Treatment Decisions After Diagnosis of Metastatic Colorectal Cancer

Treatment of metastatic colorectal cancer (mCRC) involves the use of active cytotoxic drugs (irinotecan, oxaliplatin, 5-fluorouracil [5-FU], and capecitabine) and biological agents (bevacizumab, cetuximab, and panitumumab) either in combination or as single agents. Until recently, the only biological agent with proven first-line efficacy was bevacizumab, but options have expanded from the data generated with anti–endothelial growth factor (EGFR) monoclonal antibodies. Anti-EGFR agents can be added to first-line FOLFIRI (5-fluorouracil, leucovorin [folinic acid], irinotecan) or FOLFOX (5-fluorouracil, leucovorin [folinic acid], oxaliplatin) in patients whose tumors express wild-type KRAS. These agents may improve outcomes when added to chemotherapy, particularly progression-free survival (PFS), and in the case of cetuximab, overall survival (OS) and response rates. The selection of first-line therapy should be based on the individual treatment goals after considering the efficacy and tolerability of each regimen. For patients with metastases confined to the liver, surgical resection offers a potentially curative approach. For initially unresectable lesions, treatment regimens offering high response rates may produce sufficient tumor shrinkage to permit complete resection. Regimens with high response rates are also preferable for patients requiring symptom relief or for those with large tumor burdens. The choice between intensive vs. nonintensive management also depends on other factors, including the patient's functional status, comorbidities, and desires. A sequential single-agent strategy or an intermittent approach (combination therapy followed by maintenance) may minimize toxicity and be appropriate for patients who are not surgical candidates, irrespective of treatment response. Guidelines, such as those of the National Comprehensive Cancer Network (NCCN), recommend that KRAS mutational status should be determined at mCRC diagnosis to identify candidates for anti-EGFR therapy whether they are used in first or subsequent lines of treatment.

Introduction

Colorectal cancer is the fourth most common malignancy in the United States but the second leading cause of cancer deaths. In 2010, colorectal cancer was newly diagnosed in an estimated 142,570 individuals and was responsible for 51,370 deaths.¹ Most patients present with localized or regional disease that is amenable to surgical resection, with adjuvant therapy offered to those with high-risk stage II or stage III disease.² However, approximately 20% of patients present with metastatic colorectal cancer (mCRC), and a significant proportion of previously treated patients have disease recurrence typically involving the liver or lungs.3, 4 Systemic therapy is the mainstay of management for most of these patients with mCRC, although some cases of metastatic disease restricted to the liver may be amenable to surgery.5, 6

The systemic treatment of mCRC involves the use of active cytotoxic drugs and biological agents either in combination or as single agents.² To maximize outcome, patients should receive oxaliplatin, irinotecan, and either 5-fluorouracil (5-FU) or capecitabine at some point during the course of treatment.⁷ Further improvement in mCRC management has been realized over the past 5-10 years with the introduction of biological agents—the antiangiogenic agent bevacizumab and the anti–endothelial growth factor receptor (EGFR) monoclonal antibodies cetuximab and panitumumab—and through the elucidation of key molecular markers. To take full advantage of these advances, clinicians need to individualize treatment courses based on clinical characteristics, molecular features, and therapeutic goals. This article reviews key clinical trial data with cytotoxic and biological agents and then discusses important factors influencing treatment decisions after the diagnosis of mCRC.

Several cytotoxic regimens have emerged as acceptable options for initial treatment of mCRC based on the results of key phase III studies conducted over the past decade (Table 19, 11, 13, 14, 17). Irinotecan plus bolus 5-FU and leucovorin (LV), known as the IFL regimen, was considered to be the standard of care in 2000, producing median progression-free survival (PFS) of 7 months and median overall survival (OS) of 14.8 months.⁸ In the Intergroup N9741 trial, oxaliplatin plus infusional 5-FU/LV (FOLFOX4) compared with IFL significantly improved median OS (19.5 vs. 15.0 months, respectively; hazard ratio [HR], 0.66; P = .0001), time to progression (TTP) (8.7 vs. 6.9 months; HR, 0.74; P = .0014), and overall response rate (ORR; 45% vs. 31%; P = .002).⁹ This study included a third arm with oxaliplatin plus irinotecan given once every 3 weeks (IROX), but its efficacy was generally comparable to IFL and inferior to FOLFOX4. From a safety perspective, FOLFOX4 was associated with lower rates of grade ≥ 3 diarrhea, vomiting, nausea, and dehydration, but higher rates of paresthesias compared with IFL. To compensate for the higher toxicity, the starting doses of irinotecan and 5-FU were subsequently reduced by 20%, but FOLFOX4 continued to maintain a survival benefit over the modified IFL regimen.¹⁰ The suboptimal toxicity profile of IFL led to the development of irinotecan plus infusional 5-FU/LV (FOLFIRI), which significantly improved PFS and showed a trend for improving OS relative to IFL in the BICC-C study.¹¹Oxaliplatin plus bolus 5-FU administration has also been studied in the FLOX (5-FU, leucovorin, oxaliplatin) regimen. A single-arm study showed that this regimen is feasible and its efficacy may be comparable to FOLFOX¹²; however this was a small study compared with the volume of data supporting FOLFOX. On the basis of these trials, 5-FU should be administered by an intravenous (I.V.) infusion schedule, and not solely by bolus injection, when used in combination with oxaliplatin or irinotecan.²

