 (0.63, 1.54) 0.94193.61.10 (0.74, 1.64) 0.62593.61.20 (1.04, 1.39) 0.0140.0NOS scoresN151.07 (0.74, 1.55) 0.72694.21.19 (0.79, 1.79) 0.41778.90.88 (0.46, 1.67) 0.69195.41.04 (0.60, 1.79) 0.89895.41.10 (0.94, 1.30) 0.2350.0N241.11 (1.02, 1.21) 0.015^^20.01.26 (1.03, 1.53) 0.022^^0.01.10 (0.97, 1.24) 0.1310.01.12 (1.00, 1.26) 0.042^^0.01.19 (0.98, 1.43) 0.038^^29.3Sample sizeS150.94 (0.73, 1.22) 0.11976.21.44 (0.90, 2.29) 0.12659.11.20 (0.82, 1.77) 0.34867.41.32 (0.89, 1.95) 0.16476.01.23 (0.96, 1.59) 0.10721.8S251.26 (0.94, 1.69) 0.64293.31.07 (0.83, 1.37) 0.61864.70.83 (0.53, 1.29) 0.39995.50.88 (0.59, 1.31) 0.52795.21.11 (0.97, 1.28) 0.1370.0Cancer typeBC31.20 (0.78, 1.85) 0.40587.21.51 (0.72, 3.18) 0.28178.01.21 (0.65, 2.28) 0.54687.81.27 (0.67, 2.40) 0.46289.51.14 (0.91, 1.42) 0.25823.9rs3024270 C/GTotal121.04 (0.97, 1.12) 0.26145.01.12 (1.01, 1.24) 0.025^^40.61.00 (0.91, 1.09) 0.92634.11.03 (0.95, 1.12) 0.42119.91.09 (0.95, 1.25) 0.22865.3EthnicityAsian121.04 (0.97, 1.12) 0.26145.01.12 (1.01, 1.24) 0.025^^40.61.00 (0.91, 1.09) 0.92634.11.03 (0.95, 1.12) 0.42119.91.09 (0.95, 1.25) 0.22865.3Source of controlPB21.16 (0.75, 1.80) 0.49486.51.42 (1.05, 1.93) 0.025^^84.21.01 (0.76, 1.35) 0.9360.01.15 (0.88, 1.49) 0.3090.01.30 (0.53, 3.21) 0.56892.4HB101.03 (0.98, 1.09) 0.2896.91.09 (0.98, 1.21) 0.1127.20.99 (0.91, 1.09) 0.90244.91.02 (0.94, 1.12) 0.61028.41.06 (0.95, 1.17) 0.32035.9NOS scoresN151.04 (0.96, 1.15) 0.3360.01.10 (0.95, 1.27) 0.2200.00.96 (0.84, 1.09) 0.50268.61.00 (0.88, 1.13) 0.96849.01.10 (0.98, 1.23) 0.1170.0N271.05 (0.93, 1.19) 0.45768.21.14 (1.00, 1.31) 0.05665.91.03 (0.92, 1.15) 0.6440.01.06 (0.95, 1.19) 0.2710.01.08 (0.85, 1.37) 0.55078.6Sample sizeS110.93 (0.73, 1.18) 0.5450.88 (0.54, 1.43) 0.6021.11 (0.72, 1.72) 0.6251.02 (0.68, 1.54) 0.9140.82 (0.56, 1.19) 0.295S2111.05 (0.98, 1.13) 0.20547.41.13 (1.02, 1.26) 0.01742.90.99 (0.91, 1.08) 0.84539.11.04 (0.95, 1.13) 0.42427.11.11 (0.96, 1.28) 0.14665.8Cancer typeBLC21.07 (0.96, 1.19) 0.2300.01.18 (0.94, 1.48) 0.1510.01.17 (0.95, 1.45) 0.1510.01.17 (0.96, 1.43) 0.1240.01.05 (0.89, 1.22) 0.5740.0HCC31.11 (0.84, 1.47) 0.45284.91.20 (0.97, 1.48) 0.08983.11.04 (0.86, 1.26) 0.71144.71.10 (0.92, 1.31) 0.31347.01.22 (0.73, 2.03) 0.44889.2rs3741216 A/TTotal41.66 (0.87, 3.18) 0.12795.91.16 (0.85, 1.56) 0.3480.00.91 (0.78, 1.06) 0.2360.00.95 (0.82, 1.10) 0.4710.02.42 (0.66, 8.83) 0.18195.7EthnicityAsian41.66 (0.87, 3.18) 0.12795.91.16 (0.85, 1.56) 0.3480.00.91 (0.78, 1.06) 0.2360.00.95 (0.82, 1.10) 0.4710.02.42 (0.66, 8.83) 0.18195.7Source of controlHB41.66 (0.87, 3.18) 0.12795.90.87 (0.64, 1.17) 0.3480.00.91 (0.78, 1.06) 0.2360.00.95 (0.82, 1.10) 0.4710.02.42 (0.66, 8.83) 0.18195.7NOS scoresN122.96 (0.32, 27.18) 0.3361.07 (0.65, 1.75) 0.7980.97 (0.62, 1.50) 0.8831.00 (0.67, 1.48) 0.9814.22 (0.21, 84.51) 0.34798.4N220.99 (0.87, 1.14) 0.9160.00.77 (0.52, 1.12) 0.1700.01.12 (0.95, 1.32) 0.1860.00.94 (0.80, 1.10) 0.4440.01.35 (0.92, 1.97) 0.1270.0Sample sizeS122.96 (0.32, 27.18) 0.3361.07 (0.65, 1.75) 0.7980.97 (0.62, 1.50) 0.8831.00 (0.67, 1.48) 0.9814.22 (0.21, 84.51) 0.34798.4S220.99 (0.87, 1.14) 0.9160.00.77 (0.52, 1.12) 0.1700.01.12 (0.95, 1.32) 0.1860.00.94 (0.80, 1.10) 0.4440.01.35 (0.92, 1.97) 0.1270.0Cancer typeGC21.01 (0.84, 1.21) 0.9440.01.09 (0.72, 1.64) 0.6974.70.97 (0.75, 1.24) 0.7790.00.99 (0.78, 1.25) 0.9410.01.08 (0.65, 1.70) 0.74822.9BC: breast cancer; LC: lung cancer; BLC: bladder cancer; GC: gastric cancer; CRC: colorectal cancer; PC: pancreatic cancer; OC: ovarian cancer; CC: cervical cancer; OSCC: oral squamous cell carcinoma; UCC: urothelial cell carcinoma; RCC: renal cell carcinoma. ^^P < 0.05

