The primary end point of the study was the analysis of associations between polymorphisms with putative influence on 5-fluorouracil/irinotecan activity and progression-free survival (PFS) of patients with advanced colorectal cancer treated with first-line FOLFIRI chemotherapy. Peripheral blood samples from 146 prospectively enrolled patients were used for genotyping polymorphisms in thymidylate synthase (TS), methylenetetrahydrofolate reductase (MTHFR), excision repair cross-complementation group-1 (ERCC 1) xeroderma pigmentosum group-D (XPD), X-ray cross-complementing-1 (XRCC 1), X-ray cross-complementing-3 (XRCC 3) and uridine diphosphate-glucuronosyltransferases-A1 (UGT1 A1). TS 3′-UTR 6+/6+ and XRCC3-241 C/C genotypes were associated with adverse PFS. Hazard ratio for PFS achieved 2.89 (95% confidence interval=1.56–5.80; P=0.002) in 30 patients (20%) with both risk genotypes. Risk for Grade III–IV neutropenia was significantly associated with UGT1A1*28 7/7 genotype. These promising findings deserve further investigations and their validation in independent prospective studies.
Combination chemotherapy is the mainstay of treatment for patients with advanced colorectal cancer (CRC).1 Regimens with bolus/infusional 5-fluorouracil, folinic acid modulation and oxaliplatin (FOLFOX) or irinotecan (FOLFIRI) have become the most common first-line treatments in the metastatic disease.1 No clear superiority of one of the two regimens has been demonstrated so far,2 and their optimal combination with novel anti-EGFR and anti-VEGF target therapies is to be established.1 Genomic polymorphisms in drug target genes, genes encoding DNA-repair enzymes and detoxification pathways may influence the activity of 5-fluorouracil and irinotecan.3, 4 Therefore, associations between polymorphisms and clinical end points may help to tailor chemotherapy and find an optimal drug strategy.3
Elevated thymidylate synthase (TS) protein levels may interfere in the mechanisms of action of 5-fluorouracil.5 A recent meta-analysis confirmed poorer overall survival of CRC patients with enhanced TS activity compared to cases with low TS activity.5 The tandem repeat polymorphism (VNTR) in TS 5′-untranslated region (5′-UTR), which consists of two (2R) or three (3R) 28-bp repeated sequences,6 showed enhanced mRNA translational efficiency/stability,6, 7, 8 with upregulation of TS levels.9, 10, 11 A G/C polymorphism in the 3R allele was found to determine two additional alleles at this locus (3G or 3C).7, 8 In vitro, the 3G allele has been associated with higher reporter gene activity at both DNA transcriptional and mRNA translational levels than the 3C allele.7, 8 In vivo, 3G-containing genotypes (2R/3G, 3C/3G, 3G/3G) showed correlation with high TS mRNA expression.10, 11 An additional TS polymorphism is a 6-bp insertion/deletion (6+/6−) in the 3′-untranslated region (3′-UTR).12 Mandola et al.13 found higher stability of chimeric mRNA composed of a luciferase reporter and the 3′-UTR 6+ variant compared to a corresponding 6− construct. Higher TS mRNA levels in 6+/6+ carriers than in 6−/6− carriers11 may influence chemosensitivity to 5-fluorouracil.13
Functional polymorphisms in the methylenetetrahydrofolate reductase (MTHFR) gene14, 15 have been associated with alterations in the intracellular folate pool and methylation reactions.16, 17 Increased availability of 5,10-methylenetetrahydrofolate, which is a necessary cofactor for 5-fluorouracil inhibition of TS and silencing by hypermethylation of genes required for cell survival in the presence of cytotoxic agents,18 may cause variable 5-fluorouracil activity in vivo.