With 5-FU/LV administration optimized, attention focused on comparison of FOLFOX vs. FOLFIRI. In the Gruppo Oncologico Dell'Italia Meridionale (DOIM) 9901 trial, FOLFOX4 and FOLFIRI produced similar efficacy in terms of OS (15 months vs. 14 months; P = .29), TTP (both 7 months; P = .64), and ORR (34% vs. 31%; P = .60).¹³ Further evidence for the comparable efficacy of FOLFOX and FOLFIRI was obtained from the randomized crossover GERCOR study, in which previously untreated patients with mCRC initially received FOLFOX6 or FOLFIRI and then were switched to the other regimen at disease progression.¹⁴ Median OS was 20.6 months in patients who received FOLFOX6 followed by FOLFIRI compared with 21.5 months in those who received these regimens in the reverse order (P = .99). Moreover, median PFS (8.0 vs. 8.5 months; P = .026) and ORR (54% vs., 56%) did not differ between FOLFOX6 and FOLFIRI in the first-line setting. In both trials, gastrointestinal toxicity was more common with FOLFIRI, whereas neurosensory and hematologic toxicity were more common with FOLFOX.13, 14

The oral fluoropyrimidine capecitabine has efficacy similar to that of 5-FU/LV in mCRC15, 16 and therefore studies were conducted to ascertain if it could be used in place of infusional 5-FU/LV. The Intergroup NO16966A trial evaluated whether capecitabine plus oxaliplatin (XELOX; also known as CapeOX) was noninferior to FOLFOX4 in first-line treatment of mCRC.¹⁷ An initial cohort of 634 patients was enrolled. After phase III data with bevacizumab became available, the study was subsequently amended to have a 2 × 2 factorial design, and 1400 additional patients were randomized to FOLFOX4 vs. XELOX and then to either bevacizumab or placebo. XELOX and FOLFOX4 produced similar median PFS (8.0 vs. 8.5 months; HR, 1.04) and OS (19.8 vs. 19.6 months; HR, 0.99). The statistical analysis concluded that XELOX was noninferior to FOLFOX4 as a first-line treatment for mCRC. From a safety perspective, XELOX was associated with more grade 3 diarrhea and hand-foot syndrome, whereas FOLFOX4 caused more grade 3/4 hematologic toxicity.¹⁷ The BICC-C (Bolus, Infusional, or Capecitabine with Camptosar-Celecoxib) trial included a comparison of capecitabine plus irinotecan (XELIRI; also known as CapeIRI) vs. FOLFIRI, but failed to show that the two regimens were equivalent.¹¹ Median PFS was significantly shorter with XELIRI (5.8 vs. 7.6 months; P = .015), but median OS did not differ significantly (18.9 vs. 23.1 months; P = .29). More importantly, XELIRI was associated with substantially higher rates of severe nausea, vomiting, diarrhea, dehydration, and hand-foot syndrome, making it a suboptimal regimen in comparison with FOLFIRI.¹¹

On the basis of available clinical trial data, the following cytotoxic regimens are considered acceptable options for patients with mCRC who are suitable candidates for intensive therapy: FOLFOX, FOLFIRI, XELOX, infusional 5-FU/LV, or capecitabine.² As discussed in the following sections, clinical trial evidence is available to support the use of biological agents with these regimens. The combination of xaliplatin, irinotecan, and infusional 5-FU/LV (FOLFOXIRI or FOLFIRINOX) is also considered to be an acceptable option for first-line mCRC based on comparisons with FOLFIRI in two randomized phase III trials.18, 19 FOLFOXIRI appeared associated with greater ORR, PFS, and OS, but the differences reached statistical significance only in one of the trials. As would be expected, toxicities were also increased, particularly severe neutropenia, neurotoxicity, and diarrhea, as a result of the concurrent presentation of irinotecan- and oxaliplatin-related events. FOLFOXIRI is not recommended for use in combination with a biological agent because of the lack of sufficiently mature safety and efficacy data.²

