Background

Breast cancer (BC) comprises about one fourth of all female cancers worldwide. Despite new diagnostic and treatment options, roughly 30 % of early-stage patients will progress to metastatic disease [1]. Experimental genetic research and genome-wide association studies have significantly improved our understanding of complex BC biology, the process of the disease development in particular. However, it is equally important to extend our knowledge on the course the disease takes by following its development to identify patients who are likely to have a more aggressive disease and to tailor their treatment.

It has been well established that several cytokines, including Interleukin-10 (IL-10) and Tumor Necrosis Factor α (TNFα), have a crucial role in a coordinated manner in breast carcinogenesis [2]. Genes, encoding IL-10 and TNFα cytokines, contain several nucleotide variations, namely single nucleotide polymorphisms (SNPs), which are associated with different levels of gene transcription and determine interindividual differences in IL-10 and TNFα production [3, 4].

Over the recent years, three functional SNPs, constituting substitutions of a single bases upstream of the transcriptional start site of IL10 gene, have been investigated: IL10 adenine (A) to guanin (G) substitution at -1082 bp (rs1800896), IL10 thymin (T) to cytosine (C) substitution at -819 bp (rs1800871) and IL10 A to C substitution at -592 bp (rs1800872) [5]. These SNPs affect transcriptional activity, leading to alterations in gene expression that influence IL-10 production [3, 4]. They are strongly linked together and present three major haplotypes, ATA, ACC, and GCC, which are associated with low, medium and high levels of IL10 expression respectively. GCC individuals secrete on average two or three times more IL-10 than wild type ATA individuals [6]. It was proven by several authors that IL-10 levels in blood samples of breast cancer patients correlate directly with the clinical stage of the disease [7, 8].

SNP in the promoter region of the TNFα locus has been identified at position −308, which also showed that it involves the replacement of G by A [9]. TNFα -308G > A GA and AA genotypes lead to a higher rate of TNFα gene transcription than wild type GG genotype in vitro [10]. High plasma TNFα levels in cancer patients are associated with a poor disease outcome [11]. TNFα expression significantly increases at the advanced stages of breast cancer [12]. The TNFα protein induces the expression of adhesion molecules, facilitating the invasion of metastatic tumor cells [13]. Several studies have shown a close link between TNFα -308G > A polymorphism and breast cancer risk [14].

Some investigators found genetic evidence for association between IL10 -1082A > G, −819 T > C, −592A > C and TNFα -308G > A polymorphisms and breast cancer progression in different ethnic populations [8, 15]. However, the data is not consistent [5], poorly differentiated in terms of ethnicity, cancer stage, age etc. This study, therefore, aimed to investigate the relationship between functional SNPs in IL10 and TNFα and BC clinicopathologic features and survival in a highly homogeneous group of patients, taking into account age, race and stage of the disease at the time of diagnosis to identify whether these genetic determinants may be important for BC prognosis.

Materials and Methods

Patients

Adult female primary stage I-II BC patients (≤50 years old at the time of diagnosis) in premenopausal state (n = 100) were involved in this research. Women with other malignant tumors, poor performance status, other significant comorbidities and/or incomplete medical documentation were not included in the study. Adjuvant therapy was chosen by clinicians, based on pathomorphological characteristics and validated prognosis factors, according to national recommendations. All the study subjects were Eastern European (Lithuanian).

Specimen Characteristics and Assay Methods

Samples were collected in 2009–2014. Genomic DNA was extracted from peripheral blood leukocytes by using the commercially available DNA extraction kit (Thermo Fisher Scientific), with regard to the manufacturer’s protocol. A IL10 gene promoter polymorphisms analysis was performed by using a polymerase chain reaction-based restriction fragment length polymorphism method (PCR-RFLP).

IL10 gene regions including -592A > C, −819C > T and -1082A > G polymorphic sites were amplified by using primers reported by Liu et al. [16]. For -592C > A and -819C > T polymorphisms, the same reaction mixture composition was employed. Briefly, PCR reaction was carried out in a total volume of 25 μl, containing 1x DreamTaq standard buffer, template DNA, 0.24 μM of each primer, 200 μM of each dNTP and 1.25 U of DreamTaq DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA) with annealing at 63 °C and 58 °C for -592C > A and -819C > T polymorphisms respectively. PCR reaction conditions for Il10 gene -1082G > A polymorphism were slightly modified by adding 4.0 mM MgCl2, 4 % DMSO and changing the annealing temperature to 56 °C.

