Pharmacogenetics of the Primary and Metastatic Osteosarcoma: Gene Expression Profile Associated with Outcome

Osteosarcoma (OS) is the most common malignant bone tumor in children and adolescents. In recent decades, OS treatment has reached a plateau and drug resistance is still a major challenge. Therefore, the present study aimed to analyze the expression of the genes related to pharmacogenetics in OS. The expression of 32 target genes in 80 paired specimens (pre-chemotherapeutic primary tumor, post-chemotherapeutic primary tumor and pulmonary metastasis) obtained from 33 patients diagnosed with OS were analyzed by the real-time PCR methodology. As the calibrators (control), five normal bone specimens were used. The present study identified associations between the OS outcome and the expression of the genes TOP2A, DHFR, MTHFR, BCL2L1, CASP3, FASLG, GSTM3, SOD1, ABCC1, ABCC2, ABCC3, ABCC5, ABCC6, ABCC10, ABCC11, ABCG2, RALBP1, SLC19A1, SLC22A1, ERCC1 and MSH2. In addition, the expression of the ABCC10, GGH, GSTM3 and SLC22A1 genes were associated with the disease event, and the metastasis specimens showed a high expression profile of ABCC1, ABCC3 and ABCC4 genes and a low expression of SLC22A1 and ABCC10 genes, which is possibly an important factor for resistance in OS metastasis. Therefore, our findings may, in the future, contribute to clinical management as prognostic factors as well as possible therapeutic targets.


Introduction
Osteosarcoma (OS) is the most common primary malignant bone tumor in children and adolescents [1]. The overall survival probabilities have not improved during the last 30 years. Since then, the treatment has consisted of complete tumor resection after neoadjuvant chemotherapy, followed by adjuvant chemotherapy [2]. According to the EURAMOS-1 study results, the MAP regimen (Methotrexate, Adriamycin-doxorubicin, Platinol-cisplatin) must be considered the standard chemotherapy treatment for highgrade osteosarcoma [3].
Equivalent chemotherapeutic drug doses may lead to wide interpatient variability in treatment response, and it may be due to pharmacokinetic (Absorption, Distribution, Metabolism and Elimination-ADME) and pharmacodynamic (receptors and targets) differences in drugs [4]. The biological mechanisms involved in genetic variability are the Int. J. Mol. Sci. 2023, 24, 5607 2 of 14 differences in gene expression, epigenetics and genetic polymorphism [5]. Pharmacogenetics investigations have been explored in OS to understand the variability in treatment outcomes among patients [6,7]. Many pharmacogenomic studies have been conducted in OS and are beginning to yield insights into how to modify and improve chemotherapeutic approaches [8]. However, these studies have been focused merely on single nucleotide polymorphisms (SNPs) [9,10]. Moreover, a major priority in OS management is pulmonary metastasis, as this is the primary cause of death [11].

Gene Expression Profile in Primary OS, Metastatic OS and Normal Bone
The expression of 32 target genes was investigated in OS and normal bone specimens. In some cases, this expression was different when comparing pre-chemotherapy with post-chemotherapy specimens, as well as in the primary and metastatic OS. The present investigation did not detect the expression of the HMGB1 gene, neither in the OS nor in the normal bone specimens.
The relative quantification (RQ) of the target genes with statistically significant results in all analyzed specimens: pre-chemotherapy (B), post-chemotherapy (S), metastasis (M) and normal bone (NB) are presented in Figure 1. The RQ of the target genes with nonstatistically significant results is presented in the Supplementary Figure S1. Regarding the comparisons between primary OS and normal bone, evaluated by the Mann-Whitney test, higher expression of GSTM3, GGH, ABCC10 and SLC22A1 genes in OS was observed. (p = 0.037; p = 0.042; p = 0.013; and p = 0.015, respectively).

