This study was designed to establish a relationship between common SNPs in 3’UTR of ADME genes and clinical parameters as putative risk factors of myelosuppression, gastrointestinal toxicity, nephrotoxicity and hepatotoxicity during breast cancer chemotherapy. We explored ADME genes involved in pharmacokinetic of doxorubicin, 5’FU and cyclophosphamide. Our results indicate that the risk of adverse side effects and toxicity was correlated with numerous genetic and clinical changes. In the following part of the discussion, we describe the results of this study divided into sections based on the functions of the analyzed genes and clinical parameters generating adverse effects of therapy.
We report the significant correlation of AKR1C3 rs3209896 AG with the risk of early nausea and bone narrow suppression, observed in our study as overall and early anemia. Family of AKR1C are enzymes involved in progesterone metabolism pathways [39]. Our results are consistent with previous studies which suggested the role of AKR1C3 in regulation of proliferation and differentiation of bone myeloid cells. Researches indicate that reduced expression of AKR1C3 prevents the proliferation of human myeloid leukemia cells [40]. Genetic variants of AKR1C3 gene were correlated with risk of lung and prostate cancer, [41], lymphoma [42], bladder cancer [43], B cell non-Hodgkin lymphoma and breast cancer [44]. To the date, there are no data regarding the positive impact of rs3209896 variant on cancer risk or treatment outcome. In the study of Liu et al. [42] it did not associated with the risk of childhood leukemia, similarly the group of Voon reports lack of correlation between AKR1C3 rs3209896 genotype AG, survival time (PFS and OS) and chemotoxicity in breast cancer patients treated with doxorubicin-containing chemotherapy [45].
In our group of patients, the presence of ALDH5A1 rs1054899 genotype AC increased the risk of early and recurrent severe neutropenia. Moreover, carriers of AA genotype had increased risk of recurrent nausea inducted by FAC chemotherapy. ALDH5A1 is involved in cyclophosphamide metabolic pathways [46], and its expression is a predictive factor of poor prognosis in early and advanced ovarian cancer [47]. The clinical importance of ALDH5A1 rs1054899 has been confirmed in pharmacokinetic studies in Chinese patients with epilepsy treated with valproic acid (VPA) with anticonvulsant properties [48].
We report that SNPs in UGT2 genes influenced risk of myelotoxicity (rs1131878) and gastrointestinal symptoms (rs3100). The UGT2 belongs to phase II enzymes, playing important roles in toxicology and pharmacology, participating in clinically important drug-xenobiotic and drug-drug interactions. Dysfunction and/or changed expression of UGT is thought to contribute to interindividual differences in drug disposition, drug response, detoxification, homeostasis as well as to the risk of certain diseases [49]. Reported in the current study UGT2B4 rs1131878 common homozygote AA was the independent predictor of recurrent anemia while the presence of heterozygote AG was the independent predictor of recurrent neutropenia, including severe symptoms. In literature, the presence of UGT2B4 rs1131878 heterozygote in comparison to both homozygotes increased the pancreatic cancer risk [50]. In pharmacogenomic and clinical studies the UGT2B4 rs1131878 heterozygote showed considerable but not significant associations with metabolic ratio of risperidone in autism disorder patients [51]. Also, the pharmacogenetic study of Scherer’s group presents the correlations of UGT2B4 rs1131878 heterozygotes with lower risk of colorectal cancer in users of nonsteroidal anti-inflammatory drugs, than non-users with only major alleles [52]. In our group of women, UGT2B15 rs3100 reference CC homozygote increased risk of gastrointestinal toxicity manifested by severe nausea. The rs3100 was previously reported to increase the risk high-grade prostate cancer in African American men [53]. It is known, that this variant is probably located within a miRNA binding site and therefore is crucial to expression regulation of UGT2B15 gene. The exact mechanism of such influence is not yet elucidated, but the literature data suggest, that several miRNAs (including miR-376c-3p), or perhaps indirectly another mechanism, regulate influence of rs3100 on UGT2B15 expression [54].
In our study DPYD rs291593 the presence of minor T allele was an independent risk factor of leukopenia. Moreover, the CT heterozygote was risk factor of recurrent nausea, while nephrotoxicity was linked to the presence of both homozygotes. Rare variant GG of the second studied DPYD SNP rs291592 was an unfavourable risk factor of early and recurrent leukopenia, allele G increased risk of neutropenia and early nausea. DPYD is an enzyme of 5-FU metabolism pathway [55]. In literature, SNPs within DPYD are reported as cause of enzyme deficiency resulting in 5-FU therapy toxicity. Lowered DPYD activity contribute to interindividual differences in 5-FU therapies for efficacy and toxicity [56, 57, 58]. To the date the variants rs291592 and rs291593 was not established as risk factors of chemotherapy toxicities. Conversely, the group of Etienne-Grimaldi did not confirm associations with digestive and/or neurologic and/or hematotoxicities and risk of severe toxicity under capecitabine in advanced breast cancer patients [59]. Although, the clinical significance of rs291593 has been shown by Garcia-Gonzalez et al, who reported worse disease-free survival of colorectal cancer patients carrying the GG homozygote, when compared to other genotypes [60].
