Discussion
HCC patients treated with sorafenib show a highly variable response, and patients experience resistance and adverse events in approximately 30% of cases. The molecular mechanisms underlying inter-individual variability in sorafenib response have yet to be fully elucidated, and a deeper knowledge of the underlying mechanisms and associated gene variants would allow us to tailor treatment prescriptions for better outcomes.
Thus, with this aim, we studied 5 known candidate SNPs in genes controlling tumor angiogenesis, VEGF-A (rs2010963), VEGF-C(rs4604006), HIF-1α (rs12434438), ANGPT2 (rs55633437), andNOS3 (rs2070744), and genotyped a subgroup of 23 HCC patients using the DMET plus platform to identify potential prognostic biomarkers correlated to HCC patient responses to sorafenib treatment. Moreover, in a subgroup of patients (n = 11), the serum/plasma concentration of sorafenib was determined. The average steady-state sorafenib concentration was 4.13 mg/L in responders, and 4.41 mg/L in non-responders, independent of sorafenib dosages. No significant correlation was detected between sorafenib concentration and clinical outcome (data not shown), as already reported in the literature.35
In our study, the analysis showed that the allele and genotype frequencies of SNPs in angiogenesis-related genes were significantly correlated with the response to sorafenib only for the rs2010963 C allele (p = 0.004) and the CC genotype (p = 0.046) of theVEGF-A gene, in accordance with the results of the retrospective multicenter study ALICE-2, in which the rs2010963 C allele and CC/CG genotype were significantly associated with a higher median overall survival of HCC patients receiving sorafenib.10 It is likely that VEGF-A -related genetic variants could influence the level of circulating VEGF,36,37 therefore affecting sorafenib response. Also, in the SHARP trial,38 it was reported that a low VEGF-A plasma baseline level, as a prognostic independent factor, can predict outcomes in patients with advanced HCC, both in the entire patient population and in the placebo cohort.37 Consistent with these results, in our GEO analysis, patients expressing lower levels of VEGF-A mRNA showed a better response to sorafenib therapy (GSE109211 dataset). Our findings confirmed the prominent role of the rs2010963 gene variant andVEGF-A expression as significant predictive factors for sorafenib response in HCC patients.10,39
DMET genotyping showed a statistically significant association of the sorafenib “non-responder” phenotype with the heterozygous genotypes ofSLC22A14 rs171248, rs149738, and rs183574, and the homozygous genotypes AA in SULT1A2/CCDC101 rs11401, DPYD rs2297595,FMO2 rs2020863, the TT in DPYD*9 rs1801265, the GG inADH6 rs10008281, and CYP26A1 rs7905939. Instead, the sorafenib “responder” phenotype was associated with the genotypes CC in ADH1A rs6811453, the AG in SULT1A2/CCDC101 rs11401,DPYD rs2297595, FMO2 rs2020863, and CT in DPYD*9rs1801265, as well as with the homozygous genotypes of SLC22A14rs171248 (TT), rs149738 (AA) and rs183574 (AA). These results demonstrate that the ADME genotype is correlated with different responses to sorafenib, underlying the role of the reference allele or variant in the effect on treatment response.
Moreover, to verify whether a correlation of SNPs in angiogenesis- and ADME -related genes might help to discriminate responder/non-responder patients, we applied a classifier to mine classification rules able to figure out the principal signatures for discriminating among patients belonging to the responder/non-responder to sorafenib phenotypes. The novelty of our study lies in the identification of 10 rules in different genotype associations for the identification of the non-responder phenotype, and 3 rules for the responder type. These rules may represent a genetic signature which could allow the stratification of patients who are fit for sorafenib treatment. We found that the genetic signature including 3 ADME-SNPs, SLC22A14 (rs171248), ADH1A ( rs6811453) and CYP26A1 ( rs7905939), and 3 known SNPs in angiogenesis-related genes, VEGF-A (rs2010963), VEGF-C (rs4604006), and HIF-1A (rs12434438), was correlated to sorafenib response in our dataset, allowing the discrimination between responders/non-responders according to mRECIST criteria. The signature of response, identified by a decision tree, was also validated by the GRS analysis.
However, while the role of angiogenesis related genes is well known in HCC patients treated with sorafenib, little information is reported in the literature regarding the role of selected ADME genes in this context. SLC22A14 , also known as organic cation transporter-like 2 (OCTL2), is a gene encoding a member of the organic-cation transporter family and anions (OATs), whose expression is high in the liver. As for the other SLCs, it is involved in regulating the homeostasis of metabolites and the uptake of a wide range of molecules and the disposition of drugs, as well as in promoting cell proliferation, migration, and invasion in HCC.40 The function of SLC members in sorafenib resistance is not clear, and only recently have studies begun to investigate their role in chemoresistance,41-44 highlighting the role of aberrant variants or SNPs in organic cation transporters during liver carcinogenesis, with effects on the ability of HCC to respond to sorafenib.
The ADH1A gene catalyzes the oxidation of alcohols to aldehydes and belongs to the superfamily of dehydrogenase enzymes. Several reports provide a correlation of ADH1A and other ADH s expression with increased risk of liver cancer, with an impact on the prognosis for HCC patients.20,45
CYP26A1 , a member of the cytochrome P450 enzyme superfamily, is mainly involved in retinoic acid metabolism and the synthesis of cholesterol, steroids, and other lipids, and it contributes to the development and progression of multiple cancers.46-48Previous studies in HCC have shown that CYP26A1 mRNA is downregulated in tumor tissue compared to paired-matched non-tumor tissues,49 but the role of CYP26A1 in HCC is not entirely clear despite being reported as hypovitaminosis. A, as potential result from a CYP26A1 depletion, could be correlated to higher risk of carcinogenesis.50
To support the correlation between ADME- and angiogenesis-related genes, the network and PEA analysis highlighted the association of 8/12 identified genes in topological key points with a relevant node degree score in important pathways underlying biological mechanisms implicated in HCC and sorafenib. In fact, VEGF-A, together with ADH1A ,CYP26A1 , and VEGF-C , showed a common interaction in “signal transduction pathways” which are known to be dysregulated in HCC, with consequent uncontrolled cell division and metastasis,44 alteration of intracellular regulators or extracellular signals with abnormal epigenetic modification, and gene expression in the tumor microenvironment. Moreover, all seed genes identified were involved in multiple pathways, showing their involvement in significantly affected by biological functions.
Given that the most recent AASLD guidelines on HCC systemic therapy have approved the use of drugs like bevacizumab+atezolizumab and lenvatinib as first-line therapies, in addition to sorafenib, the opportunity to verify a score like the GRS observed in our responder patients or the genetic signature identified in our retrospective study might help clinicians to select patients with higher chances to benefit from sorafenib treatment. In the presence of a favorable GRS, physicians could treat advanced HCC patients with sorafenib as soon as possible. Conversely, patients with an unfavorable GRS might not be excellent candidates for sorafenib.