Fig. 2Forest plots for the association between H19 rs2107425 polymorphism and cancer risk in five models. A: allele model; B: dominant model; C: heterozygote model; D: homozygote model; E: recessive model

Correlation between rs217727 G/A polymorphism and cancer risk

Intriguingly, we obtained thirty studies about the relationship between rs217727 polymorphism and cancer risk with 14,215 patients and 20,247 controls. Overall, the rs217727 polymorphism was not significantly correlated with cancer risk (Table 2; Fig. 3). The allelic, homozygote, dominant and recessive models of rs217727 notably increased the risk of lung cancer (A vs. G: OR = 1.16, 95% CI = 1.06–1.27, P = 0.002; AA vs. GG: OR = 1.38, 95% CI = 1.14–1.67, P = 0.001; AA + GA vs. GG: OR = 1.16, 95% CI = 1.01–1.33, P = 0.031; AA vs. GG + GA: OR = 1.31, 95% CI = 1,03-1.66, P = 0.028) and oral squamous cell carcinoma (A vs. G: OR = 1.31, 95% CI = 1.14–1.50, P = 0.000; AA vs. GG: OR = 1,89, 95% CI = 1.18-3.00, P = 0.008; GA vs. GG: OR = 1.27, 95% CI = 1.07–1.50, P = 0.006; AA + GA vs. GG: OR = 1.36, 95% CI = 1.16–1.60, P = 0.000; AA vs. GG + GA: OR = 1.67, 95% CI = 1.04–2.68, P = 0.035, Table 2). Additionally, the rs217727 mutation significantly decreased the risk of hepatocellular carcinoma (GA vs. GG: OR = 0.68, 95% CI = 0.49–0.93, P = 0.017; AA vs. GG + GA: OR = 0.73, 95% CI = 0.54-1.00, P = 0.048, Table 2), suggesting that the rs217727 mutation may be an important protective factor for liver cancer, but a key risk factor for lung cancer and oral squamous cell carcinoma. The pooled results indicated that the homozygote and recessive models of rs217727 have a positive association with cancer risk in larger sample size (AA vs. GG: OR = 1.17, 95% CI = 1.02–1.33, P = 0.022; AA vs. GG + GA: OR = 1.14, 95% CI = 1.03–1.28, P = 0.015, Table 2). Heterogeneity was shown to exist in all five gene models, and results demonstrated that heterogeneity significantly decreased or disappeared in lung cancer and oral squamous cell carcinoma.

Fig. 3Forest plots for the association between H19 rs217727 polymorphism and cancer risk in five models. A: allele model; B: homozygote model; C: heterozygote model; D: dominant model; E: recessive model

Correlation between rs2839698 G/A polymorphism and cancer risk

A total of twenty-six studies with 12,413 cancer patients and 18,650 controls were included to examine the association between H19 SNP rs2839698 and cancer risk. The rs2839698 polymorphism remarkably enhanced the risk of cancer in the allelic, homozygote, dominant and recessive models (A vs. G: OR = 1.10, 95% CI = 1.01–1.20, P = 0.039; AA vs. GG: OR = 1.29, 95% CI = 1.09–1.52, P = 0.003; AA + GA vs. GG: OR = 1.18, 95% CI = 1.01–1.23, P = 0.036; AA vs. GG + GA: OR = 1.18, 95% CI = 1.01–1.39, P = 0.042, Table 2; Fig. 4). Next, stratification analyses by cancer type showed the rs2839698 mutation significantly increased the risk of gastric cancer (A vs. G: OR = 1.33, 95% CI = 1.13–1.56, P = 0.000; AA vs. GG: OR = 1.76, 95% CI = 1.26–2.46, P = 0.001; AA + GA vs. GG: OR = 1.27, 95% CI = 1.03–1.57, P = 0.024; AA vs. GG + GA: OR = 1.74, 95% CI = 1.27–2.40, P = 0.001), hepatocellular cancer (A vs. G: OR = 1.17, 95% CI = 1.03–1.34, P = 0.015; GA vs. GG: OR = 1.30, 95% CI = 1.08–1.56, P = 0.006; AA + GA vs. GG: OR = 1.29, 95% CI = 1.08–1.93, P = 0.005), renal cell carcinoma and ovarian cancer, leukemia and lymphoma (Table 2). Similarly, a positive association was detected between the allelic, homozygous, and dominant models and cancer susceptibility in the Asian descent (A vs. G: OR = 1.10, 95% CI = 1.00-1.21, P = 0.041; AA vs. GG: OR = 1.30, 95% CI = 1.09–1.54, P = 0.003; AA + GA vs. GG: OR = 1.12, 95% CI = 1.02–1.24, P = 0.024, Table 2). When stratifying by source of control, quality score and sample size, the significantly increased cancer risk was discovered in hospital-based control (AA vs. GG: OR = 1.30, 95% CI = 1.07–1.59, P = 0.009; AA + GA vs. GG: OR = 1.14, 95% CI = 1.02–1.28, P = 0.025), population-based control (AA vs. GG + GA: OR = 1.28, 95% CI = 1.02–1.59, P = 0.032) and large simple size (A vs. G: OR = 1.11, 95% CI = 1.01–1.21, P = 0.030; AA vs. GG: OR = 1.28, 95% CI = 1.07–1.53, P = 0.006; AA vs. GG + GA: OR = 1.25, 95% CI = 1.09–1.45, P = 0.002, Table 2). Heterogeneity results suggested that heterogeneity consisted in the five genetic models of overall analysis. Interestingly, we found that heterogeneity notably diminish or disappear in hepatocellular carcinoma, bladder, gastric, and lung cancer.