In our study, CYP1B1 rs162562 common genotype AA modulated gastrointestinal toxicity and early nausea risk. Cytochrome P450 1B1 (CYP1B1) is a monooxygenase involved phase I metabolism and is expressed in a variety of tumor tissues [61, 62]. CYP1B1 plays the main role in the metabolism of estrogen and androgens substrates and wide variety of xenobiotics. This enzyme catalyses 4-hydroxyl-estrogens, which is a key reaction to hormonal carcinogenesis [63]. The are no reports describing the significant connection of rs162562 with clinical outcome of patients, while the data from disease risk assessing case-control studies are contradictory. The rs162562 polymorphism was studied by a group of Gu et al., but they did not confirm associations with prostate cancer risk [64]. In study of Burdon however, rs162562 together with rs10916 were associated under a dominant model in the severe cases of primary open-angle glaucoma in a Caucasian population. These observations confirmed that common variation are cause a common disease [65].
We observed correlation of GSTM3 rs3814309 rare CC genotype with an increased risk of early neutropenia. GSTM3 belongs to a family of phase II metabolizing enzymes that catalyze the conjugation of glutathione (GSH) to a wide variety of xenobiotic [63]. It plays an important role in antioxidant, detoxification and clearance function of reactive oxygen species in tumor tissues [66]. GSTM3 polymorphisms are seen as risk factors of age-related and oxidative stress-related diseases [48, 67] Also, polymorphic variants of GSTs genes family are associated with an increased risk of cancer, including breast carcinoma [68]. The functional SNP GSTM3 rs3814309 in the 3’UTR region is putatively associated with a few miRNAs interactions [69, 70]. Moreover, role of GSTM3 rs3814309 was described as the switch of GSTM1 and GSTM4, which works by negative control of methylated sites [71]. Unfortunately, despite of the known function of GSTM3 rs3814309, there are no data describing its exact clinical impact, whether on the treatment efficiency or cancer risk [70].
In our study NR1/2 rs3732359 allele G was associated with hepatotoxicity and recurrent vomiting. The transcription factor NR1/2 is extremely important sensor of endo- and exogenous chemicals, which activates variety of detoxification enzymes and pathways [72]. It should be noted, that NR1/2 is the main mediator of expression of CYP3A4, the major metabolizer of drugs [73]. In literature, NR1/2 rs3732359 and rs3732360 were associated with higher CYP3A4 activity in vivo [74]. Polymorphisms in 3’UTR of NR1/2 was also shown to increase resistance to chemotherapy in breast cancer patients, and the NR1/2 rs3732359 role was described in detoxification mechanisms of xenobiotics and in influence of on bone marrow hematopoietic capacity [75–78]. In the study of Zeng, NR1/2 rs3732359, together with several other NR1/2 SNPs, significantly affected the voriconazole concentrations in patients with hematological malignancies and hematopoietic stem cell transplantation [79]. The group of Ren classified NR1/2 rs3732359 with SLC15A1 rs2297322 and FMO3 rs2266782 as multiple novel predictive biomarkers of docetaxel-induced myelosuppression specific to Chinese Han patients [80]. Another study underlines the importance of NR1/2 rs3732359 with platelets and/or absolute neutrophil count (ANC) from baseline in cycle 1, a significant reduction in nadir haemoglobin, either dependent or independent of the effects on the pharmacokinetics of docetaxel in nasopharyngeal cancer patients [81]. Also, variants NR1/2 rs3732359 and rs3732360 exhibited higher median oral midazolam clearance compared with homozygous reference genotypes for these SNPs [74, 81]. In patients with SULT4A1 rs138057 allele G we observed higher risk of recurrent nausea and severe, early severe vomiting and recurrent nausea. The carriers of NOS3 rs2566508 genotypes TT/GG correlated with the observed higher risk of early and recurrent nausea. In our knowledge this is first study reported associations of 3’UTR polymorphism in doxorubicin metabolizing enzyme SULT4A1 [82]. Function of this protein product is metabolism of endogenous chemicals, neurotransmitters, drugs and xenobiotics. Expression is known in cerebellum, hypothalamus and cortex [83, 84]. Another identified in our study gene increasing the risk of early and recurrent nausea was NOS3. In breast cancer tumours it was positively associated with ER and PR status. SNPs in NOS3 in breast cancer carriers shoved associations with invasive cancer and poor prognosis [85].
