Predictive and Prognostic Markers in Colorectal Cancer

Abstract

Colorectal cancer (CRC) is the third most commonly diagnosed cancer in both men and women in the United States, with an estimated 153,760 new cases predicted for 2007. Since the 1960s, 5-fluorouracil (5-FU) has remained the mainstay of therapeutic options in the treatment of advanced CRC, with response rates of 20% to 25%. The introduction of newer agents such as oxaliplatin and irinotecan in combination with 5-FU has increased response rates to 40% to 50% in advanced disease and improved overall survival. The development of monoclonal antibodies targeting the epidermal growth factor receptor or vascular endothelial growth factor has demonstrated additional clinical benefit for patients with metastatic disease. However, many patients succumb to their disease, and a significant proportion will experience severe chemotherapy-associated toxicities while deriving little or no benefit. To improve the treatment of CRC, efforts must be directed toward the identification of patients who are likely to respond to a specific therapy, those who will experience severe toxicities, and those who will benefit from chemotherapy in the adjuvant setting. However, the utility of individual markers of response, toxicity, and disease recurrence remains in question. Efforts are now under way to develop multimarker profiles that can more accurately predict disease response. In this review, we discuss both predictive and prognostic markers identified in the treatment of CRC in terms of their robustness and their ability to assist the clinician in developing the most efficacious and least toxic therapeutic strategy for each patient.

Colorectal cancer (CRC) remains the second leading cause of cancer-related death in the Western world, with an estimated 52,180 deaths expected in the United States in 2007.¹ Despite the recent introduction of agents such as the topoisomerase I inhibitor irinotecan and the platinum-based agent oxaliplatin, 5-fluorouracil (5-FU) remains the most effective therapeutic agent in CRC treatment; it is the preferred partner for irinotecan and oxaliplatin, achieving response rates of 40% to 50% and prolonging overall survival.^2–4 Novel biologic agents—such as the monoclonal antibodies cetuximab, which targets the epidermal growth factor receptor (EGFR), and bevacizumab, an inhibitor of vascular endothelial growth factor (VEGF)—have recently demonstrated additional clinical benefit for patients with metastatic CRC (mCRC).⁵

Selection of the most beneficial treatment regimens in CRC remains a challenge and is hindered by a lack of predictive and prognostic markers. In addition, drug resistance remains a major stumbling block to effective cancer treatment. In recent years, research on a global scale has attempted to define subsets of biochemical markers that may be useful predictors of response to treatment (evaluated through clinical response, toxicity, and time to disease progression) and prognostic markers to determine the aggressiveness of the disease and the likelihood of recurrence after surgery. The science of pharmacogenomics is emerging as an increasingly useful molecular tool to investigate the disparity in drug efficacy by analysis of patient variables such as genetic polymorphisms in drug targets, metabolizing enzymes, transporters, and influential receptors.⁶ Accordingly, the identification of accurate and validated predictive and prognostic markers combined with an increasing arsenal of therapeutic agents will provide the clinician with the knowledge and the means of tailoring a targeted and effective therapy to the patient’s molecular profile while minimizing life-threatening toxicities.

MAJOR GENETIC ABERRATIONS

Microsatellite Instability

Microsatellite instability (MSI) is found in many different types of cancer and characterized by a change in length of DNA microsatellites due to the insertion or deletion of repeating units. This phenomenon is caused by defects in mismatch repair (MMR) genes such as MLH1, MSH2, or MSH6, or methylation of the MLH1 promoter. A study in stage II and III CRC showed that patients with high microsatellite instability (MSI-H) had improved survival and that patients with MSI were more likely to exhibit better recurrence-free survival than those with microsatellite stable (MSS) phenotypes.⁷ Additional studies have analyzed the relationship between MSI and CRC prognosis and concluded that CRC patients exhibiting MSI had a significantly better prognosis compared to those with intact MMR but did not benefit from the administration of 5-FU therapy in the adjuvant setting.^8,9

In a recent report, the presence of a mutation in transforming growth factor— beta-RII (TGF-β-RII) was shown to improve survival in patients who also possess MSI-H. The 5-year survival rate for patients with MSI-H tumors and the TGF-β-RII mutation was 74% following adjuvant 5-FU–based therapy compared to 46% in patients with MSI-H tumors lacking the mutation in TGF-β-RII. Interestingly, 61% of stage III colon cancers in this study exhibited the TGF-β-RII mutation, indicating that this high-frequency mutation may be useful in combination with MSI status as a prognostic marker for adjuvant therapy.¹⁰

Loss of Heterozygosity of 18q and 17p

It is reported that allelic deletions involving chromosomes 18q and 17p occur in more than 70% of colorectal cancers. Such deletions are thought to signal the existence of a tumor suppressor gene in the affected region. The tumor suppressor gene p53, often referred to as the “guardian of the genome” due to its central role in detection of genotoxic stress, is located on 17p and is mutated in 40% to 60% of colorectal cancers.¹¹ p53 status has been rigorously analyzed as both a prognostic and predictive marker in CRC, with conflicting results.