Irinotecan is a prodrug that is converted into an active metabolite 7-ethyl-10-hydroxycamptothecin (SN-38) by carboxylesterase, which has a 100- to 1000-fold higher cytotoxicity than irinotecan. SN-38 specifically targets DNA topoisomerase I (Topo I) and stabilizes the Topo I–DNA complex. Collision of this complex with moving DNA replication forks causes lethal DNA damages and cell death.19 SN-38 is metabolized in the liver by uridine diphosphate-glucuronosyltransferases (UGTs) to an inactive metabolite, SN-38 glucuronide (SN-38G).19 Because glucuronidation is the major route of detoxification and elimination of active metabolite SN-38, inherited differences in irinotecan glucuronidating capacity may have an important influence on the pharmacokinetics and toxicity of this drug.3, 4 SN-38 is largely metabolized to the inactive glucuronide (SN-38G), via the action of several UGTs, including hepatic UGT1A1, UGT1A6, UGT1A9 and extrahepatic UGT1A7. UGT1A1 is the main isoform involved in the formation of SN-38G. A common genetic polymorphism in UGT1A1 results from a dinucleotide (TA) insertion of the UGT1A1 promoter into the TATA box and the occurrence of seven TA repeats instead of the usual six. The UGT1A1*28 allele with the additional TA repeat is associated with reduced gene expression and functional activity, as measured by SN-38 and bilirubin glucuronidation.4 UGT1A1*28 has been correlated with increased risk of severe toxicity after irinotecan chemotherapy,20, 21, 22, 23 and it is unclear whether this variant may also influence treatment outcomes.23
In experimental models, SN-38 showed induction of DNA damages, which require the activation of the nucleotide-excision repair (NER), the base-excision repair (BER) and the homologous recombination repair (HR) pathways.24, 25 In vivo, the NER enzyme Excision Repair Cross-Complementation Group 1 (ERCC1) showed possible association with irinotecan efficacy.26 In vitro and in vivo studies showed that ERCC1 118 C/T, xeroderma pigmentosum group D (XPD) 312 G/A and 751 A/C, X-ray cross-complementing-1 (XRCC1) 399 G/A and X-ray cross-complementing-3 (XRCC3) 241 T/C are functional variants in genes belonging to the NER, BER and HR pathways.27 These polymorphisms are under investigation in pharmacogenetic studies in solid neoplasms and they may influence clinical outcomes of CRC patients treated with irinotecan.
We investigated possible associations between polymorphisms (Table 1) and progression-free survival (PFS) of patients with metastatic CRC treated with first-line FOLFIRI chemotherapy. PFS was preferred since overall survival might be influenced by selective use of second-line treatments with or without new target therapies, loco-regional treatments and deaths attributed to causes unrelated to colorectal cancer.28 Additional associations between genotypes and response/toxicity to chemotherapy were explored.
The analysis was performed in 146 patients and their characteristics are shown in Table 2. At the time of the final analysis (March 2007), disease progression occurred in all patients. In the whole group, the median PFS time was 8 months (range 1.5–22 months).
Clinical outcomes and genotypes
Genotypes were in Hardy–Weinberg equilibrium and their overall frequencies were consistent with those observed in previously reported investigations in Caucasians (Table 3). No significant associations between polymorphisms and demographic, clinical, or pathological characteristics were observed (data not shown). According to the RECIST criteria for the evaluation of tumor response,29 no significant association was detected between genotype frequencies and patients subdivided into responders and non-responders, but suboptimal disease control rate (patients with complete response, partial response, stable disease vs patients with disease progression) emerged for carriers of TS 3′-UTR 6+/6+ and XRCC3 241 C/C genotypes (Table 4).
In the univariate analysis and the multivariate model (Table 5), TS 3′-UTR 6+/6+ and XRCC3 241 C/C genotypes were significantly associated with increased risk of progression. Combinations of the TS 3′-UTR 6+/6+ and XRCC3 241 C/C genotypes were analyzed in 146 patients. Both risk genotypes were present in 30 patients (Group 2), one of the two risk genotypes was present in 51 patients (Group 1), no risk genotype was found in the remaining 65 patients (Group 0). Median PFS times in Group 0 patients, Group 1 patients and Group 2 patients were 10.2, 7.9 and 6 months, respectively. The adverse effect with shorter PFS in Group 2 patients was shown in the multivariate model (hazard ratio of 2.89; 95% CIs=1.56–5.80, P=0.002). Kaplan–Meier curves of the three groups are shown in Figure 1.
All patients were assessable for safety. Observed maximum toxicity per patient is shown in Table 6. Grade IV toxicity was uncommon with three patients experiencing grade IV neutropenia, one patient grade IV vomiting and one patient grade IV diarrhea. No toxic death occurred.