Gene Expression Profile Associated with Clinical Parameters
The clinical parameters were associated with gene expression, evaluated by the Mann-Whitney test (Table 1). Tumors from patients who were metastatic at diagnosis presented higher expression of ABCB1, ABCC6, ABCC10, BCL2L1 and SLC19A1 genes (p = 0.039; p = 0.048; p = 0.048; p = 0.026; and p = 0.010, respectively) than tumors from patients who were non-metastatic at diagnosis. Poor responders' tumors presented higher expression of ERCC1 and TOP2A genes (p = 0.021 and p = 0.036, respectively) and lower expression of ABCC3, FASLG and SLC22A1 genes (p = 0.031; p = 0.017; and p = 0.014, respectively) than good responders' tumors. The sizes of tumors resected in surgery were also associated with the expression of the investigated genes. Large tumors of 12 cm or more presented lower expression of ABCG2, CASP3 and MSH2 genes (p = 0.027; p = 0.033; and p = 0.045, respectively) than small tumors. The local control was conducted with surgery that could be either conservative or an amputation. Tumors from patients who underwent amputation presented higher expression of ABCC11, DHFR, ERCC1, GSTM3, SLC19A1 and TOP2A genes (p = 0.002; p = 0.007; p = 0.0042; p = 0.022; p = 0.002. and p = 0.010, respectively) and lower expression of FASLG, MTHFR and SLC22A1 genes (p < 0.0001; p = 0.003; and p = 0.024, respectively) than tumors from patients who underwent conservative surgery. In the relapse analyses, one patient was excluded because he had disease progression during the first treatment and died before he reached remission. Tumors from patients who relapsed presented higher expression of TOP2A (p = 0.038) and lower expression of ABCC3, ABCC5 and FASLG genes (p = 0.026; e p = 0.051; p = 0.050, respectively) than patients who not relapsed. Regarding the ABCC5 result, this was only a trend of statistical significance.

Gene Expression Profile Associated with Clinical Parameters
The clinical parameters were associated with gene expression, evaluated by the Mann-Whitney test (Table 1). Tumors from patients who were metastatic at diagnosis presented higher expression of ABCB1, ABCC6, ABCC10, BCL2L1 and SLC19A1 genes (p = 0.039; p = 0.048; p = 0.048; p = 0.026; and p = 0.010, respectively) than tumors from patients who were non-metastatic at diagnosis. Poor responders' tumors presented higher expression of ERCC1 and TOP2A genes (p = 0.021 and p = 0.036, respectively) and lower expression of ABCC3, FASLG and SLC22A1 genes (p = 0.031; p = 0.017; and p = 0.014, respectively) than good responders' tumors. The sizes of tumors resected in surgery were also associated with the expression of the investigated genes. Large tumors of 12 cm or more presented lower expression of ABCG2, CASP3 and MSH2 genes (p = 0.027; p = 0.033; and p = 0.045, respectively) than small tumors. The local control was conducted with surgery that could be either conservative or an amputation. Tumors from patients who underwent amputation presented higher expression of ABCC11, DHFR, ERCC1, GSTM3, SLC19A1 and TOP2A genes (p = 0.002; p = 0.007; p = 0.0042; p = 0.022; p = 0.002. and p = 0.010, respectively) and lower expression of FASLG, MTHFR and SLC22A1 genes (p < 0.0001; p = 0.003; and p = 0.024, respectively) than tumors from patients who underwent conservative surgery. In the relapse analyses, one patient was excluded because he had disease progression during the first treatment and died before he reached remission. Tumors from patients who relapsed presented higher expression of TOP2A (p = 0.038) and lower expression of ABCC3, ABCC5 and FASLG genes (p = 0.026; e p = 0.051; p = 0.050, respectively) than patients who not relapsed. Regarding the ABCC5 result, this was only a trend of statistical significance.