Retention of 18q alleles in MSS cancers points to a favorable outcome after adjuvant chemotherapy with 5-FU–based regimens for stage III colon cancer. The 18q chromosome contains DCC (deleted in colon cancer), which is a cell adhesion molecule whose elevated expression can lead to enhanced tumor growth and metastatic potential.¹² Several studies have determined that cancers with chromosome 18q loss appear to be associated with worse disease-free and overall survival.^13,14 The Eastern Cooperative Oncology Group (ECOG) study 5202 is an ongoing prospective clinical trial that is randomizing patients with stage II disease based on their MSI and 18q status to observation vs. chemotherapy, with the intention of prospectively determining the prognostic value of molecular markers.¹⁵

MOLECULAR PREDICTORS OF RESPONSE

Fluoropyrimidines

Since its introduction in 1957, the fluoropyrimidine 5-FU has been the mainstay of therapeutic regimens in the treatment of CRC. Upon entry to the cell, it is converted to its active metabolite, 5-fluoro-2-deoxy-uridine monophosphate (FdUMP), whose primary mechanism of action is inhibition of thymidylate synthase (TS) by formation of a ternary complex. This blocks the de novo synthesis of thymidine, an essential component for DNA synthesis, and initiates DNA damage (Figure 1).¹⁶ A significant number of studies have correlated intratumoral TS levels with response to fluoropyrimidine -based therapy.^17–19 The study by Salonga et al demonstrated for the first time that patients with CRC who responded to 5-FU therapy could be segregated by analysis of three genes in the 5-FU pathway. Those patients with low expression levels of TS, thymidine phosphorylase (TP), and dihydropyrimidine dehydrogenase (DPD) all responded, and those with elevated expression in at least one of the three genes did not.¹⁹ A meta-analysis performed by Popat et al analyzed 20 studies of over 3,000 patients and concluded that those with elevated TS expression demonstrated poorer overall survival compared with those whose tumors expressed low levels.²⁰

Figure 1. Mechanisms of action of 5-FU.

Abbreviations: 5-FU = 5-fluorouracil; CH2THF = 5,10-methylenetetrahydrofolate; DPD = dihydropyrimidine dehydrogenase; dUTP = deoxyuridine triphosphate; dUTPase = dUTP pyrophosphatase; FdUDP = 5′-fluoro-2′-deoxyuridine-5′-diphosphate; FdUMP = 5′-fluoro-2′-deoxyuridine-5′-monophosphate; FUDR= 2′-deoxy-5-fluorouridine; FUMP= 5′-fluorouridine-5′-monophosphate; FUTP= 5′-fluorouridine-5′-triphosphate; LV = leucovorin; TP = thymidine phosphorylase; TS = thymidylate synthase

Regulation of TS expression was determined to be critical for the efficacy of fluoropyrimidines, therefore identifying genetic alterations that regulate TS gene expression is crucial for developing predictive markers. Functional genomic polymorphisms exist within the 5′ region and the 3′-UTR of the TS gene. The 5′ polymorphic variant results in the majority of patients possessing a series of 28bp repeats termed TS2R (2 repeats) or TS3R (3 repeats). TS2R is associated with lower TS enzyme expression in vitro and in vivo and has been associated with increased clinical benefit in fluoropyrimidine treatment. Conversely, the 3R/3R genotype has been associated with increased TS mRNA expression, a significantly reduced response rate, and increased toxicity during 5-FU therapy.^21,22 More recently, additional polymorphisms within this 28bp repeat have been identified. Mandola et al described a G–C nucleotide transition located only in the TS3R allele, which disrupts transacting factors from binding a functional E-Box element and leads to reduced TS mRNA expression.^23,24

Low TS levels are associated with improved survival—one might hypothesize, then, that these tumors would be more sensitive to 5-FU–based therapy, and several studies that used immunohistochemistry (IHC) techniques have suggested this.^17,25 However, several larger studies have concluded that elevated TS levels are more likely to benefit from 5-FU–based chemotherapy.^26,27 A possible explanation for these conflicting observations is that TS levels detected by IHC can demonstrate significant variation as a result of antibody specificity and tissue handling/preparation.