In this study, we showed a correlation of genetic variants of genes AKR1C3, ALDH5A1, UGT2B4, UGT2B15, DPYD, CYP1B1, NR1/2, SULT4A1 and NOS3 with bone narrow suppression, gastrointestinal toxicity, hepatotoxicity and nephrotoxicity. Identified genes control cell proliferation and differentiation, influence hormone regulation and play a superior role in relation to many genes.
In the current study, ERCC1 rs3212986 genotype GT and rs1046282 genotype TC and CC determined a higher risk of hematologic toxicities. Moreover, ERCC4 rs2276464 allele C increased the risk of recurrent nausea, while ERCC4 rs4781563 genotype AA modulates risk of early nephrotoxicity. The ERCC1 rs3212986 genotype GT is crucial in drug-induced damage repair. In literature, ERCC1 rs3212986 is also described in associations with treatment response, cancer risk and overall survival [86–89] and with poor response to chemotherapy and shorter survival time of advanced NSCLC [90]. Moreover, this SNP is associated with the development of breast cancer in Chinese population [91], and influences the response to chemotherapy and clinical outcome of gastric cancer [92]. Soares’ group noted that ERCC1 rs3212986 is risk factor of late gastrointestinal toxicity in cervical cancer patients. Recessive rs3212986 genotype AA presented a four-fold increased risk of developing late gastrointestinal toxicity compared to C allele carriers [93]. Goekkurts’ team noted that grade 3–4 anemia, leukopenia and neutropenia was associated with polymorphisms within ERCC1 gene in gastroesophageal adenocarcinoma [94]. The pathogenicity of this variant may be due to changes in mRNA stability and dysfunction of the DNA repair process [93, 95]. ERCC1 polymorphism rs1046282 CC identified in our study as risk factor of early neutropenia in literature indicated over two-fold lower risk of hepatocellular carcinoma, and the occurrence of rs1046282 TC genotype could reduce the DNA load of hepatitis B virus and might affect the replication of HBV DNA in liver cancer patients [96].
Our results show, that two polymorphisms in ERCC4 gene were responsible for elevated risk of recurrent nausea (rs2276464) and early nephrotoxicity (rs4781563). To date these results are not backed up by the literature data, in fact the lack of correlation of rs2276464 variant with gastric, melanoma and prostate cancer patients was confirmed by a few collaborations [66, 97, 98].
Our study indicates that hematologic, gastrointestinal and kidney-related toxicity symptoms could be inducted by SNPs-related decrease of ERCC1 and ERCC4 expression. Polymorphisms located in the 3’UTR of ERCC1 and ERCC4 could modulate the major regulatory region of gene and influence posttranscriptional modification of mRNA, which in turn affects the stability and function of mRNA, the ability of DNA repair is reduced, and cancer cells become prone to proliferation [96].
In this study, we evaluated several polymorphisms in selected membrane transporter genes as possible risk factors of treatment-induced toxicity. Strong influence was shown for variants in four genes of ATP-binding cassette (ABC) superfamily (ABCA1, ABCB1, ABCC1, ABCC5). These carriers are known for their impact on drug resistance [99].
We found a higher risk of recurrent leukopenia in carriers of ABCA1 rs4149339 homozygotes CC and TT. Although the impact of this polymorphism on sensitivity to cytotoxic drugs has not been reported yet, the results of Zhao’s group indicate the functionality of said SNP. They reported elevated risk of coronary artery disease with dyslipidemia in Chinese population for the carriers of the ABCA1 rs4149339 genotype CC [100]. In our study, two common polymorphism in ABCC1 gene influenced FAC toxicity. The common homozygote GG of the rs129081 variant increased risk of hematological related symptoms: early and recurrent anemia and overall leukopenia (together with heterozygote GC). Correlation between ABCC1 rs129081 and drugs’ pharmacokinetics has not yet been reported, as shown Steeghs and others in study of telatinib in patients with solid tumors [101].