Fig. 4Forest plots for the association between H19 rs2839698 polymorphism and cancer risk in five models. A: allele model; B: homozygote model; C: heterozygote model; D: dominant model; E: recessive model

Correlation between rs3741219 A/G polymorphism and cancer risk

To explore the association between H19 rs3741219 polymorphism and cancer risk, we included 10 studies with 5305 patients and 6974 controls. Compared with AA + GA genotype, the GG allele of rs3741219 polymorphism was correlated with cancer susceptibility in overall analysis (AA vs. GG + GA: OR = 1.14, 95% CI = 1.01–1.29; P = 0.037, Table 2; Fig. 5). Stratified analyses indicated that the rs3741219 mutant remarkably enhanced the risk of hepatocellular carcinoma and ovarian cancer, but also decreased the risk of Glioma tumor. We next performed stratification analysis by source of control and sample size, the pooled results indicated no association between 3,741,219 polymorphism and cancer risk. Beyond that, subgroup analyses by quality score strongly showed an elevated cancer risk in higher quality score (G vs. A: OR = 1.11, 95% CI = 1.02–1.21, P = 0.015; GG vs. AA: OR = 1.26, 95% CI = 1.03–1.53, P = 0.022; GG + GA vs. AA: OR = 1.12, 95% CI = 1.00-1.26, P = 0.042; GG vs. AA + GA: OR = 1.19, 95% CI = 0.98–1.43, P = 0.038, Table 2). It manifested that heterogeneity mainly appeared in the five gene models of overall analysis and Asian population. Moreover, there was no heterogeneity existing in population-based and higher quality score.

Fig. 5Forest plots for the association between H19 rs3741219 polymorphism and cancer risk in five models. A: allele model; B: homozygote model; C: heterozygote model; D: dominant model; E: recessive model

Correlation between rs3024270 C/G polymorphism and cancer risk

Through integrating 12 potential studies embodying 5402 patients and 9159 controls, we found a significant relationship of rs3024270 polymorphism with cancer risk under homozygous model (GG vs. CC: OR = 1.12, 95% CI = 1.01–1.24, P = 0.025, Table 2; Fig. 6). The homozygous and recessive models of rs3024270 were significantly correlated with the increased risk of colorectal cancer (GG vs. CC: OR = 1.28, 95% CI = 1.01–1.61, P = 0.041; GG vs. CC + GC: OR = 1.29, 95% CI = 1.04–1.58, P = 0.019). There was no significant association between the rs3024270 polymorphism and cancer susceptibility in stratification analysis by ethnicity and quality score. We found that the rs3024270 polymorphism was positively related to cancer risk in hospital-based controls under the homozygote model (GG vs. CC: OR = 1.42, 95% CI = 1.05–1.93; P = 0.025, Table 2). Except for the recessive model (I^2^ = 65.3%, P = 0.001), there was no heterogeneity in other models.