An elevated level of TS alone is unlikely to be sufficient to predict response to 5-FU. A subset of patients with low TS expression do not respond to 5-FU–based therapy and may have additional mechanisms of resistance explained by modulation of additional 5-FU–metabolizing enzymes such as DPD, methylenetetrahydrofolate reductase (MTHFR), orotate phosphoribosyl transferase (OPRT), and deoxyuridine triphosphate nucleotidohydrolase (dUTPase).

Dihydropyrimidine dehydrogenase catalyses the rate-limiting step in the catabolism of fluoropyrimidines with more than 80% of 5-FU administered degraded by DPD in the liver. Variation in expression levels of DPD therefore has a direct effect on the bioavailability of 5-FU.²⁸ Patients who possess inactivating mutations of the DPD gene are deficient in DPD enzyme activity and encounter profound systemic toxicities that may prove fatal.²⁹ A previously mentioned study by Salonga et al demonstrated that patients with CRC who responded to 5-FU therapy had low expression levels of TS, TP, and DPD and nonresponders had elevated expression in at least one of the three genes.¹⁹ Dihydropyrimidine dehydrogenase has also demonstrated prognostic significance in several studies, demonstrating that patients with low expression had longer disease-free recurrence and increased survival than those with high expression.^30,31 While the utility of DPD as a marker of toxicity has been firmly established, its role in predicting response to 5-FU is complicated by widely variable expression levels in tumor and normal tissues.³²

Additional enzymes involved in fluoropyrimidine metabolism have been implicated in determining drug sensitivity including OPRT, the enzyme catalyzing the reduction of FUDP to the active metabolite FdUMP which irreversibly binds to TS. A recent study directly linked OPRT expression to 5-FU sensitivity, concluding that increased OPRT mRNA expression (P = .0008) and high OPRT/DPD mRNA ratio (P = .003) predicted response to fluoropyrimidine- based therapy in a small cohort of patients with mCRC.³³

Methylenetetrahydrofolate reductase is an enzymatic determinant of intracellular folate levels. A polymorphic region of the MTHFR gene termed 677T results in decreased enzymatic activity and subsequent increased levels of 5,10 methylenetetrahydrofolate and is associated with a significantly improved response to 5-FU.³⁴

dUTPase is a key regulator of intracellular dUTP pools and has been shown to be an important determinant of cytotoxicity mediated by TS inhibitors by regulation of dUTP pools and prevention of detrimental uracil misincorporation into genomic DNA in the absence of thymine.³⁵ A small retrospective study determined that elevated expression of dUTPase was associated with resistance to 5-FU, shorter time to progression, and reduced overall survival.³⁶

Irinotecan

Irinotecan targets the DNA topoisomerase I (topo I) enzyme causing inhibition followed by accumulation of DNA damage and subsequent cell death and is currently approved for the treatment of mCRC. Irinotecan itself exerts little cytotoxicity and requires conversion to SN-38 (7-ethyl-10-hydroxyl-camptothecan) by carboxylesterase (CES), an enzyme found in serum, liver, intestine, and other tissues (Figure 2).³⁷ Irinotecan therapy is frequently associated with high toxicities, including chronic life-threatening diarrhea, and treatment mortality of almost 7% has been reported. ³⁸ SN-38 is detoxified primarily by the hepatic isoform 1A1 of the uridine diphosphate glucuronosyltransferase (UGT) enzyme to SN-38 glucuronide (SN38G), which is excreted in bile and urine.³⁹ A common polymorphic variant exists in the UGT1A1 promoter, resulting in an additional TA repeat in the TATA sequence termed UGT1A1*28 and is associated with a significant decrease in SN-38 glucuronidation and a reduced ability to detoxify SN-38.⁴⁰ Additional studies demonstrated that the TA alleles could accurately predict severe gastrointestinal and bone marrow toxicity⁴¹ and that the frequency of UGT1A1 polymorphic variants were significantly different in black patients compared with white patients.⁴² The irinotecan label has recently been modified to warn that patients homozygous for the UGT1A1*28 allele are at high risk of developing severe neutropenia and that a lower initial dosing schedule should be implemented in this genotype.⁴³

Figure 2. Schematic of the steps involved in the activation and metabolism of CPT-11 (irinotecan) as well as its proposed mechanisms of action and resistance.