No significant association was found except for myelotoxicity and UGT1A1*28. Carriers of the UGT1A1*28 7/7 genotype showed higher risk for suffering from grade III/IV neutropenia than carriers of the UGT1A1*28 6/6 genotype (Table 7). A total of 1188 cycles of the FOLFIRI regimen were administered during the study, with a median of eight cycles per patient (range, 2–20 cycles). The median targeted doses of irinotecan and 5-flurouracil delivered were 80% of planned and 90% of planned, respectively. Analysis of dose intensity in each genotype showed no significant association with a trend toward reduction of irinotecan dose intensity in carriers of the UGT1A1*28 7/7 genotype. The median targeted doses of irinotecan delivered were 84, 80 and 76% of planned in carriers of the UGT1A1*28 6/6, 6/7 and 7/7 genotypes, respectively.
In our study population of patients with advanced CRC treated with first-line FOLFIRI regimen, the presence of unfavorable genotypes in TS 3′-UTR and XRCC3 241 was significantly associated with reduced PFS and the UGT1A1*28 7/7 variant showed predictive role for irinotecan-induced neutropenia . So far, pharmacogenetic studies in patients treated with irinotecan-based chemotherapy have been performed with single UGT1A1*28 analysis or a limited number of genetic variants. We prospectively studied the association between clinical outcomes and a comprehensive panel of genetic variants in major genes involved in TS regulation, folate metabolism, irinotecan metabolism and DNA repair. This approach is relevant, especially if we consider that chemotherapy drugs exert their effects through a multi-step, multi-genic cascade and, today, the majority of patients receive combination chemotherapy with drugs exploiting independent biological pathways (that is, the FOLFIRI association).
So far, clinical studies including variants in TS30, 31, 32, 33, 34 and MTHFR35, 36, 37 have not shown univocal results. It is to be said that findings across clinical studies with TS and MTHFR polymorphisms in CRC patients are poorly comparable. Differences in the clinical setting (patients treated with adjuvant or first/second line palliative chemotherapy), in the 5-flurouracil schedule (bolus/infusional administration) and in the chemotherapy regimens (5-flurouracil alone, with folinic acid modulation or coupled with other drugs) represent major limitations for interpreting these data. In some cases, results should be looked at with caution for potential methodological pitfalls such as the overall survival end point in the adjuvant setting when a limited number of events have occurred coupled with missing analysis of influent managements for patients with metastatic disease,32 or the lack of multivariate analysis for modeling marginally significant survival associations.33 We found a biologically plausible association between TS 3′-UTR 6+/6+ genotype and adverse PFS. In fact, higher TS mRNA levels in 6+/6+ carriers than in 6−/6− carriers13 have been described in preclinical models and in vivo studies. The same association did not emerge for the putative unfavorable 3R and 3G alleles in TS 5′-UTR. According to recent findings, the TS 3′-UTR variant possess additional specific biological functions that may explain its prevailing role as TS predictive genetic marker. TS 3′-UTR 6−/6− genotype was found to be significantly associated with increased folate concentrations in vivo.38 Folate is a necessary cofactor for 5-fluorouracil TS inhibition and its increased availability may favor 5-fluorouracil activity in TS 3′-UTR 6−/6− carriers. The mRNA-binding, decay-promoting protein AUF1 has been found to suppress TS mRNA levels with higher affinity for TS 3′-UTR 6− allele mRNA.39 According to these findings, it could be hypothesized that the TS 3′-UTR 6− allele may reduce an event-predisposing effect of the 3R allele (3G considering the C/G SNP in the 3R allele). The 5′-UTR VNTR-G/C double polymorphism and the 3′-UTR 6+/6− locus are not in complete linkage disequilibrium (global r2=0.42). The estimated most frequent haplotypes in our study population were: 2R/6+ (35%), 3G/6− (29%) and 3C/6+ (15%). In fact, these figures indicate that a proportion of patients are carriers of both the unfavorable 3G and the favorable 6− alleles and the latter may antagonize the adverse functional effect of the 3G allele.