Bevacizumab is a humanized monoclonal antibody directed against vascular endothelial growth factor (VEGF), a key factor in tumor angiogenesis. Initial evidence for its activity in first-line treatment of mCRC was obtained in the pivotal phase III AVF 2107 trial, in which 813 patients were randomized to IFL plus either bevacizumab 5 mg/kg or placebo every 2 weeks.²⁰ The addition of bevacizumab compared with IFL alone significantly improved OS (20.3 vs. 15.6 months, respectively; HR, 0.66; P < .001), PFS (10.6 vs. 6.2 months; HR, 0.54; P < .001), and ORR (44.8% vs. 34.8%; P = .004). From a safety perspective, the bevacizumab arm had a higher rate of grade 3 hypertension, with grade 4 diarrhea and leukopenia also being slightly more common.²⁰ In addition, 6 patients (1.5%) treated with bevacizumab plus IFL had gastrointestinal perforations.

The profile of bevacizumab when used in combination with contemporary first-line regimens was determined in other trials (Table 220, 22, 27, 28). As FOLFOX emerged as the standard cytotoxic treatment of choice (supported by the results of N9741; see earlier), the ECOG3200 trial became relevant in first-line treatment. Nominally, this trial enrolled patients in their second line of therapy; they had previously been treated with irinotecan and a fluoropyrimidine but were oxaliplatin and bevacizumab naive. Patients were randomly allocated to received FOLFOX4 plus bevacizumab 10 mg/kg, FOLFOX 4 alone, or bevacizumab alone every 2 weeks. Compared with FOLFOX4 alone, the addition of bevacizumab to a FOLFOX4 regimen significantly improved OS (12.9 vs. 10.8 months, respectively; HR, 0.75; P = .0011), PFS (7.3 vs. 4.7 months; HR, 0.61; P < .0001), and ORR (22.7% vs. 8.6%; P < .0001).²¹In comparison, single-agent bevacizumab exhibited little activity, with a median PFS of 2.7 months and ORR of 3.3%. Median OS was 10.2 months but may have been influenced by treatment administered after disease progression.²¹

The most recent trial of bevacizumab plus oxaliplatin therapy was NO16966. As noted previously, this trial was amended to have a 2 × 2 factorial design with patients randomized to FOLFOX4 vs. XELOX and then to bevacizumab or placebo.²² Bevacizumab was administered at a dose of 5 mg/kg every 2 weeks when given with FOLFOX4 or 7.5 mg/kg every 3 weeks when given with XELOX. The addition of bevacizumab to oxaliplatin-based therapy significantly improved the primary PFS endpoint (9.4 vs. 8.0 months; HR, 0.83; P = .0023).²² In the planned subset analysis, bevacizumab significantly improved PFS when added to XELOX (P = .0026) but not when added to FOLFOX4 (P = .187). In the survival analysis, median OS was 21.3 months in the bevacizumab arm compared with 19.9 months in the placebo arm (HR, 0.89; P = .077)—a much smaller difference than in the AVF 2107 trial. Grade 3/4 adverse events were slightly more common in the bevacizumab arm, reflecting small increases in gastrointestinal events, cardiac disorders, and hand-foot syndrome, as well as events known to be associated with bevacizumab, including thromboembolic events and hypertension.²² The clinical use of bevacizumab is lagging in terms of patient selection refinement; several small, exploratory studies have proposed candidate markers, mostly among angiogenic and inflammatory cytokines, but valid data are lacking.23, 24, 25