Abbreviations: β Gluc = beta-glucuronidase; BCRP = breast cancer resistance protein; CE = carboxylesterase; ds = double strand; MDR1 = multidrug resistance protein 1; Topo I= topoisomerase I; ss = single strand; UGT = uridine diphosphate glucuronosyltransferase

Reliable markers predicting the efficacy of irinotecan in patients with CRC have yet to be clearly defined. However, reports have suggested that increased levels and activity of the target enzyme topo I is correlated with response to irinotecan in human colon cancer cell lines,⁴⁴ but the clinical importance has yet to be established. Expression levels of CES in peripheral blood mononuclear cells are reported to determine the cellular accumulation of the active metabolite SN-38. Patients expressing high levels of CES mRNA demonstrated a significantly higher activation ratio of irinotecan to SN-38 (P = .013) and displayed increased incidence of grade 3/4 neutropenia.⁴⁵

A small retrospective study (N = 33) analyzed mRNA expression levels in a subset of selected targets including the irinotecan drug target (topo I), tumor angiogenic factors (interleukin- 8 [IL-8], VEGF, EGFR), DNA repair components (excision repair cross complementing group 1 [ERCC1]), and drug detoxification factors (GST-P1). This study reported that gene expression levels of EGFR, ERCC1, and GST-P1 may be useful in predicting clinical outcomes in patients with mCRC treated with first-line irinotecan-based chemotherapy. ⁴⁶ A larger study also investigated the significance of topo I expression in 62 patients with advanced disease receiving first-line 5-FU/irinotecan and concluded that expression levels of topo I had no influence on objective response, time to progression, and overall patient survival.⁴⁷

The elimination of irinotecan and its metabolites occurs primarily through biliary and intestinal efflux with the assistance of the ATP-binding cassette transporter superfamily. These transporters include MDR1 P-glycoprotein (ABCB1), breast cancer resistance protein (ABCG2), and the multidrug resistance–associated protein 1 and 2 (ABCC1 and ABCC2). Several of these members have been implicated in influencing the pharmacokinetics of irinotecan. In particular, ABCB1 1236 C→T transition was reported to increase significantly patient exposure to irinotecan (P = .038) and SN-38 (P = .031), which may in turn affect both toxicity and drug efficacy.⁴⁸

Oxaliplatin

Oxaliplatin is a third-generation platinum analog in which the 1,2-diaminocyclohexane (DACH) ligand substitutes the amine groups of cisplatin. Oxaliplatin also exhibits a relatively favorable toxicity profile and in many instances is clinically favorable to cisplatin, with reduced toxicity and substantial activity against cisplatin-resistant tumors.⁴⁹ Cytotoxic platinum compounds form positively charged species that cause DNA-damaging cross-links blocking both DNA replication and transcription and initiating apoptosis.⁵⁰

Several distinct mechanisms are proposed to mediate response to oxaliplatin including increased drug inactivation and efflux, decreased cellular uptake/accumulation, enhanced tolerance to pt-DNA damage, and an increase in the efficiency of DNA repair mechanisms (Figure 3). The nucleotide excision repair (NER) pathway represents the only mechanism described to date for removal of the bulky pt-DNA adducts induced by oxaliplatin treatment.⁵¹ One conserved member of the NER pathway implicated in mediating response to oxaliplatin is ERCC1. ERCC1 forms a complex with xeroderma pigmentosum group F (XPF), which recognizes and cleaves the 5′ damaged DNA strand in lesion repair. A number of studies have correlated ERCC1 gene expression levels with clinical outcome in patients receiving a platinum-based regimen.^52–54

Figure 3. Schematic of the steps involved in the activation and metabolism of oxaliplatin as well as proposed mechanisms of action and resistance.