Following PCR, the amplicons underwent digestion with different restriction endonucleases. RsaI restriction endonuclease (Thermo Fisher Scientific Baltics, Lithuania) was used for a -592C > A polymorphism analysis. In the presence of A allele, RsaI yielded 175 and 237 bp fragments, while C allele remained uncut (412 bp). MaeIII restriction endonuclease was implemented for a -819C > T polymorphism detection. The presence of MaeIII restriction site indicated C allele (125 and 84 bp fragments), while T allele remained undigested (209 bp). For a -1082G > A polymorphism identification, PCR products were incubated with MnlI enzyme (Thermo Fisher Scientific Baltics, Lithuania), which cut G allele into 106 and 33 bp fragments, while A allele remained uncut (139 bp). The results were visualized on 2 % agarose gel containing ethidium bromide.

The primer sequences for a TNFα -308G > A fragment amplification were reported by Kaur et al. [17]. PCR reaction was carried out in a total volume of 25 μl, containing 1x DreamTaq standard buffer, template DNA, 0.24 μM of each primer, 200 μM of each dNTP, 4.0 mM MgCl2, 4 % DMSO and 1.25 U of DreamTaq DNA polymerase (Thermo Fisher Scientific, Waltham, MA, USA). The annealing temperature for TNFα -308G > A polymorphism was 63 °C.

Restriction endonuclease NcoI was used to detect the TNFα -308G > A polymorphism. With regard to TNFα -308G > A promoter polymorphism, G allele was represented by 87 bp and 20 bp fragments, while A allele by 107 bp fragment. Restriction endonuclease products were separated on 3 % agarose gels containing ethiduim bromide.

Study Design

A prospective cohort study was conducted at the Oncology Institute of Lithuanian University of Health Sciences. A full ethical approval was obtained from the Kaunas Regional Bioethics Committee (protocol number BE-2-13) and the Lithuanian Data Protection Agency (protocol number 2R-2246). Every subject has signed informed consent forms before commencing the study. For a case selection, the information of the period of 2001–2011 about primarily BC patients was retrieved from the Pathology Department at the Hospital of Lithuanian University of Health Sciences. The patients were matched by disease stage, age of disease onset and menopausal status. The patients' clinicopathological information was obtained from their medical files. The patients were monitored according to the clinical monitoring protocol till 1st May 2014. The median follow-up was 70 months. Disease progression was defined as a local breast cancer recurrence in the affected breast and distant metastases in visceral organs, skeleton, skin or the central nervous system. Date of cancer histological verification was considered as time zero for survival analysis. The SNPs selected for associations with the known breast cancer prognostic factors and cancer progression were as follows: IL10 -1082A > G, −819 T > C, −592A > C, and TNFα -308G > A. This study was conducted adhering to recommendations for tumor marker prognostic studies [18, 19].

Statistical Analysis

A Hardy–Weinberg Equilibrium for the genotype distribution of the selected SNPs was tested in all cases by using the Pearson X2 test and the Fisher Exact test. To evaluate if the frequencies of alleles and genotypes correspond with the data of earlier studies, we retrieved information from a population of the International HapMap project of Northern Europeans from Utah (CEU) (HapMap Data rel 28 Phasell + III, August10, on NCBI B36 assembly, dbSNP b126, http://hapmap.ncbi.nlm.nih.gov). IL10 haplotypes were inferred from promoter IL10 SNPs by Bayesian methods as implemented in the Phase software (version 2.1; Department of Statistics, University of Washington, Seattle, Washington, USA) [20, 21]. For demonstration of linkage disquelibrium (LD) SNP block was performed using Haploview v4.1. The block followed the haplotype block definition of solid spine of LD as implemented in Haploview v4.1 [22]. Statistical analyses were performed by using SPSS® for Windows software version 20.0 (Released 2011. Armonk, NY: IBM Corp.). P value of less than 0.05 was considered significant. Bonferroni-corrected alpha level was used in association analysis for multiple comparisons. The Pearson Chi-square or the Fisher Exact test was used for categorical data. Associations between genotype and disease-free survival (DFS), metastasis-free survival (MFS) and overall survival (OS) were investigated by using Kaplan-Meier’s method and estimated by performing a log-rank test. The association analysis included genotype, allelic models and haplotype model for IL10 SNPs. Cox regression models were used to adjust the analysis for potential confounders. SNPs were re-evaluated in a model adjusted for the known breast cancer prognostic values, which included age group (30–40 years, 41–50 years), tumor size (T1, 2), lymph node status (N0, 1), histological grade (G1, 2, 3) and intrinsic subtype (Luminal A, Luminal B, HER2 enriched, Basal-like), by carrying out a multivariate regression analysis as well as computing odds ratios and 95 % confidence intervals (95 % CI).