A number of other antiangiogenic agents are being evaluated, but none has shown a significant survival benefit in the first-line setting in phase III clinical trials. For example, cediranib is an antiangiogenic drug that blocks VEGF signaling by inhibiting the tyrosine kinase activity of all three VEGF receptors.²⁶ The HORIZON II and III trials initially evaluated cediranib at doses of 20 mg/d and 30 mg/d, respectively, but an interim analysis by an independent monitoring committee led to selection of the 20 mg/d dose for further evaluation.27, 28 Subsequent patients enrolled in these trials were allocated to receive cediranib 20 mg/d or the control. In HORIZON II, patients with previously untreated mCRC were assigned to cediranib or placebo in combination with FOLFOX4, modified FOLFOX6, or XELOX per the investigator's preference.²⁷ Addition of cediranib to oxaliplatin-based therapy significantly improved median PFS (8.6 vs. 8.2 months; HR, 0.84; P = .012) but not median OS (19.7 vs. 18.9 months; HR, 0.94; P = .57) or ORR (50.6% vs. 49.7%; P = .90). The cediranib arm had a higher incidence of adverse events, which led to a reduction in chemotherapy dose intensity.²⁷ Specifically, higher rates of grade ≥ 3 diarrhea, hypertension, neutropenia, and thrombocytopenia were seen in the cediranib arm.

In the HORIZON III trial, patients received modified FOLFOX6 in combination with either cediranib 20 mg/d or bevacizumab 5 mg/kg every 2 weeks.²⁸ Comparable efficacy was seen in the cediranib vs. bevacizumab arms for median PFS (9.9 vs. 10.3 months; HR, 1.10; P = .119), OS (22.8 vs. 21.3 months; HR, 0.94; P = .55), and ORR (46.3% vs. 47.3%; P = .67). However the safety and quality of life assessments favored the bevacizumab arm: higher rates of grade ≥ 3 adverse event rates, specifically neutropenia and diarrhea, were seen in the cediranib arm, which led to reduced delivery of chemotherapy. Interestingly, these results suggest that the activity of newer antiangiogenic agents combined with modern chemotherapy is on par with bevacizumab, but their benefit was not demonstrated with statistical significance, leaving these agents in the investigational realm.

Another promising antiangiogenic agent in mCRC is aflibercept, for which the phase III trial VELOUR trial (aflibercept added to second-line FOLFIRI) produced positive results.²⁹ The first trial expected to provide data on the efficacy of this agent in combination with first-line FOLFOX is AFFIRM (clinicaltrials.gov ID NCT00851084), expected to be completed late in 2011.

The anti-EGFR monoclonal antibodies are the first therapeutic class for mCRC tailored for use in patients with a specific molecular marker, namely wild-type KRAS (Table 330, 31, 32, 33, 34, 35, 36). The efficacy of cetuximab in first-line treatment of patients with mCRC whose tumors express wild-type KRAS was demonstrated in the CRYSTAL study.30, 31 A total of 1198 patients with EGFR-positive tumors were randomly assigned to receive FOLFIRI with or without cetuximab (400 mg/m² initially; then 250 mg/m² weekly). In the entire study cohort—regardless of KRAS mutational status—addition of cetuximab significantly improved PFS (8.9 vs. 8.0 months; HR, 0.85; P = .048) compared with FOLFIRI alone.³⁰ The mutation status of codons 12 and 13 of the KRAS gene was determined for 1063 patients, representing 89% of the study cohort. Wild-type KRAS was found in 666 patients (63%), and mutant KRAS in 397 patients (37%).³¹ In patients with wild-type KRAS, cetuximab plus FOLFIRI significantly improved all efficacy parameters compared with FOLFIRI alone: OS (23.5 vs. 20.0 months; HR, 0.796; P = .0093); PFS (9.9 vs. 8.4 months; HR, 0.696; P = .0012); and ORR (57.3% vs. 39.7%; P < .001).³¹ In contrast, no difference in efficacy was evident between treatment arms in the mutant KRAS subgroup. Grade 3 acne-like rash occurred in approximately 16% of patients in the cetuximab arm; however patients with wild-type KRAS in whom rash developed early during treatment had prolonged OS compared with those who did not (26.4 vs. 19.1 months).³¹ Other adverse events occurring at slightly higher rates in the cetuximab arm included infusion-like reactions and diarrhea.