Abbreviations: ERCC1 = excision repair cross complementing protein 1; XRCC1 = xray repair cross complementing 1; XPD = xeroderma pigmentosum group D; GSH = glutathione; GST = glutathione-S-transferase; MT = metallothionein molecules; Oxali = oxaliplatin; pt = platinum

In a study of 50 patients with advanced disease refractory to 5-FU/irinotecan chemotherapy, ERCC1 mRNA expression level (along with TS) was determined to be an independent predictive marker of survival in 5-FU/oxaliplatin chemotherapy (P < .001).⁵⁵ Furthermore, polymorphic variants within the ERCC1 gene have been identified and associated with clinical outcome in patients receiving 5-FU/oxaliplatin.⁵⁶ The frequency of these polymorphisms varies with race and may account for reduced response rates in black patients when compared with white patients.⁴²

Interestingly, xeroderma pigmentosum group D (XPD) is another gene that codes for an important protein in the NER pathway. A polymorphic variant at position 751 results in a lysine to glutamine substitution and has been linked with significantly lower response rates in a study of 73 patients receiving 5-FU/oxaliplatin. Patients with the Lys/Lys genotype demonstrated a median survival of 17.4 months whereas those possessing the Lys/Gln and Lys/Gln demonstrated 12.8 and 3.3 months respectively (P = .02).⁵⁷

X-ray repair cross complementing 1 (XRCC1) is involved in the repair of single-strand breaks following base excision repair and has been demonstrated to mediate repair of alkylating agent-induced DNA damage.^58,59 A study by Stoehlmacher et al analyzed gene expression of the XRCC1 polymorphic variant found at codon 399 in a 61-patient study of 5-FU/oxaliplatin.⁶⁰ Seventy-three percent of patients with the favorable Arg/Arg genotype responded to treatment, and patients who possessed at least one Gln allelic polymorphism in XRCC1 were 5.2-fold more likely to fail 5-FU/oxaliplatin chemotherapy. This polymorphic was also identified as demonstrating significant heterogeneity according to race—again a possible determinant of race-specific response.⁴² One study demonstrated the potential benefits of performing multivariate analysis of multiple gene polymorphisms in patients with refractory colorectal cancer and identified a gene dosage effect on 5-FU/oxaliplatin treatment response with patients with two or more unfavorable XPD, TS, ERCC1, and GST-P1 polymorphisms having a significantly reduced OS.⁶¹ The precise clinical effect of modulation of components of the DNA repair pathways in platinum drug efficacy remains to be clearly established.

The majority of oxaliplatin that enters cells never becomes associated with DNA. A possible explanation for this is inactivation via formation of conjugates between glutathione, detoxification by the glutathione-S-transferase (GST) family, and exportation of the complexes via the ABC superfamily.^62,63 Studies have identified polymorphisms in GST enzymes that have been correlated with response to platinum agents. One polymorphic variant results in an amino acid substitution at position 105 of GST-P1, which is reported to diminish enzymatic activity. A study of 107 patients determined that those homozygous for the val/val genotype demonstrated significantly longer survival compared with those heterozygous or homozygous for the wild type⁶⁴ and an additional study linked the GST-P1 105V polymorphism to sensory neurotoxicity in patients receiving FOLFOX.⁶⁵

NOVEL THERAPIES

Epidermal Growth Factor Receptor

The epidermal growth factor receptor (EGFR; ErbB-1; HER-1) is a transmembrane receptor with an intracellular tyrosine kinase domain. Its main activating ligands are epidermal growth factor (EGF) and transforming growth factor (TGF)-α. The EGFR pathway is activated after homodimerization of two EGFRs or heterodimerization with additional members of the ErbB family. Phosphorylation of the tyrosine kinase leads to activation of several intracellular signaling pathways, including phosphoinositide 3-kinase/protein kinase C (PI3k/PKC), mitogen-activated protein kinase (MAPK), and signal transducer and activator of transcription 3 (STAT3), that results in downstream effects on cell cycle regulation, proliferation, migration, angiogenesis, and inhibition of apoptosis (Figure 4).⁶⁶ Dysregulation of the EGFR pathway by mutation, overexpression, or EGFR stimulation by an excess in growth factors promotes growth and progression of many tumors, including CRC, and is associated with aggressive disease and poor prognosis.⁶⁷

Figure 4. Simplified diagram illustrating multiple pathways and downstream effects modulated by EGFR and VEGFR signaling in colorectal cancer.