Results

Sample Characteristics

The analysis included 100 primary, young, premenopausal, early stage breast cancer patients. The frequency data for clinical and tumor biological factors is shown in Table 1. All the patients were genotyped for a panel of four SNPs: IL10 -1082A > G, −819 T > C, −592A > C, and TNFα -308G > A. The genotypes were found to be in Hardy-Weinberg equilibrium in all the four SNPs. A strong LD was confirmed for IL10 -819 T allele with IL10 -592A allele and IL10 -819 C allele with IL10 -592C allele (Fig. 1). Our cohort statistically has the same genotype distribution as the HapMap CEU population. The allele and genotype frequencies determined in our study and, for comparison, HapMap CEU population are shown in Table 2.

Table 1 Frequencies of clinical and tumor biological factors
Full size table
Fig. 1
figure 1

Linkage disequilibrium and haplotype block. Numerical values are given of r2 values, whereas the colors are given to encode D’ (dark grey encodes strong evidence of LD). Block followed the haplotype block definition of solid spine of LD as implemented in the Haploview v.4.1 [22]

Full size image
Table 2 Allele and genotype frequencies of the IL10 and TNFα gene promoter regions. Data from our study and HapMap CEU population
Full size table

Inferential Analysis

The estimation of associations between the known BC prognostic variables and the studied polymorphisms in genotype model revealed a significant link between IL10 -592A > C SNP and ER status (P = 0.017). The carriers of heterozygous AC genotype had 3.231 times higher probability of ER positive BC phenotype than CC genotype carriers (95 % CI 1.282 - 8.141; P = 0.011) and 4.500 times higher than AA genotype carriers (95 % CI 1.032 - 19.630; P = 0.037). The allelic model showed no close relationships of IL10 -592A > C SNP with tumor biological and clinical prognostic factors. The analysis of IL10 -1082A > G, IL10 -819 T > C and TNFα -308G > A SNPs in both genotype and allelic models showed no significant links with clinicopathological features.

Phasing revealed three main, well-known haplotypes, namely GCC, ACC and ATA. A few uncommon haplotypes were confirmed (ACA and GCA), which were not included in the association analysis. The haplotype frequency data are shown in Table 3. The haplotype analysis confirmed the ACC haplotype connection with younger age (30–40 years) of disease onset (P = 0.017). Non-carriers of ACC haplotype 2.951 times more frequently belonged to older patient subgroup (41 – 50 years) than carriers (95 % CI 1.198 – 7.273; P = 0.017). GCC and ATA haplotypes did not show any significant associations with the known breast cancer prognostic factors.

Table 3 Relative haplotype frequencies of IL10 promoter polymorphism on the total number of chromosomes
Full size table

Survival Analysis

In the median follow-up time of 70 months (range 28–157), progression was observed for 24 patients. 76 cases were censored. Of those who progressed, 20 had distant metastases. 14 patients with progressive disease died, all due to cancer related death. The data of Cox’s proportional hazards regression analysis is shown in Table 4. Kaplan-Meier and Cox's regression analysis did not reveal any significant relationships between the analyzed IL10 -1082A > G, −819 T > C, −592A > C SNPs and phased haplotypes and PFS, MFS and OS in our study. Cox’s regression analysis of TNFα -308G > A SPN has shown a significant disadvantage of GA genotype vs. two others in PFS (P = 0.020, hazard ratio (HR) = 3.049, 95 % CI = 1.195-7.778) and MFS (P = 0.045, HR = 2.819, 95 % CI = 1.021-7.780). During a further analysis of this SNP, we evaluated only the major GG genotype vs. heterozygous GA because of a small number of AA genotypes in our population. GG genotype of the TNFα -308G > A polymorphism was significantly associated with a longer PFS by carrying out the Kaplan-Meier analysis, which is graphically shown in Fig. 2 (P = 0.014). Mean PFS was 119 months in GG genotype group (95 % CI 108–129) vs. 86 months in GA genotype group (95 % CI 56–116).