The activity of anti-EGFR agents with first-line FOLFOX4 was shown in the PRIME trial with panitumumab and in the OPUS trial with cetuximab. In the PRIME trial, 1183 patients with mCRC, regardless of EGFR expression or KRAS status, were randomly assigned to FOLFOX4 with or without panitumumab 6 mg/kg every 2 weeks.³² KRAS status was ascertained in 1096 patients, representing 93% of the study cohort. In the subset of 656 patients (60%) with wild-type KRAS, panitumumab plus FOLFOX4 significantly improved PFS compared with FOLFOX4 alone (9.6 vs. 8.0 months; HR, 0.80; P = .02), and showed trends for improving OS after extended follow-up (23.9 vs. 19.7 months; HR, 0.88; P = .17) and ORR (55% vs. 48%; P = .068).³³ In contrast, the addition of panitumumab had a negative impact on the survival endpoints in the mutant KRAS subgroup. Several grade 3/4 adverse events occurred more frequently in the panitumumab arm, including skin toxicity, diarrhea, hypokalemia, fatigue, mucositis, and hypomagnesemia.³²

In the randomized phase II OPUS trial, KRAS status was determined in 315 patients with EGFR-positive mCRC, representing 93.5% of the study cohort.³⁴ In the wild-type KRAS subset of 179 patients (57%), cetuximab plus FOLFOX4 significantly improved PFS (8.3 vs. 7.2 months; HR, 0.567; P = .0064), and ORR (57% vs. 34%; P = .0027) compared with FOLFOX4 alone.³⁴ Although OS did not differ significantly between treatments (22.8 vs. 18.5 months; HR, 0.855; P = .39), the numerical improvement in median OS was consistent with the findings from the CRYSTAL and PRIME trials.

In contrast to the positive efficacy results in these trials, two phase III trials did not find a survival benefit with addition of cetuximab to first-line oxaliplatin-based therapy. In the MRC COIN trial, 1630 patients were randomized to oxaliplatin-based therapy (either FOLFOX or XELOX at the investigator's discretion) with or without cetuximab. KRAS mutation status was determined in 1316 patients, representing 81% of study cohort; wild-type KRAS was identified in 729 patients (55%). Addition of cetuximab produced increased ORR (59% vs. 50%; P = .015) but did not significantly improve OS (17.0 vs. 17.9 months; HR, 1.038; P = .68) or PFS (both 8.6 months; HR, 0.96; P = .60) in the wild-type KRAS subset.³⁵ However the choice of chemotherapy influenced the effect of cetuximab on PFS (interaction P = .07 in wild-type KRAS subgroup); cetuximab tended to improve PFS when FOLFOX was used (HR, 0.77; P = .056) but not when XELOX was given (HR, 1.06; P = .56).³⁵ One of the likely factors in that difference is the capecitabine dose reductions required to control toxicity in patients receiving the cetuximab/XELOX combination (but not in the XELOX-only group), which may have jeopardized the overall antitumor activity of that cetuximab triplet.

In NORDIC VII, an oxaliplatin bolus/5-FU regimen known as NORDIC FLOX was used either continuously until disease progression or intermittently for 16 weeks and then reintroduced at disease progression.³⁶ A total of 571 patients with mCRC were randomized to receive continuous FLOX with or without cetuximab or intermittent FLOX with continuous weekly cetuximab. KRAS status was determined in 498 patients (87% of the study cohort) and wild-type KRAS was identified in 303 patients (61%). In the KRAS wild-type subgroup, cetuximab plus continuous FLOX did not significantly affect OS (20.1 vs. 22.0; HR, 1.14; P = .66), PFS (7.9 vs. 8.7 months; HR, 1.07; P = .66), or ORR (46% vs. 47%; P = .87) compared with continuous FLOX alone. Comparable activity was seen in patients with wild-type KRAS who received cetuximab plus intermittent FLOX: median OS was 21.4 months and the ORR was 51%. The unexpected finding in NORDIC VII was the greater benefit with cetuximab in the cohort of patients with KRAS mutant tumors. The lack of statistical power for the KRAS subgroup analysis probably accounts for this outlier observation; however further investigation in this group of patients, if possible, could shed additional light on the mechanisms of response/resistance in these molecularly defined groups.

The efficacy of bevacizumab and the anti-EGFR agents in first-line mCRC prompted 2 phase III clinical trials of double monoclonal antibody therapy. In the PACCE study, patients were randomized to receive first-line chemotherapy and bevacizumab with or without panitumumab.³⁷ The cytotoxic regimen (oxaliplatin/5-FU/LV or irinotecan/5-FU/LV) and the doses of the individual agents were selected by the investigator to mimic clinical practice. A total of 1053 patients were randomized to treatment, including 823 patients to the oxaliplatin cohort and 230 patients to the irinotecan cohort. The study was discontinued early after a planned interim analysis showed reduced PFS and increased toxicity in the panitumumab arm. In the final analysis, median PFS and OS were shorter in the panitumumab arm in the entire study cohort as well as in the subset with wild-type KRAS.³⁷ Similarly, in the CAIRO2 trial, the addition of cetuximab to XELOX plus bevacizumab resulted in shorter PFS in the entire study cohort and in the wild-type KRAS subset compared with XELOX plus bevacizumab.³⁸ No difference in OS or ORR was evident between treatment arms. On the basis of these studies, double monoclonal antibody therapy with bevacizumab and an anti-EGFR agent is not recommended.²