Abbreviations: EGF = epidermal growth factor; EGFR = epidermal growth factor receptor; MAPK = mitogen-activated protein kinase; PI3K = phosphoinositide 3- kinase; STAT = signal transducer and activator of transcription; VEGF = vascular endothelial growth factor; VEGFR = vascular endothelial growth factor receptor

It is reported that 50% to 70% of colorectal cancers exhibit EGFR expression, but as yet there is no conclusive evidence with regard to the role of EGFR as a prognostic marker.⁶⁸ In a study of 249 patients with CRC, almost 50% demonstrated high levels of EGFR expression with no correlation to tumor stage or progression, but variation in expression between primary tumor and lymph node metastases was observed.⁶⁸ A recent retrospective analysis in 134 patients investigated EGFR as a prognostic indicator in stage II CRC and concluded that it was an independent predictor of recurrence (P = .05) and that elevated expression was associated with poor survival (P = .01).⁶⁹ A study by Zhang et al also identified two EGFR polymorphisms that together predicted pelvic recurrence in patients previously treated with chemoradiation.⁷⁰ Clearly, more research is needed to define the precise prognostic role of EGFR in CRC.

Targeting EGFR

Monoclonal antibodies represent an important clinical option in the inhibition of EGFR by binding to the extracellular domain and inhibiting activation of intracellular signaling cascades. Importantly, adverse effects during treatment are limited primarily to mild skin toxicity.⁷¹ The most promising monoclonal antibody targeting EGFR is cetuximab. The antitumor activity of cetuximab has been attributed to several distinct mechanisms including direct inhibition of tyrosine kinase activity and blockade of the EGFR signaling pathways, which results in proapoptotic, antiangiogenic, and anti-invasive effects.⁷²

A series of phase II/III clinical trials evaluated the clinical efficacy of cetuximab and demonstrated that it has activity when used in combination with irinotecan and as a single agent in advanced CRC. Interestingly, these studies found no correlation between response to treatment and EGFR expression, though this may be a reflection on the inherent limitations of the IHC techniques used to determine expression levels.^5,72 A recent study by Lauraint-Puig et al identified mutations in k-ras as a significant predictor of response to cetuximab therapy (P = .0003) and overall survival (P = .016) in a group of 30 mCRC patients.⁷³ More recently, further data were presented indicating that k-ras mutations preclude tumor shrinkage in CRC.⁷⁴ Interestingly, patients displaying undetectable levels of EGFR protein have demonstrated response to cetuximab therapy, again a reminder that presence of the target does not always dictate the response to the drug.^75,76

Skin toxicity is frequently observed in patients receiving EGFR-inhibitor treatment; this adverse effect has been associated with response and overall survival and may serve as a visible marker of antitumor activity and therapeutic efficacy.^5,71,77 A small pilot study analyzed mRNA expression levels of key genes by quantitative reverse-transcription polymerase chain reaction in an effort to identify molecular determinants of cetuximab efficacy. Higher expression levels of VEGF were associated with resistance to cetuximab (P = .038) and the combination of low gene expression levels of COX-2, EGFR, and IL-8 was significantly associated with overall survival (13.5 vs. 2.2 months, P = .028). Interestingly, this study also correlated skin toxicity with survival and identified low mRNA levels of COX-2 as a marker of grade 2/3 skin reactions during cetuximab therapy.⁷⁸

In a follow-up study, the same group attempted to determine the effects of polymorphisms in EGFR-signaling pathways in mediating response to single-agent cetuximab therapy. Their results identified a significant association between the cyclin D1 (CCND1) A870G polymorphism and overall survival in 39 CRC patients. Those patients possessing the AA homozygous genotype survived for a median of 2.3 months whereas those homo- or heterozygous for the G allele survived for a median of 8.7 months (P = .019). Analysis of the CCND1 and EGF polymorphisms together resulted in patients with favorable genotypes (EGF any A allele and CCND1 any G allele) demonstrated a median survival of 12 months, whereas patients with any two unfavorable genotypes (EGF GG or CCND1 AA) showed a median survival time of 4.4 months (P = .004).⁷⁹ It is important to point out that the authors noted the hypothesis-generating nature of these studies and the need for further validation in larger prospective clinical trials.