Table 4 Cox’s univariate model. Unajusted hazard ratios for PFS, MFS, OS with each of the SNPs in genotype, allelic and haplotype model
Full size table
Fig. 2
figure 2

Kaplan–Meier curves for progression-free survival of TNFα -308G > A polymorphism GG and GA genotypes

Full size image

As far as MFS is concerned, the benefit of GG genotype vs. GA was also demonstrated by Kaplan-Meier curves (P = 0.037, Fig. 3). The mean time of MFS was 122 months in GG genotype group (95 % CI 112–132) vs. 93,7 months in GA genotype group (95 % CI 64–124). The period of follow-up is rather short to evaluate OS differences, however, preliminary data also shows unequal survival between GG and GA genotypes of TNFα -308G > A SNP (P = 0.036) (Fig. 4).

Fig. 3
figure 3

Kaplan–Meier curves for metastasis-free survival of TNFα -308G > A polymorphism GG and GA genotypes

Full size image
Fig. 4
figure 4

Kaplan–Meier curves for overall survival of TNFα -308G > A polymorphism GG and GA genotypes

Full size image

After adjusting to age group, tumor size, histological grade, lymph node status, ER, PR, HER2 status and intrinsic subtype, TNFα GA genotype of TNFα -308G > A SNP remained a significant negative prognostic factor for PFS (P = 0.005, HR = 4.631, 95 % CI = 1.587-13.512), MFS (P = 0.010, HR = 4.708, 95 % CI =1.445 – 15.345) and OS (P = 0.037, HR = 4.829, 95 % CI =1.098 – 21.243), which is shown in Table 5.

Table 5 Cox’s multivariable model. Adjusted hazard ratios for PFS, MFS, OS with each of the known BC prognostic factor and TNFα -308G > A
Full size table

Discussion

In this prospective cohort study of 100 premenopausal female patients with early-stage breast cancer, we investigated associations between functional SNPs in IL10 and TNFα genes, previously implicated in breast cancer occurrence, spread and survival. We found that the SNP genotype frequency data of IL10 -1082A > G, −819 T > C, −592A > C and TNFα -308G > A correspond to HAPMAP project CEU population data and obey the Hardy-Weinberg law of genetic equilibrium.

IL10 -1082A > G polymorphism did not show any significant correlation with tumor characteristics, lymph node status and the course of the disease. In the Asian population, Kong et al. showed a larger tumor size for those with AA genotype at position −1082 in comparison to other genotypes and a significantly lower lymph node involvement in patients harboring at least one G allele of this SNP [15]. However, supporting our results, none of the reported European studies showed this SNP to be associated with tumor phenotype or survival [8, 23–26]. Despite the fact that in earlier studies the −1082 G allele (which had also been related to higher IL10 expression [10]) was associated with a lower breast cancer risk [27], it seems not to have a major impact on a further course of the disease in our study.

Carriers of IL10 -592A > C heterozygote AC genotype and IL10 -819 T > C CT genotype had a higher probability of ER positive BC type than homozygote variants. Our data conflict with other authors who did not find any associations of these SNPs with ER status [15, 23, 28]. Furthermore, in the Chinese population, Jingyan et al. [29] did not reveal any significant locus–locus interaction between ER coding genes and IL10 -1082, IL10 -819, or IL10 -592 SNPs, which could explain associations of these SNPs with ER status. However, there is lack of data on this topic in the European population in literature.