The initial characterization of KRAS exon 2 mutations as determinants of lack of benefit from anti-EGFR monoclonal antibodies may be the first rather than the definitive step in the characterization of this marker. All the mutations currently evaluated together as a single entity are likely not equivalent in their predictive effect. In particular, recent reports investigating the G13D mutation (the most frequent single alteration within the group of exon 2 mutations) suggest that this point mutation may have a prognostic effect and may be associated with poorer outcomes regardless of treatment. This G13D mutation, however, may not be specifically predictive of lack of benefit from anti-EGFR therapy, as are the other exon 2 mutations39, 40 as suggested from pooled data obtained in patients receiving first-line and later lines of cetuximab therapy. Pooled data in first-line treatment from the cetuximab trials OPUS and CRYSTAL show an OS HR of 0.80 for this small group of patients, closer to the HR of 0.81 for the KRAS wild-type cohort than to the HR of 1.14 observed in the group including all other KRAS exon 2 mutations.⁴⁰ The challenge in this case will be to generate reliable data with the appropriate statistical power for a patient population that includes approximately 8% of all patients; this issue is the most pressing considering that these patients are currently excluded from therapy and that diagnostic tests to identify them are already commercially available.

Additional biomarkers are desirable to improve selection of which patients with wild-type KRAS tumors are most likely to benefit from anti-EGFR therapy. At present the role of other biomarkers with available data remains dubious. The mutational status of BRAF, a gene that encodes a serine-threonine kinase downstream from KRAS, has received the most attention. The BRAF V600E point mutation encodes an active B-RAF kinase, which triggers downstream signaling and consequently bypasses EGFR regulation. In the correlative analysis of the CRYSTAL trial, BRAF mutations were identified in 59 of 625 patients with wild-type KRAS (9.4%).³¹ Patients with BRAF mutations had poorer survival regardless of whether cetuximab was administered. In the subset with wild-type KRAS/BRAF, the addition of cetuximab improved OS (25.1 vs. 21.6 months; HR, 0.830; P =.0549) and PFS (10.9 vs. 8.8 months; HR, 0.673; P = .0013) compared with FOLFIRI alone. Although the results were not statistically significant due to the small sample size, the addition of cetuximab produced similar numerical benefit in patients with wild-type KRAS/mutant BRAF for OS (14.1 vs. 10.3 months; HR, 0.908; P = .74) and PFS (8.0 vs. 5.6 months; HR, 0.934; P = .87). Comparable results based on BRAF status were obtained in a pooled analysis of the CRYSTAL and OPUS trials.⁴¹ Therefore BRAF mutation is an adverse prognostic factor in mCRC, but it cannot be considered a predictive marker of cetuximab benefit, and consequently should not be used to exclude patients from anti-EGFR therapy. Mutations on another member of the RAS family, NRAS, could also have predictive value, and this potential association was studied in the COIN trial. Results, however, showed no clear affirmation that this marker has a specific predictive effect.³⁵ In addition there is a growing list of markers (including EGFR FISH, mutations on PI3K, PTEN expression) under investigation for their predictive value for anti-EGFR monoclonal antibodies, mostly in exploratory studies in refractory patients; among them, levels of ligand expression (amphiregulin or epiregulin) appeared to be specifically predictive in a correlative study to the phase III NCIC CO.17 trial of third-line cetuximab vs. best supportive care,⁴² but data from first-line trials are lacking.