Gene Expression Profile Associated with OAS and EFS
As shown in Figure 2, patients with high expression of the ABCC5 and BCL2L1 genes in the pre-chemotherapy biopsy had a trend towards worse OAS (p = 0.051; HR = 3.42) and EFS (p = 0.058; HR = 3.27), respectively, compared with patients with low expression of the ABCC5 and BCL2L1 genes. Moreover, patients with high expression of the ABCC3 gene in the pre-chemotherapy biopsy had worse EFS compared with patients with low expression of the ABCC3 gene (p = 0.048; HR = 3.41). Patients with high expression of the TOP2A gene in the post-chemotherapy specimens had worse OAS (p = 0.015; HR = 5.37) and EFS (p = 0.005; HR = 6.36) compared with patients with low expression of the TOP2A gene. Furthermore, patients with low expression of the RALBP1A gene in the post-chemotherapy specimens had a trend towards worse OAS compared with patients with high expression of the RALBP1 gene (p = 0.051; HR = 3.40). Patients with low expression of the BCL2L1 and MTHFR genes in the metastasis had worse OAS (p = 0.018, HR = 3.53; and p = 0.027, HR = 3.27, respectively) and worse EFS (p = 0.019, HR = 3.29; and p = 0.024, HR = 3.16, respectively) compared with patients with high expression of the BCL2L1 and MTHFR genes. Moreover, patients with low expression of the ABCC2, RALBP1 and SOD1 genes in the metastasis specimens had worse EFS (p = 0.048, HR = 3.16; p = 0.022, HR = 3.26; and p = 0.027, HR = 3.14, respectively) compared with patients with high expression of the ABCC2, RALBP1 and SOD1 genes. HR = 3.16, respectively) compared with patients with high expression of the BCL2L1 and MTHFR genes. Moreover, patients with low expression of the ABCC2, RALBP1 and SOD1 genes in the metastasis specimens had worse EFS (p = 0.048, HR = 3.16; p = 0.022, HR = 3.26; and p = 0.027, HR = 3.14, respectively) compared with patients with high expression of the ABCC2, RALBP1 and SOD1 genes.

Discussion
The expression of genes analyzed in the present study was investigated for the first time in paired OS specimens. When comparing OS samples obtained pre-and posttreatment, as well as from metastases, we detected different levels of expression of the selected genes. Moreover, this study showed that the genes related to a treatment response could be associated with OS tumorigenesis.
TOP2A is a target for several anticancer agents, such as doxorubicin, and a variety of mutations in this gene have been associated with the development of drug resistance. This nuclear enzyme is involved in processes such as chromosome condensation, chromatid separation and the relief of torsional stress that occurs during DNA transcription and replication [12]. A meta-analysis showed that high TOP2A expression is associated with a worse prognosis in many types of cancer [13]. In OS, the presence of TOP2A amplification tends to relate to a worse overall survival rate [14]. The present study showed an association between high expression of TOP2A and poor response, amputation and relapse. Moreover, high expression was also associated with worse OAS and EFS.
DHFR, GGH and MTHFR are genes involved in the methotrexate pathway and response [15]. Methotrexate resistance in human OS cells is associated with an