In the present investigation, UGT1A1*28 confirmed a predictive role for irinotecan-associated toxicity. Carriers of the UGT1A1*28 7/7 homozygous genotype showed higher risk for suffering Grade III/IV neutropenia than UGT1A1*28 6/6 carriers. The association the UGT1A1*28 variant and side-effects from irinotecan-based chemotherapy has been described in previous studies20, 21, 22, 23 and according to present and recently reported data, it seems that the side-effect that is more frequently associated with UGT1A1*28 status is myelotoxicity.23, 40, 41, 42, 43 Toffoli et al.23 also found a favorable association between UGT1A1*28 7/7 better response rate and survival. This association did not emerge in the present as well as in recently reported studies.41, 42, 43 The UGT1A1*28 variant influences the irinotecan pharmacokinetics with enhanced bioavailability of its active metabolite SN-38.40 It is plausible that a more prolonged exposure of normal tissues to SN-38 determines a higher frequency of side-effects. However, enhanced systemic SN-38 levels may not parallel an increased amount of SN-38 in tumor lesions, where the altered blood supply may condition the bioavailability of the drug. Also, specific mechanisms of drug resistance and DNA repair may be more relevant for predicting chemosensitivity to irinotecan. Multidrug resistance is a major obstacle to successful cancer treatment. One mechanism by which cells can become resistant to chemotherapy is the expression of ABC transporters that use the energy of ATP hydrolysis to transport a wide variety of substrates across the cell membrane. Pgp (ABCB1), MRP1, MRP2 (ABCC2) and ABCG2 are human ABC transporters primarily associated with the multidrug resistance phenomenon.44 ABCG2 and MRP2 have been recently described as multidrug-resistance pumps for irinotecan.44 Naturally occurring variants in ABCG2 and MRP2 genes have been identified that might affect the function and expression of the protein.45 These variants, together with additional functional polymorphisms in UGTs (UGT1A6, UGT1A9, UGT1A7),46, 47 may contribute to variable irinotecan bioavailability and they should be considered in future studies for assessing associations with clinical end points in irinotecan-treated patients.
The present investigation included the analysis of functional polymorphisms in the DNA repair pathways, and among them, the XRCC3 241 C/C genotype showed association with adverse PFS. This finding is in agreement with data from experimental models, where the XRCC3 241 T allele (Thr/Met amino-acid variation) was found to be associated with defective DNA repair.48, 49 Notably, SN-38 specifically targets Topo I and stabilizes the Topo I–DNA complex. Collision of this complex with moving DNA replication forks causes lethal double-strand DNA breakage and cell death.19 XRCC3 product50, 51 plays an important role in DNA repair by homologous recombination of DSBs at DNA replication forks,52 whereas ERCC1, XRCC1 and XPD products are mainly involved in the NER and BER pathways.53 Specificity of the XRCC3 DNA repair function for SN-38 DNA damage may explain the observed association between the XRCC3 polymorphism and the studied clinical outcomes. Notably, BER and NER pathways seem implicated in the repair of DNA damage from other drugs used in metastatic CRC, like oxaliplatin.54
In conclusion, adverse PFS was observed in carriers of XRCC3 241 C/C and TS 3′-UTR 6+/6+ genotypes. We did not plan statistical corrections for multiple comparisons and it is to be acknowledged that significant associations due to chance could not be ruled out. However, we wish to emphasize the exploratory nature of the present investigation and the necessity of including multiple polymorphisms in pharmacogenetic studies, because of the use of combination chemotherapy with drugs exploiting different mechanisms of action. Although significant in the multivariate model, the observed associations from single genetic variants did not indicate a major genotype/phenotype effect, but notably, the combination of the two unfavorable genotypes achieved a remarkable HR for PFS (P=0.002). The non-randomized nature is an additional limitation of our investigation and prospective studies are needed for confirming and refining our findings on the predictive/prognostic role of these promising genetic variants. Validation of pharmacogenetic data will represent a relevant step for optimizing anti-cancer chemotherapy.