Our results of the IL10 -819 T > C and -592A > C SNP association analysis with other known BC prognostic factors and survival confirm a few other authors’ findings, i. e. those SNPs are neither related with clinicopathological tumor data (except ER status as mentioned earlier) nor with PFS, MFS or OS [15, 23, 25, 30]. However, our data contradict the study of Slattery et al. [31], who have recently showed the IL10 -819 TT genotype as a potential factor for lower cancer risk with OR of 0.79 and Gerger et al. [8], who revealed A-allele of the IL10 -592C > A polymorphism to have a prognostic value of the reduced DFS with 1.45 risk ratio; yet, controversially, this allele was earlier proved to be linked with a lower BC risk [28].

Due to strong linkage disequilibrium between IL10 -819 T > C and -592C > A SNPs, the presence of ATA haplotype could be determined by analyzing the -592C > A polymorphism: the -592A allele indicated the presence of the ATA haplotype, whereas the -592C allele indicated its absence. Phasing revealed three main, well-known haplotypes, namely GCC (41 %), ACC (32.8 %) and ATA (26.2 %). An association between ACC haplotype and younger age of disease onset was found. In the Asian population, as earlier reported [15], the authors discovered ATA haplotype to be associated with a significantly increased risk of lymph node metastasis and a higher tumor size at the time of diagnosis. We did not reproduce these results in the Lithuanian population. ATA haplotype in our study did not show any distinction from other haplotypes in the association and survival analysis. The literature on survival differences among breast cancer patients with different IL10 haplotypes is extremely poor. Data from one small Iranian study support our results [32].

Functional IL10 polymorphisms are of particular interest when describing BC because IL-10 has both potentially cancer-promoting immunosuppressive and potentially cancer-inhibiting antiangiogenic properties. Despite the fact that Langsenlehner et al. [28] revealed that genetically programmed low IL10 expression may be protective in susceptibility to breast cancer, according to our data it seems to have no importance to a further development of the disease.

TNFα -308G > A SNP has showed the greatest prognostic potential for BC of all the analyzed SNPs. GA genotype (earlier reported as a high plasma TNF producer) in BC patients was found to be significantly associated with a poor disease outcome, while wild GG genotype, usually linked to low plasma TNF levels, was associated with a better prognosis. The multivariate regression model indicated TNFα -308G > A SNP as an independent prognostic factor for PFS, MFS and OS. As a biological background for these results may serve the fact, that TNFα protein induces an epithelial-mesenchymal transition, namely the process through which cancer cells at the invasive front of primary tumors undergo a phenotypic conversion to invade and metastasize through the circulation and generate a metastatic lesion at distant tissues or organs [33]. A chronic and consistent presence of TNFα in tumors leads to procancerous consequences in many malignant diseases [34]. TNFα is overexpressed in approximately 90 % of patients with recurrent disease [12]. Similarly, Mestiri et al. discovered that the low producer TNFα -308G > A AA genotype was often associated with the reduced DFS and/or overall survival in patients with breast cancer [35]. Azmy et al. revealed that the carriage of low producer -308A allele might predispose to a more aggressive disease [36]. A study in Tunisia concluded that individuals with the AA genotype were more susceptible to and had worse prognoses in BC [32]. An Italian study did not demonstrate any association between TNFα -308G > A polymorphism genotypes and BC [27]. Murray et al. [25] failed to confirm TNF alpha polymorphisms as a potential indicator for time to recurrence in Caucasians, African Americans and Hispanics. Controversially, a meta-analysis of Caucasian and Asian ethnicities reported by Fang et al. [14] suggested that the G allele of TNFα -308G > A is a risk factor for breast cancer development, especially for Caucasians. A contrasting nature of the results of all these studies may be accounted for by sampling error or by differences in ethnicity of patient groups.

We take into consideration a limited sample size, the risk of other confounders and nonrandom sampling. However, this study supports the relevance of TNFα germline polymorphisms to BC prognosis and our findings hold promise for further investigations, preferable on larger cohorts from different ethnic origins.

Conclusions

In conclusion, our findings suggest that IL10 -1082A > G, −819 T > C, −592A > C SNPs have no sufficient data of association with the prognosis of BC. Contrary, the TNFα -308 polymorphism might modulate the risk and could contribute to the identification of patients at a higher risk of BC recurrence, metastasis and overall survival in Lithuanian early-stage breast cancer patients. To confirm the validity and utility of these polymorphisms as clinical prognostic biomarkers, future studies of a wider European population are needed.