We reported a higher risk of recurrent neutropenia in carriers of ABCC1 rs212091 genotype AG. Moreover, ABCC1 rs212091 GG conditioned recurrent severe vomiting. Cao and colleagues reported that ABCC1 rs212091 alleles GG, AG vs AA had a lower risk of developing myelosuppression in Chinese acute myeloid leukemia patients. The functional significance of ABCC1 rs212091 AG is correlated with creating a putative miRNA binding site for has-miR-1303 [102, 103]. This polymorphic variant may result in adverse events during treatment or potential influence microsatellite instability in tumorigenesis but functional studies are required [104]. Moreover, rs212091 is associated with virological failure in antiretroviral drugs therapy [105]. Polymorphism rs212091 AG was also associated with the protective function of G allele and better prostate cancer survival. Literature data suggest, that rs212091 together with rs35605 regulates RNA splicing of ABCC1 and leads to the opposing effects on expression and function [106].
The results of our study showed a significant increased risk of recurrent severe and early neutropenia in carriers of ABCB1 rs17064 genotype AT. At the same time, the risk of recurrent and severe vomiting correlated with the presence of minor T allele. Margier et al., described associations ABCB1 rs17064 with decreased fasting vitamin D status in a healthy male. This group speculated, that rs17064 affects ABCB1 expression, which may lead to altered vitamin D transport and status [107]. This SNP was examined in a few studies without associations with essential hypertension [108], depression and antidepression-response in Mexican Americans [109], ovarian outcome [110], and kidney allograft failure [111].
Our results showed, that ABCC5 rs3805114 reference allele AA was the only genetic factor independently associated with overall vomiting, accompanied with premenopausal age of patients who received FAC treatment. Also, the rs3805114 heterozygote AC was responsible for elevated risk of early nephrotoxicity. This SNP had been also associated with ratio of CSF-to-plasma raltegravir concentration and exposure in healthy, HIV-negative adults of European descent [112]. However, Yuan et al., reported no relationship between this genotype and response in patients with chronic hepatitis B patients treated with entecavir [113].
The results of our study showed a higher risk of early anemia and early vomiting in carriers of RALPB1 rs12680 allele C. The RALBP1 belongs to xenobiotic-transporting ATPase family [114] It is an anti-apoptotic protein that mediates multidrug-resistance. RALBP1 is the major transporter of doxorubicin in lung cancer cells. In breast cancer cell lines doxorubicin transport activity may be related to lower RALBP1 transporter activity [115]. In our knowledge, it is the first report about SNPs in RALPB1 gene and chemotherapy outcome in breast cancer patients.
Multiagent chemotherapy has a various range of side effects including anemia [116]. These toxic myelosuppression manifestations increase risk of fatigue, malaise and infection. [117]. Bone narrow suppression is the dose-limiting factor because of potentially life-threatening neutropenia, leukopenia and thrombocytopenia complications, i.e. the severe infections and bleeding [118]. Moreover, haematological toxicity and side effects lead to limited practical application of drug and treatment delays. FAC chemotherapy regimen induced grade I or II anemia in 55% of patients and grade II or IV in 11% of previously treated patients with metastatic breast cancer [119]. We reported an increased risk of overall anemia in breast cancer patients who underwent above six cycles of FAC regimen. Concordant reports come from the study by Pourali et al., where anemia was observed in 38.2% of patients who underwent 6 cycles of chemotherapy and 50.9% in patients treated with 8 cycles [116]. In addition, group of Hassan et al. observed that prevalence of anemia among solid tumor patients was increased after each chemotherapy cycle (after the 1st cycle anemia was observed in 0.5% patients, 4th cycles of chemotherapy inducted anemia in 43.6% patients) [120]. Furthermore, the danger of developing systemic toxicity during prolonged cancer chemotherapy is universally accepted, despite the improvement of outcomes and lowering the recurrence rate for longer cycles. Such dependence has been shown by group of Mayama, where the risk of haematological toxicities in endometrial cancer patients decreased with fewer treatment cycles [121].
Vomiting and nausea are some of the most common side effects reported during chemotherapy treatments. Chemotherapy-induced nausea and vomiting (CINV) is a common adverse effect that impacts treatment outcomes, quality of life, nutrition, decreased ability to work and treatment regimens [122, 123]. Furthermore, reported in literature risk factors of nausea and vomiting included chemotherapy type, female sex, history of morning or motion sickness, pregnancy-related nausea and vomiting and history of low alcohol use [122–127]. In previous reports, the young age of (< 50 years) cancer patients was one of the general risk factors for developing CINV [122–126]. In the present study, premenopausal age was a strong independent risk factor of nausea and vomiting. Similar observations came from the study of Nawa-Nishigaki et al., who observed that breast cancer patients younger than 55-year-old treated anthracycline/cyclophosphamide regimen were a significant risk for both nausea and vomiting [128]. These data agree with the results of other studies conducted in group of pre-menopausal age women [129].