Fig. 6Forest plots for the association between H19 rs3024270 polymorphism and cancer risk in five models. A: allele model; B: homozygote model; C: heterozygote model; D: dominant model; E: recessive model

Correlation between rs3741216 A/T polymorphism and cancer risk

In general, four eligible studies with 2049 patients and 1808 controls were included to detect the relation between rs2107425 polymorphism and cancer risk. The pooled results suggested that the rs2107425 polymorphism was not related to cancer risk in five genetic models (T vs. A: OR = 1.66, 95% CI = 0.87–3.18, P = 0.127; TT vs. AA: OR = 1.66, 95% CI = 0.85–1.56, P = 0.348; AT vs. AA: OR = 0.91, 95% CI = 0.78–1.06, P = 0.236; AT + TT vs. AA: OR = 0.95, 95% CI = 0.82–1.10, P = 0.471; TT vs. AA + AT: OR = 2.42, 95% CI = 0.66–8.83, P = 0.181, Table 2; Fig. 7). Similarly, when stratifying analyses by ethnicity, cancer type, quality score, and source of control, we did find any correlation between the rs3741216 polymorphism and cancer risk. The result of heterogeneity test exhibited I^2^ = 95.9 and 95.7, implying that heterogeneity clearly exists in both homologous and recessive models, and thus random-effects model was used to examine the correlation.

Fig. 7Forest plots for the association between H19 rs3741216 polymorphism and cancer risk in five models. A: allele model; B: homozygote model; C: heterozygote model; D: dominant model; E: recessive model

FPRP results

We investigated determinants of FPRP across a range of probabilities to determine whether a given association between H19 SNPs and cancer risk is deserving of attention or is noteworthy. In this respect, we found that our main results were further supported by FPRP analysis. As shown in Table 3, with a prior probability < 0.25, the H19 rs2839698 polymorphism was associated with the risk of cancer under allele, homozygote, dominant and recessive models. Similarly, with a prior probability of 0.25, the homozygote model of H19 rs3024270 polymorphism was associated with cancer risk and the recessive model of H19 rs3024270 polymorphis was associated with cancer risk (P < 0.2).

Table 3False-positive report probability analysis of the noteworthy resultsPrior probabilitySNPGenetic modelOR (95% CI)PPower0.250.10.010.0010.0001rs2107425Allele0.98 (0.91, 1.06)0.6141.0000.6480.8470.9840.9981.000Homozygote1.01 (0.88, 1.17)0.8941.0000.7290.8900.9890.9991.000Heterozygote0.96 (0.85, 1.07)0.4611.0000.5800.8060.9790.9981.000Dominant0.97 (0.87, 1.08)0.5781.0000.6340.8390.9830.9981.000Recessive0.98 (0.91, 1.06)0.5781.0000.6480.8470.9840.9981.000rs217727Allele1.06 (0.99, 1.14)0.1161.0000.2590.5120.9200.9910.999Homozygote1.12 (0.97, 1.30)0.1361.0000.2900.5510.9310.9330.999Heterozygote1.07 (0.97, 1.17)0.1381.0000.2920.5540.9320.9930.999Dominant1.08 (0.98, 1.19)0.1201.0000.2650.5190.9220.9920.999Recessive1.09 (0.96, 1.24)0.1911.0000.3630.6310.9500.9950.999rs2839698Allele1.10 (1.01, 1.20)0.0321.0000.087^^0.2230.7590.9690.997Homozygote1.29 (1.09, 1.52)0.0020.9640.007^^0.0210.1940.7090.961Heterozygote1.06 (0.97,1.17)0.2471.0000.4260.6900.9610.9961.000Dominant1.11 (1.01, 1.23)0.0461.0000.122^^0.2940.8210.9790.998Recessive1.18 (1.01, 1.39)0.0480.9980.125^^0.3000.8250.9790.998rs3741219Allele1.07 (0.88, 1.30)0.5001.0000.5980.8170.9800.9981.000Homozygote1.18 (0.94, 1.48)0.1520.9810.3170.5830.9390.9940.999Heterozygote0.97 (0.71, 1.33)0.8500.9990.7200.8850.9880.9991.000Dominant1.06 (0.79, 1.41)0.6890.9910.6740.8610.9860.9991.000Recessive1.14 (1.01, 1.29)0.0381.0000.102^^0.2540.7890.9740.997rs3024270Allele1.04 (0.97, 1.12)0.3001.0000.4730.7290.9670.9971.000Homozygote1.12 (1.01, 1.24)0.0291.0000.080^^0.2070.7420.9670.997Heterozygote1.00 (0.92, 1.09)0.9281.0000.7360.8930.9890.9991.000Dominant1.03 (0.95, 1.12)0.4891.0000.5950.8150.9800.9981.000Recessive1.09 (0.95, 1.25)0.2281.0000.4060.6730.9580.9961.000rs3741216Allele1.66 (0.87, 3.18)0.1260.3800.5000.7500.9710.9971.000Homozygote1.16 (0.85, 1.56)0.7730.6910.7310.8910.9890.9991.000Heterozygote0.91 (0.78, 1.06)0.2261.0000.4040.6700.9570.9961.000Dominant0.95 (0.82, 1.10)0.4931.0000.5970.8160.9800.9981.000Recessive2.42 (0.66, 8.83)0.1810.2340.6890.8740.9870.9991.000^*^P < 0.2