Discussion
The expression of genes analyzed in the present study was investigated for the first time in paired OS specimens. When comparing OS samples obtained pre-and posttreatment, as well as from metastases, we detected different levels of expression of the selected genes. Moreover, this study showed that the genes related to a treatment response could be associated with OS tumorigenesis.
TOP2A is a target for several anticancer agents, such as doxorubicin, and a variety of mutations in this gene have been associated with the development of drug resistance. This nuclear enzyme is involved in processes such as chromosome condensation, chromatid separation and the relief of torsional stress that occurs during DNA transcription and replication [12]. A meta-analysis showed that high TOP2A expression is associated with a worse prognosis in many types of cancer [13]. In OS, the presence of TOP2A amplification tends to relate to a worse overall survival rate [14]. The present study showed an association between high expression of TOP2A and poor response, amputation and relapse. Moreover, high expression was also associated with worse OAS and EFS.
DHFR, GGH and MTHFR are genes involved in the methotrexate pathway and response [15]. Methotrexate resistance in human OS cells is associated with an amplification and/or overexpression of its target, the DHFR [6]. We observed that patients who under-went amputation presented metastasis with higher DHFR expression than patients who underwent conservative surgery. Increased levels of GGH led to a decreased accumulation of polyglutamated MTX and MTX resistance [16]. In OS, the ratio between the patients and the controls for the polymorphisms GGH_452T/C, GGH_401T/C and GGH_16T/C was greater than 1.5. The GGH_401C/T variant enhanced promoter activity, increasing protein expression [9]. In the present study, it was observed that OS presented higher GGH expression than in normal bone. The rs1801133 polymorphism of the MTHFR has been the most frequently studied in OS and leads to a C to T substitution, resulting in decreased enzymatic activity. In OS, the TT genotype was significantly associated with toxicity [15]. However, we observed that low expression in OS metastasis was associated with worse OAS and EFS. Moreover, the patients who underwent amputation had lower MTHFR expression in the primary tumor compared with the patients who underwent conservative surgery.
Regarding apoptosis, BCL2L1, CASP3 and FASLG were investigated in the present study. The longer isoform of BCL2L1 acts as an apoptotic inhibitor and the shorter isoform acts as an apoptotic activator [17]. We found that high BCL2L1 expression in the primary tumor was associated with metastasis at diagnosis and a worse EFS. Nevertheless, in the metastatic tumor, low expression was associated with worse OAS and EFS. This discrepancy in our results could be explained by the theory that the primary tumor expresses the longer isoform (anti-apoptosis) and a metastatic tumor expresses the shorter isoform (proapoptosis), since chemotherapy drugs stimulate the production of the shorter isoform [18]. The G allele of the variant rs2720376, linked with lower CASP3 expression, was associated with a lower EFS in OS [10]. In this study, low CASP3 expression was associated with large tumors. The variant rs763110, linked to a lower FasL expression, was associated with a lower EFS in OS [10]. We found that low FASLG expression was associated with a poor response, amputation and relapse.
Regarding genes related to detoxification of the chemotherapeutic drugs, GSTM1 and GSTP1 presented no association with the outcome in OS. The polymorphism in GSTM3 (AA versus BB) has been associated with OS risk [19]. The present study observed an association between high GSTM3 expression and amputation. Moreover, OS presented higher GSTM3 expression than normal bone. The GSTM3 polymorphism could confer different efficiencies in the metabolism of carcinogens and has been shown to modulate various cancers' risk [20]. The null GSTT1 genotype was associated with OS risk [19]. Resistant cell lines of OS showed lower SOD1 expression than their parental cells [21]. We found that metastasis presented lower SOD1 expression than the primary tumor, and patients with low expression of the SOD1 gene in the metastasis had worse EFS.
Genes involved in the repair of DNA adducts induced by cisplatin, which thereby influence cisplatin efficacy, have been investigated by the largest number of studies on OS [7]. ERCC1 positivity has presented an association with poor EFS and OAS in OS [22]. The present study showed an association between high expression and poor response and amputation. The ERCC2 rs1799793 polymorphism was related to the high risk of OS development [23]. This study showed that metastasis presented lower ERCC2 expression than primary tumors. We found that low MSH2 expression was associated with large primary tumors. Metastasis specimens presented lower expression than primary tumors. In OS, the variant rs4638843 in MSH2 was associated with a worse EFS [10]. Moreover, a wide investigation of childhood cancers found germline mutations of MSH2 in OS [24]. Taken together, our results showed that metastasis in OS presents low expression of the ERCC2 and MSH2 genes compared with pre-chemotherapy biopsy, which could be related to decreased ability to repair DNA damage in metastasis, possibly resulting in genetic alterations accumulation and more aggressive cancer [25].
High efflux transporter gene expression and low influx transporter gene expression are the main resistance mechanisms related to cisplatin, doxorubicin and methotrexate, in vitro. Moreover, many polymorphisms in these genes have been related to treatment response in OS [6,7,[26][27][28]. The present study, for the first time, investigated the expression of transporter genes in paired specimens. The results showed that the tumor biopsy presented high ABCC6 and ABCC10 expression, and metastasis presented high ABCB1 expression when metastasis was present at diagnosis. The patients with high ABCC3 and ABCC5 expression in biopsy presented worse EFS and OAS, respectively, and patients with low ABCC2 expression in metastasis presented worse EFS.The patients with low RALBP1 expression in surgery and metastasis presented worse OAS and EFS, respectively.
OS presented higher ABCC10 and SLC22A1 expression than normal bone. However, low SLC22A1 expression was associated with a poor response and amputation, probably due to its influx function. This is the first investigation regarding SLC22A1 and OS. SLC22A1 could be activated by miR-21, which is overexpressed in OS and was associated with tumorigenesis [29]. Moreover, metastasis presented higher ABCC1, ABCC3 and ABCC4 expression and lower SLC22A1 and ABCC10 expression than the primary tumor. This pattern could contribute to the lower intracellular concentration of the chemotherapeutic drugs. Consequently, it could contribute to the mechanism of resistance in metastasis, which is the main cause of death in OS patients. Therefore, with the knowledge of the metastasis profile, it is possible to develop new strategies for these patients. CBT-1 ® is an adjunct to chemotherapy in all cancer types with multi-drug resistance. Eight clinical trials are evaluating CBT-1 ® in patients with many cancer types, such as acute myelogenous leukemia, breast, non-Hodgkin's lymphoma, Hodgkin's disease, lung, and sarcoma [30,31]. Moreover, CBT-1®was able to revert the ABCB1/ABCC1-mediated resistance against doxorubicin in OS cell lines [32]. In the future, it could be interesting to evaluate CBT-1 ® in metastatic OS patients In conclusion, the present study identified associations between OS outcome and expression of the genes TOP2A, DHFR, MTHFR, BCL2L1, CASP3, FASLG, GSTM3, SOD1, ABCB1, ABCC2, ABCC3, ABCC5, ABCC6, ABCC10, ABCC11, ABCG2, RALBP1, SLC19A1, SLC22A1, ERCC1 and MSH2. In addition, the pre-chemotherapy biopsy from OS patients had higher gene expression of ABCC10, GGH, GSTM3 and SLC22A1 compared with bone specimens obtained from healthy subjects, and the metastasis specimens showed a high expression profile of ABCC1, ABCC3 and ABCC4 and low expression of SLC22A1 and ABCC10, which is possibly an important factor for resistance in OS metastasis. In summary, we found that the expression of genes related MAP pharmacokinetic and pharmacodynamic modeling, as well as cell death and DNA damage repair processes are associated with OS tumorigenesis and MAP response in OS patients. Therefore, in the future, our findings may contribute to clinical management as prognostic markers and also as possible therapeutic targets.