Antibody-dependent cellular cytotoxicity (ADCC) is another mechanism by which IgG1 antibodies exert their antitumor effects by initiation of an immune response. The Fcγreceptor family is reported to mediate the ADCC component of cytotoxicity,⁸⁰ and polymorphisms within these genes have been demonstrated to modulate tumor response to rituximab in follicular lymphoma.⁸¹ A small retrospective study analyzed the effects of two polymorphisms within FcγRIIIa (158 Val/Phe) and FcγRIIa (131 His/Arg) in 39 EGFR-expressing mCRC patients and determined that patients with FcγRIIa 131 His/His or His/Arg demonstrated better time to progression and overall survival than those with the 131 Arg/Arg genotype (P = .037) following single-agent cetuximab. This report indicates that ADCC may represent an important mechanism of action of the EGFR inhibitor cetuximab and that analysis of the FcγRIIIa polymorphisms may be useful markers for clinical outcome in mCRC patients treated with single-agent cetuximab, but these pilot studies must be validated in larger and prospective trials.⁸²

Recently the US Food and Drug Administration (FDA)-approved panitumumab, another anti-EGFR monoclonal antibody, for the treatment of mCRC. Clinical trials are already reporting promising results for panitumumab in the treatment of advanced CRC,⁸³ and retrospective analysis from these trials will yield preliminary information on potential markers of response.

Vascular Endothelial Growth Factor

Vascular endothelial growth factor-A is a secreted ligand with specific receptors that are primarily expressed by angioblasts and endothelial cells.⁸⁴ The VEGF pathway plays a major role in tumor growth and angiogenesis. The formation of new blood vessels carrying oxygen, nutrients, growth factors, and hormones is an important factor required for proliferation of solid tumors. Activation of the VEGF/VEGFR axis triggers multiple intracellular signaling pathways such as the PI3-kinase, MAP-kinase, and focal adhesion pathways. This cascade of signaling results in increased vascular endothelial cell proliferation, enhanced permeability, invasion, migration, and survival (Figure 4). Overexpression of VEGF and increased circulating VEGF have been associated with tumor progression and poor prognosis in several gastrointestinal tumor types including CRC.^85–87 Many solid tumors are also reported to secrete elevated levels of VEGF to stimulate vascularization and initiate metastasis.⁸⁸ In a sizable study of 121 patients with stage II colon cancer, VEGF-positive tumors were associated with a 50% chance of disease recurrence compared to 11% in VEGF-negative tumors.⁸⁹ VEGF-positive tumors also exhibit a 4.5- fold higher rate of recurrence in resected stage III CRC than VEGF-negative tumors.⁹⁰

Targeting VEGF

Due to its central role in promoting tumor growth, angiogenesis, and metastasis, VEGF has become a target for therapeutic intervention. Bevacizumab is a recombinant humanized monoclonal antibody targeted against VEGF. It was the first antiangiogenic drug approved by the FDA for treatment of mCRC in combination with 5-FU–based therapy, where it provides a statistically significant and clinically meaningful improvement in overall and progression-free survival when compared with 5-FU–based therapy alone in mCRC.^91,92

Although these reports demonstrate the potential efficacy of antiangiogenic therapy, the identification of biomarkers to determine those who will benefit from these agents is of great interest. Preclinical data suggest that dysregulation and mutation in the Ras/Raf/Mek/Erk93 and p53 pathways may modulate the efficacy of anti-VEGF therapies such as bevacizumab.⁹⁴

A retrospective study by Ince et al analyzed the expression and mutation status of k-ras, b-raf, and p53 to predict which patients were more likely to respond to bevacizumab. The authors reported no statistical significance between mutations of k-ras, b-raf, or p53 (mutation and overexpression) and the increase in median survival associated with the addition of bevacizumab to irinotecan/5-FU/leucovorin (IFL) therapy, but did note that patients with no mutations in k-ras or b-raf had statistically significant better overall survival irrespective of treatment received.⁹⁵

An additional study analyzed host factors involved in the negative regulation of angiogenesis as potential predictors of response to antiangiogenic therapies including microvessel density (MVD), the antiangiogenic thrombospondin (THBS) family member THBS-2, and VEGF. The results of the study indicated that patients with mCRC will benefit from the addition of bevacizumab regardless of VEGF, THBS-2 expression, and MVD.⁹⁶ Future studies aim to analyze biomarkers involved in activation of VEGF signaling (Src kinases), downstream molecules, such as carcinoembryonic antigen-related cell adhesion molecule and neuropilin-1 (a novel VEGF receptor), and VEGF-independent pathways such as IL-8 and adrenomedullin.