Materials and methods
This prospective study involved eight Medical Oncology Units in Central Italy. Eligibility criteria were: cytologically or histologically confirmed metastatic CRC, presence of at least one bi-dimensionally measurable lesion, Karnofski Performance Status (KPS)70 and indication to first-line, FOLFIRI palliative chemotherapy (irinotecan 180 mg/m2 intravenously for 2 h on day 1 followed by 5-fluorouracil 400 mg/m2 bolus and 5-fluorouracil 600 mg/m2 22-h continuous infusion on days 1 and 2, plus lederfolin 100 mg/m2 on days 1 and 2, every 2 weeks).1 Previous adjuvant chemotherapy was allowed, but it had to be completed more than six months before study inclusion. Pretreatment evaluation included a complete medical and clinical–physical examination, KPS evaluation, baseline measurement of tumor size based on CT scans, X-ray or other radiographic means (comprising full assessment of all known metastatic disease), serum chemistries (adequate hematologic function, serum creatinine, less than 1.5 times the upper limit of normal and bilirubin, less than 1.5 times the upper limit of normal), CEA. Objective response was evaluated after four cycles of treatment and then every 2 months. Side-effects were graded according to the National Cancer Institute Common Toxicity Criteria.
Patients' characteristics and their outcomes were unknown to investigators performing genetic analyses. The results of genotyping were disclosed to clinical investigators after data analysis. The study was approved by local Ethical Committees and patients provided signed informed consent.
Analysis of polymorphisms
A blood sample from each enrolled patient was used for genotyping and it was collected before starting chemotherapy. Genomic DNA was extracted from 200 μl whole blood using the QiaAmp kit (Qiagen, Valencia, CA, USA). All polymorphisms were investigated using a PCR – restriction fragment length polymorphism (RFLP) technique except variants in UGT1A1*28. UGT1A1*28 genotypes were determined using a 35-cycle PCR (1 min at 94°C, 1 min at 60°C and 1 min at 72°C) and the number of TA repeats in the 253-bp PCR product was determined using capillary electrophoresis on an ABI 310 (Applied Biosystems, Foster City, USA). The assays for studying polymorphisms were performed as described previously; details of the studied genetic variants, primer sequences and restriction enzymes are shown in Table 1.
The primary end point was the exploratory analysis of the possible association between genotypes and PFS of patients with metastatic CRC treated with FOLFIRI regimen. Additional associations between genotypes and response/toxicity to chemotherapy were investigated.
The planned accrual duration was 24 months with additional 6-month follow-up time after the end of the accrual. Assuming the presence of an unfavorable pharmacogenetic profiling (one or more adverse genotypes) in at least one-third of the patients, 103 events allow to detect an hazard ratio of 1.8 associated with the group having unfavorable genotypes (80% power and 5% type I error for a two-tailed test). Before performing clinical correlations, genotype frequencies were checked for agreement with those expected under the Hardy–Weinberg equilibrium. Linkage disequilibrium (LD) between loci was assessed via the GLUE interface (www.hgmp.mrc.ac.uk) using the Unphased software package. LD provides information about non-random association between two or more alleles. LD was estimated by r2, which can range from 0 (random co-inheritance of alleles) to 1 (complete LD).
Unconditional logistic regression models were used to assess the relative risk of grade 3 to 4 toxicity between patients with different genotypes and to control for confounding factors (including sex, age, pattern of metastatic disease and prior adjuvant chemotherapy). PFS was defined as the time from the start of chemotherapy to first appearance of disease progression or death for any cause; patients known to be alive and without progression at the time of analysis were censored at their last available follow-up assessment. Each genotype was independently analyzed and a combined analysis was planned if multiple genotypes showed a significant association. The association between genotypes and PFS was estimated by computing hazard ratios and their 95% confidence intervals from univariate and multivariate Cox regression models. Among clinical and pathologic features (age, sex, KPS, CEA, prior adjuvant chemotherapy, histology, pattern of metastatic disease), mucinous histology, high CEA levels and low KPS were significantly associated with increased risk of progression. These three features, together with the number of metastatic sites (deemed as an additional clinically relevant parameter) were included in the multivariate model for adjusting genotype associations. The χ2-test was used to compare proportions of patients for demographic and genotype factors. PFS curves were estimated using the Kaplan–Meier method statistical significance was set at P<0.05.
We thank the Consorzio Interuniversitario per le Biotecnologie (CIB) and Fanoateneo for their financial support.
About this article