Patients and Specimens
We investigated 80 paired specimens obtained from 33 patients with diagnoses of OS. These patients were admitted to the Pediatric Oncology Institute (IOP/GRAACC/UNIFESP) between 2006 and 2016. The average age at diagnosis was 13 years old. Of 33 OS patients, 14 patients presented pulmonary metastasis. Thus, we investigated 33 biopsy specimens (pre-chemotherapy), 33 surgery specimens (post-chemotherapy) and 14 pulmonary metastasis specimens. Five normal bone tissues were used as a control; they were obtained from orthopedic surgeries of five healthy individuals that underwent trauma and did not present either genetic disorders or bone diseases. This study had the Research Ethics Committee approval from the Federal University of Sao Paulo (N • 0189/2016), and all patients agreed to participate by informed consent. All patients were treated following the GLATO (Grupo Latino Americano de Tratamento de Osteossarcoma-Latin American Group of Osteosarcoma Treatment) protocol of 2006, which is based on high doses of cisplatin, doxorubicin and methotrexate. All clinical data are summarized in Table 2.

Gene Expression (qRT-PCR)
The expression of 32 genes involved with pharmacogenetics was measured by a quantitative reverse transcription PCR (qRT-PCR). All frozen tissues were submitted to an RNA extraction using TRIzol ® Reagent (Thermo Fisher Scientific, Waltham, MA, USA). The cDNA was synthesized using SuperScript ® Vilo™ Master Mix (Invitrogen, Waltham, MA, USA). The qRT-PCR was performed in triplicate using TaqMan ® Gene Expression Assays (Thermo Fisher Scientific, Waltham, MA, USA) ( Table 3). The ACTB and GAPDH genes were used as endogenous controls. Normal bone was used as a calibrator. Table 3. TaqMan ® Gene Expression Assays.

Function
Gene Assay