For the patient presenting with mCRC, one of the first key decisions is whether he or she is a candidate for potentially curative resection or will be managed with a palliative approach. When metastases are limited to the liver, hepatic resection is the only treatment that has been shown to provide long-term survival, and consequently it should be offered to all patients with resectable disease.⁵ In cases that are initially unresectable, first-line chemotherapy may produce sufficient tumor shrinkage to permit complete resection. Accordingly, the goal is to select a regimen offering high response rates, thereby maximizing the likelihood for resection. The quadruplet regimen FOLFIRINOX or FOLFOXIRI has shown high response rates accompanied by promising resection rates (82.4%) in patients with limited but originally unresectable disease⁴³ and seemed to increase R0 rates when compared with FOLFIRI in a subset analysis of a phase III trial enrolling unselected patients with initially unresectable disease (15% vs. 6% overall, 36% vs. 12% in patients with liver-only disease).¹⁹ The use of an anti-EGFR agent with first-line FOLFIRI or FOLFOX has consistently shown higher response rates compared with chemotherapy alone (Table 3), whereas this has not been seen consistently in more recent studies with bevacizumab.³² An important aspect in the treatment planning for patients in whom surgery may be therapeutic goal is to anticipate and minimize the potential liver toxicities of the systemic treatment used for tumor shrinkage. Steatosis (accumulation of lipid vesicles and impairment of liver function) is an effect associated with 5-FU⁴⁴; the addition of oxaliplatin (as in FOLFOX) has been linked to increased risk for sinusoidal obstruction syndrome45, 46, 47, 48; and irinotecan-related steatohepatitis (steatosis with lobular inflammation) is a more severe event that increases both morbidity and mortality after liver resection.⁴⁹ The length of therapy is directly related to the risk for these hepatotoxicities, so it is important to plan for the appropriate number of treatment cycles to elicit a response and avoid prolongation of therapy that could be potentially counterproductive in the context of surgery. Moreover, because the patient will undergo surgery within weeks of completing treatment, it may be prudent to avoid bevacizumab and other antiangiogenic agents because of bleeding and thromboembolic concerns.

In a phase II trial, patients with unresectable metastases or ≥ 5 metastatic sites were randomly assigned to receive cetuximab plus either FOLFOX6 or FOLFIRI, and then a multidisciplinary team reassessed resectability after 16 weeks and every 2 months thereafter.⁵⁰ Confirmed objective responses were achieved in 66 of 106 patients (62%), and subsequently 36 patients (34%) in this cohort underwent R0 resection. Cetuximab has also been investigated in combination with FOLFIRINOX in a small phase II trial with patients who had initially unresectable liver metastases, some of them with extrahepatic disease, for a 60% rate of complete resection.⁵¹ Moreover, data from the CRYSTAL and OPUS trials show that adding cetuximab to first-line chemotherapy increases surgical rates and R0 resection rates (Figure 1).31, 34, 52 In the CRYSTAL trial, surgery for metastases was performed in 7.9% of patients with wild-type KRAS who received cetuximab plus FOLFIRI vs. 4.6% of those who received FOLFIRI alone (ORR, 1.82; P = .063); R0 resection rates were 5.1% vs. 2.0%, respectively (ORR, 2.65; P = .0265).³¹ In the group with liver-limited disease at study entry, R0 resection rates favored the cetuximab arm in the CRYSTAL trial (13.2% vs. 5.6%; ORR, 2.58; P = .125) and in the OPUS trial (16.0% vs. 4.3%; ORR, 4.00; P = .21).⁵² Although these rates appear to be low, it is important to recognize that only 21% of patients had metastases confined to the liver. These numbers highlight the challenges facing studies of optimal strategies for the management of liver metastases: absolute numbers are low in trials that enroll unselected patients, and specialized trials are hard to conduct due to the limited percentage of eligible patients. In patients with mutant KRAS, for whom anti-EGFR therapy is not an option, bevacizumab with chemotherapy may be viable for improving resection rates. In a recent report of 21 patients with nonoptimally resectable liver metastases, bevacizumab plus FOLFOX6 produced high resectability rates without increasing postsurgical complications.⁵³

Symptom relief may be an important goal for highly symptomatic patients and those who have a large disease burden. A more intensive approach that optimizes the likelihood of achieving an objective response may be needed to reduce symptoms, and again treatment with an anti-EGFR agent in combination with a cytotoxic regimen should be considered.

If cetuximab or panitumumab is included in first-line therapy, the choice of partnering a cytotoxic regimen may be important. The results obtained with irinotecan-based therapy have been consistently favorable in first-line and other lines of therapy, whereas mixed data have emerged from trials using oxaliplatin as a backbone. It is unclear whether this discrepancy is related to the different properties of irinotecan and oxaliplatin. Interestingly, no similar dichotomy regarding irinotecan vs. oxaliplatin as a partner has arisen for bevacizumab, since major trials with first-line irinotecan (AVF 2107) and oxaliplatin (NO16966) have been both positive.