DISCUSSION

While analysis of individual genes in predicting response to therapy and disease prognosis has on occasion proven useful, it is becoming increasingly apparent that response to therapy and disease progression are largely driven by complex, multifaceted pathways; hence, analysis of any one single marker is unlikely to accurately predict response or progression with sufficient resolution and consistency. The 2006 Update of the ASCO (American Society of Clinical Oncology) Guideline for the Use of Tumor Markers in Gastrointestinal Cancer (www.asco.org) recently reviewed the current data assessing the use of tumor markers in screening, treatment, and surveillance of CRC. They concluded that there is insufficient evidence to recommend the routine use of p53, TS, DPD, TP, ras, 18q LOH, or MSI status in CRC treatment, primarily due to conflicting reports, disparity in detection/measurement of genes and enzyme activity, and variation in data analysis and interpretation.⁹⁷ In fact, with much conflict in published literature, the distinct lack of validated markers for colorectal cancer is alarming.

In an effort to address this, the emergence and increasing resolution of gene expression profiling and microarray analysis has already had a significant effect on the classification and prognosis of multiple forms of cancer. Expression profiling in colon cancer can now readily identify and discriminate with increasing resolution between tumors of different stage and prognosis.⁹⁸ Furthermore, efforts are now under way to identify subsets of genes that can accurately predict response to chemotherapy⁹⁹ and to build “prognosis predictors” to classify patients accurately with intermediate-stage disease, allowing for more appropriate administration of adjuvant therapy.¹⁰⁰

The reality remains that the use of pharmacogenetic profiling in CRC to predict clinical response, progression, and toxicity is still a developing field. To make progress, there must be more coordinated evaluation of these markers before genetic information can become a routine part of clinical CRC treatment. Retrospective analyses have clearly demonstrated the proof of principle in this approach. However, the design of new prospective trials must encompass a more comprehensive and disciplined unilateral approach with defined protocols, primary end points, and increased statistical power. Follow-up studies are also required to identify the functional significance of the many mutations and polymorphic variants that exist in the patient population; such functional information will inevitably assist in unraveling the complex and multifaceted mechanisms of drug metabolism and cytotoxicity. The pharmacoeconomic benefits of the novel therapeutic agents including bevacizumab, cetuximab, and panitumumab must also be fully quantified and markers of response identified and rigorously validated so that the use of these agents can be targeted to those who will derive greatest benefit.

The rejuvenation of the cancer stem cell hypothesis is another exciting area of research that must be pursued. A growing body of evidence suggests that cancers develop from a small subset of cells with self-renewal properties analogous to organ stem cells that acquire epigenetic and genetic changes required for tumorigenicity or may represent proliferative progenitors that acquire self-renewal capabilities.¹⁰¹ If such cells retain the hallmarks of tissue stem cells in being rare and infrequent in replication, they may represent a population of cells intrinsically resistant to conventional therapies. Future therapies would therefore require objective targeting of the minority stem cell population that fuels tumor growth and regeneration and not the bulk of highly proliferating tumor cells. Indeed, once functional assays are developed to identify possible stem cell populations in solid tumors, including CRC, analysis of these cells may be the most important prognostic marker in deciding which treatment route to pursue in the clinic.

A recent workshop convened by the American Association for Cancer Research (AACR), and attended by a wide variety of cancer biology experts, highlighted the importance of pursuing the cancer stem cell field; in fact, the AACR is in the process of forming a task force dedicated to expediting progress in this promising area. Such a field offers a real possibility of identifying novel targets that could overcome drug resistance, improve therapeutic efficacy, and potentially make cancer treatment curative while preventing adverse toxicities.¹⁰²

The continuing evolution of high-throughput technologies such as microarray gene profiling, proteomic profiling, and the newly developed metabolomics field will improve the resolution and sensitivity with which we can detect such markers. These technologies will be invaluable in the mechanistic analysis of the promising new classes of drugs emerging in CRC—the insulin growth factor inhibitors, the mTOR inhibitors, and the histone deacetylase inhibitors. Together, this systems biology approach will give clinicians the ability to tailor a specific therapy to the pharmacogenetic profile of the patient and their disease and will revolutionize CRC treatment as a whole.

Goals

Identify patients who will benefit from chemotherapy. Select new chemotherapeutic combinations based on disease-specific molecular targets.

Recommendations

Implement pharmacogenomics early in drug development to assist in determining drug metabolism and avoid life-threatening toxicities. Include molecular markers in all clinical trials to establish predictive and prognostic markers (including surrogate markers) and validate target inhibition.

Challenges

Validating molecular markers and their association with clinical outcomes in prospective trials Refining technologic platforms and bioinformatics to accommodate the complexity of the multifaceted molecular map that may determine outcome Implementing these findings and methods in everyday clinical practice