In our group of patients, advanced breast cancer women were at higher risk of severe recurrent vomiting incidents. Tumor factor of TNM classification is crucial to treatment response [130]. In our report, the higher risk of gastrointestinal toxicity correlated with tumor size over 20 mm in the greatest dimension (i.e. T component greater than T1) although the correlation of tumor size with recurrent severe vomiting is difficult to explain.
We reported, that presence of metastases correlate with higher risk of hepatotoxicity. The presence of metastases is a key parameter of a poor prognosis and main cause of mortality from breast cancer [131–133]. In our group the liver was the fourth location of metastases, found in 8 patients, therefore the location itself cannot be responsible for the hepatic toxicity. One could assume, that elevated liver enzymes levels in patients with advanced disease may reflect the activity of chemotherapy on multiple metastatic locations in the body.
We report, that PR expression was an unfavorable factor correlated with recurrent neutropenia in breast cancer patients. Moreover, negative ER expression correlated with nephrotoxicity. The status of predictive and prognostic factors ER, PR, HER2 and grade are the main prognostic factors in breast cancer patients. Progesterone receptor signaling pathways are strongly correlated with estrogen receptor signaling pathway. The lack of progesterone receptor expression was reported as the factor for poor treatment outcome in breast cancer [134, 135]. Lack of PR expression leads to estrogen receptor activation and the activation of pathways and cell proliferation [136, 137]. ER, PR positivity and HER2 gene negativity are associated with better prognosis in breast cancer patients. In turn, ER and PR negativity and HER2 positivity correlated with poor outcome and more advanced disease [138]. These factors determine prognosis of disease and the choice of optimal therapy. In our study, overexpression of HER2 was associated with neutropenia. Prognostic and predictive factor HER2 is overexpressed in 10–25% of breast cancers and is associated with a more aggressive form of breast cancer and poor outcome [139]. Neutropenia induced by chemotherapy is one of the basic problems in the treatment of cancer patients as it can lead to infections and increase the risk of discontinuing therapy.
In this study, for 24 toxicities the statistically significant cumulative models were constructed, where the risk of adverse symptoms progressed with the growing number of risk factors. It should be noted, that each model consist of factors belonging to at least two different functional groups, i.e. drug metabolizers, transporters, DNA repair machinery and patients’ clinical characteristics. Thus, the elevated toxicity risk seen for the carriers of maximum number of high-risk variants, is the reflection of changes in network of mechanisms crucial for FAC-drugs’ distribution and activity. In our results it is also evident, that the clinical factors’ impact on toxicity risk do not prevail over the genetic ones, as could be expected. The clinical determinants are the components of over half of constructed cumulative models (14 of 24), predominantly for gastrointestinal and kidney- and liver-related toxicities. The strongest clinical influence were seen in models of overall hepato- and nephrotoxicity, where apart from the genetic factors, there were two clinical determinants, that added to the symptoms’ risk. At the highest nephrotoxicity risk were the postmenopausal carriers of the variants in steroid hormones and 5FU metabolizers (AKR1C3 and DPYD), with negative tumour ER expression. Even stronger dependence was detected for overall hepatotoxicity, where the only genetic factor (detox controller NR1/2) was accompanied with patients’ postmenopausal age and pre-existed metastases. This could be in part the reflection of the very group characteristics, as the 32% of metastases were located in the liver (Table 2). The most complex model was constructed for recurrent nausea, where the presence of variants in four metabolizer genes (SULT4A1, DPYD, NOS3, ALDH5A1), together with the DNA repair (ERCC4) was responsible for high symptom’s risk. This result indicates the importance of those systems for the developing of prolonged nausea. Surprisingly though, the cumulative model of recurrent nausea was the only one out of GI toxicities without the age-related compound. Similarly complex, but in regard to the presence of factors from all four functional groups, was the model of early nephrotoxicity risk.
The results of analyses focused on the accumulation of high-risk factors on FAC toxicity symptoms confirmed the universally accepted multifactorial nature of those events. It is crucial to emphasize, that in several models (early and recurrent anemia, recurrent nausea, early nephrotoxicity), there were any symptom’s cases for the non-carriers of high risk factors. At the same time, for models of early neutropenia and severe and severe recurrent vomiting, all patients carrying the maximum number of high risk factors developed toxic symptoms. Those observations indicate, that the patients who share the same diagnosis and treatment regime, could be divided into groups, which differ in expected oversensitivity to treatment. Such tool for patients’ selection could be useful in treatment tailoring for improvement its tolerance and effectiveness. Although the current FAC chemotherapy protocols are not flexible in terms of dose reduction, but there are many premedication scenarios that could be implemented for patients at risk of treatment-related toxicity.