Sensitivity analysis and publication bias

Sensitivity analysis was conducted by eliminating each individual study. As shown in Fig. 8, the pooled OR and 95% CI were not materially changed, indicating that our results were relatively robust. After excluding several studies inconsistent with HWE, we found substantial alteration under the allele and heterozygous models in rs3741216 polymorphism (allelic: I^2^ = 0.0%, P~(heterogeneity)~ = 0.649; I^2^ = 0.0, P~(heterogeneity)~ = 0.678; I^2^ = 0.0%, P~(heterogeneity)~ = 0.737; I^2^ = 0.0%, P~(heterogeneity)~ = 0.681; I^2^ = 0.0%, P~(heterogeneity)~ = 0.708, Table 4). Other three gene polymorphisms were not substantially changed. In addition, funnel plot was symmetrical, and no evident publication bias was observed by using the Begg’s test and Egger’s test (Table 5; Fig. 9).

Table 4After excluding studies inconsistent with HWE, the associations between four H19 polymorphisms and cancer risk under five genetic modelsGenetic modelrs217727G/Ars2839698G/Ars3741219T/Crs3741216A/TAlleleOR (95%CI) P I^2^ (%) P (heterogeneity) 1.06 (1.00, 1.13)0.07169.80.0001.06 (0.98, 1.15)0.13277.10.0001.02 (0.84, 1.24)0.85089.60.0000.99 (0.87, 1.14)0.9160.00.649HomozygoteOR (95%CI) P I^2^ (%) P (heterogeneity) 1.13 (1.00, 1.28)0.05360.60.0001.21 (1.04, 1.41)0.01466.20.0001.14 (0.91, 1.43)0.26961.60.0000.31 (0.89, 1.92)0.1700.00.678HeterozygoteOR (95%CI) P I^2^ (%) P (heterogeneity) 1.04 (0.95, 1.14)0.36968.20.0001.04 (0.95, 1.13)0.39261.30.0000.96 (0.70, 1.32)0.79592.70.0000.89 (0.76, 1.06)0.1860.00.737DominantOR (95%CI) P I^2^ (%) P (heterogeneity) 1.67 (0.98, 1.17)0.15770.30.0001.07 (0.98, 1.17)0.12466.60.0001.00 (0.75, 1.35)0.97892.40.0000.94 (0.80, 1.10)0.4440.00.681RecessiveOR (95%CI) P I^2^ (%) P (heterogeneity) 1.11 (0.99, 1.24)0.07359.20.0001.12 (0.96, 1.32)0.14772.90.0001.12 (0.99, 1.27)0.0740.00.7631.35 (0.92, 1.97)0.1270.00.708

Table 5Publication bias of the five genetic models for H19 gene polymorphismsVariablesAllelicHomozygousHeterozygousDominantRecessiveP B PEP B PEP B PEP B PEP B PErs2107425C/T0.360 0.3360.583 0.4360.246 0.2860.300 0.2870.583 0.532rs217727G/A0.454 0.5150.592 0.4940.475 0.4890.354 0.4450.748 0.540rs2839698G/A0.252 0.3170.315 0.3930.338 0.3510.388 0.3630.252 0.448rs3741219A/G0.371 0.4040.371 0.2650.592 0.5710.371 0.3460.474 0.311rs3024270C/G1.000 0.8671.000 0.8750.876 0.8610.876 0.7750.533 0.791rs3741216A/T0.308 0.1701.000 0.2001.000 0.8151.000 0.8151.000 0.803P B: P-value of Begg’s rank correlation test. ^^P < 0.05. PE: P-value of Egger’s linear regression test. ^^P < 0.05

Fig. 8Sensitivity analysis for H19 rs2839698 polymorphism and cancer risk in five models. A: allele model; B: homozygote model; C: heterozygote model; D: dominant model; E: recessive model

Fig. 9Begg’s funnel plot and Egger’s linear regression plot for detecting the publication bias in rs2839698 polymorphism. (a1) Begg’s funnel plot and (b1) Egger’s linear regression plot in the allele model; (a2) Begg’s funnel plot and (b2) Egger’s linear regression plot in the homozygote model; (a3) Begg’s funnel plot and (b3) Egger’s linear regression plot in the homozygote model; (a4) Begg’s funnel plot and (b4) Egger’s linear regression plot in the heterozygote model; (a5) Begg’s funnel plot and (b5) Egger’s linear regression plot in the recessive model

Discussion

Cancer is one of the leading causes of mortality, seriously affecting public health all over the world. However, the pathogenesis of cancer remains poorly explicit. It is widely accepted that cancer may be influenced by genetic mutations [81]. As a newly identified non-coding RNAs, lncRNAs are extensively present in human genome. Numerous studies have confirmed that lncRNAs play essential roles in diverse biological activities, such as cell cycle processes, epigenetic regulation, transcriptional regulation, stress response and pluripotency maintenance [16, 18]. A large number of SNPs located in the lncRNAs may affect gene expression and function by altering its secondary structure or the targeted microRNAs, ultimately, leading to the occurrence and progression of cancer [82–84].

H19 belongs to a class of maternally expressed lncRNA at 2.3 kb length, which is located at imprinted region on chromosome 11p15.5. Differentially methylated region (DMR) usually refers to the upstream of the transcription initiation site of H19, which servers a vital role in the regulation of H19/IGF2 expression [85, 86]. It has been reported that H19 expression is prominently decreased after birth, and only exhibits in cardiac and skeletal muscles [82]. Accumulating evidence has shown that H19 gene polymorphisms are linked to cancer risk. Verhaegh et al. first reported that H19 rs2839698 variants significantly reduced the risk of bladder cancer among Caucasians, especially in non-muscle invasive bladder cancer [41]. Also, some studies have reported that the rs3741219 polymorphism was not associated with the risk of cancer, the rs2839698 polymorphism significantly increases the risk of gastrointestinal cancer, and the rs2107425 polymorphism had a protective effect on Caucasian population [81, 87, 88]. In order to accurately assess the association between H19 polymorphisms and the risk of cancer, we conducted a comprehensive analysis of all relevant potential studies.

Our findings suggested that the rs2839698, rs3741219 and rs3024270 polymorphisms, but not rs2107425, rs217727, or rs3741216 polymorphisms were associated with cancer risk in overall analysis. Among these, the rs2839698 polymorphism was dramatically related to increased cancer risk among Asians, while the rs210742 polymorphism was significantly associated with reduced cancer risk among Caucasians, indicating that ethnic differences in genetic backgrounds might influence the correlation. Using the RNA secondary structure prediction website, Gong et al. verified that the rs2107425 variant changed the minimum free energy of its centroid secondary structure and increased genetic susceptibility to cancer by impacting the H19 function and stability [51]. Further experimental functional studies are necessary to prove the exact mechanism. We found that rs2839698 SNP was positively associated with cancer susceptibility among Asians.

In the present study, the rs217727 mutation positively associated with increased risk of oral squamous cell carcinoma and lung cancer, but reduced the risk of hepatocellular carcinoma. Moreover, the rs2839698 polymorphism was significantly correlated with increased risk of gastric cancer, which was consistent with a previous study [88]. There were significant correlations between the rs2839698 polymorphism and cancer risk in hospital-based control, sample size and quality score subgroups. These results provided evidence that rs2839698 could modify cancer susceptibility based on ethnicity and cancer type. Furthermore, the discrepancy between our results and previous studies may be attributed to different genetic backgrounds and sample sizes. As for the rs3741219 polymorphism, a marginally notable correlation was discovered under recessive model in overall analysis. The positive results of higher quality score showed remarkable association with rs3024270 polymorphism. Moreover, we did not observe any relationships between rs3741216 rs3024270 and cancer in overall and subgroup analyses.

Among these H19 SNPs, rs217727, rs2839698, rs3741219, and rs3741216 located in exon region, as well as rs3024270 in intron region. SNPs at exon region are more likely to alter the H19 conserved folding structure or complementary sequences to target genes, and thus modify its binding affinity with interacting elements, while SNPs at intronic region may participate in alternative splicing and regulation of H19 transcript [86, 89]. Li et al. found that genetic variants of rs2839689, rs217727, rs2735971 and rs3024270 were closely associated with changes of H19 secondary structure in colorectal cancer [57]. It has been reported that the rs217727 polymorphism affected interactions between miRNAs and H19 and induced formation of target miRNA sites, such as hsa-miR-4804-5p and hsa-miR-8071, leading to the loss of hsa-miR-3960 and hsa-miR-8071 binding sites [73]. In addition, the rs2839698 mutation causes the loss of hsa-miR-24-1-5p and hsa-miR-24-2-5p, hsa-miR-566, and miR-675 [71, 75]. We speculated that the rs2839698 variation might hinder the binding of these targeted miRNA sites to the H19 3’-UTR, and then disrupt the reciprocal repression-regulatory loop between them, resulting in the tumorigenesis and progression.

Several limitations should be taken into account in the present study. First, heterogeneities were observed in most of the H19 SNPs, and subgroup analyses by source of control, cancer type, and ethnic diversity failed to completely eliminate these heterogeneities. Second, with regard to rs3741219, rs3024270 and rs3741216 polymorphism, all the included subjects were from Asian, and except for one study from Caucasians in rs217727 and rs2839698 polymorphism, other studies were involved with Asian population, which may not be applicable to other populations. Third, each type of cancer with only one study was assigned to subgroup analysis by other cancers, and the number of included studies for certain H19 polymorphisms was relatively limited in the subgroup analysis. Finally, due to the lack of available data on some factors such as alcohol consumption, smoking, lifestyle and effects of haplotype, we cannot evaluate the impact of gene-environmental and gene-gene interactions.

Conclusions

In conclusion, this meta-analysis demonstrated significant associations between H19 rs2839698 and rs3024270 and overall cancer risk. We found that H19 rs2107425 may be a protective factor for the Caucasian population, while rs2839698 may be a hazard factor for the Asian descent. Therefore, studies with larger sample sizes, diverse races and different cancer types are needed to further and better validate our findings.

Abbreviations: BC = breast cancer, LC = lung cancer, BLC = bladder cancer, GC = gastric cancer, CRC = colorectal cancer, PC = pancreatic cancer, OC = ovarian cancer, HCC: hepatocellular carcinoma, CC = cervical cancer, OSCC = oral squamous cell carcinoma, UCC = urothelial cell carcinoma, RCC = renal cell carcinoma, SNP = single nucleotide polymorphism, CI = confidence interval, HWE = Hardy-Weinberg equilibrium, NOS = Newcastle Ottawa Scale; OR = odds ratio.

Electronic supplementary material

Below is the link to the electronic supplementary material.

Supplementary Material 1

Supplementary Material 2

Supplementary Material 3

Supplementary Material 4