It does appear, however, that the choice of fluoropyrimidine may be an important factor in the overall efficacy of these complex regimens. The BICC-C study showed that bevacizumab plus FOLFIRI outperformed bevacizumab plus IFL.¹¹ In the clinical trials in which the anti-EGFR agents had a beneficial effect on outcome in patients with wild-type KRAS, they were administered with a first-line regimen containing infusional 5-FU/LV (with FOLFIRI in the CRYSTAL trial, with FOLFOX4 in the PRIME and OPUS trials, and with modified FOLFOX in the subset analysis from the MRC COIN). In contrast, in the studies where no benefit was seen, the chemotherapy arm included either capecitabine (XELOX in the other subset from MRC COIN trial) or bolus 5-FU (NORDIC FLOX in the NORDIC VII trial). It is not clear why capecitabine or the way in which 5-FU is administered should influence the benefit of anti-EGFR therapy. However, the greater toxicity of bolus 5-FU vs. infusional 5-FU and the toxicity seen with the capecitabine doses used in XELOX may be contributing factors.

Another practical issue to consider in certain treatment choices is that of schedule convenience. Bevacizumab administration can be easily synchronized with most complex chemotherapy regimens and so can panitumumab (biweekly schedule for both); in the case of cetuximab, standard administration is weekly, although a growing number of studies have used double-dose biweekly schedules without jeopardizing efficacy.54, 55

FOLFIRI and FOLFOX appear to be roughly equivalent, and outcome does not appear to be affected depending on which one is given first.¹⁴ It could make more sense to use FOLFIRI initially because first-line regimens tend to be given longer than second-line regimens, and therefore oxaliplatin toxicity, specifically neuropathy, may be more manageable in the second-line setting. Nevertheless, oxaliplatin regimens are more commonly used first in the United States, whether due to the habit formed after the quick adoption by N9741 when the irinotecan choice was based on IFL, or other reasons.

The choice among first-line regimens may be influenced by previous adjuvant therapy. Key factors include the time interval between adjuvant therapy and disease recurrence, how well the patient tolerated the adjuvant regimen, and the presence of any residual toxicity. Patients who progress within 6-12 months after adjuvant therapy are unlikely to benefit from the same regimen. For patients who progress after more than 1 year, it is open to debate whether the same regimen should be resumed or an alternative regimen started. Patients who received oxaliplatin in the adjuvant setting may have residual neuropathy.⁵⁶ Although the neuropathy tends to improve slowly over time, it recurs rather quickly once oxaliplatin is restarted. Therefore in patients with residual neuropathy, it is prudent to start mCRC treatment with a different regimen.

Several other factors should be considered when deciding between an intensive vs. nonintensive management strategy, including functional status, comorbidities, and the patient's own desires. Significant quality of life issues may be problematic with complex regimens, including neurotoxicity with oxaliplatin, rash with anti-EGFR agents, and hypertension with bevacizumab. Accordingly, a sequential approach using single-agent therapy in the first-line setting may be appropriate for patients who will not be surgical candidates irrespective of treatment response. The CAIRO trial showed that survival did not differ with first-line capecitabine, second-line irinotecan, and third-line XELOX as compared with first-line XELIRI and second-line XELOX.⁵⁷ Similarly, the MRC FOCUS (Medical Research Council Fluorouracil, Oxaliplatin and Irinotecan: Use and Sequencing) trial showed that first-line therapy with 5-FU can be administered before subsequent combination therapy without compromising outcome.⁵⁸ For asymptomatic patients, it may be possible to delay chemotherapy until significant symptoms develop. In a meta-analysis of two small trials, delaying chemotherapy until development of predefined symptoms did not adversely impact survival or quality of life compared with early initiation of chemotherapy.⁵⁹ These analyses have to be layered with the data showing that delaying bevacizumab may have an impact on OS.60, 61 Therefore the better option for less intensive therapy may be a fluoropyrimidine plus bevacizumab. If performance status is a concern, it is worth noting, however, that data specifically generated in patients with compromised performance status is lacking; the recent subset analysis from PRIME (with FOLFOX4 and panitumumab) would suggest that intensive therapy is counterproductive in these patients, but the limited nature of this analysis precludes any implication ready for practice.⁶²

Intermittent strategies are being evaluated in an effort to minimize toxicity without compromising efficacy. In the OPTIMOX1 trial, 620 patients with mCRC were randomized to FOLFOX4 or alternatively to an oxaliplatin stop-and-go strategy, in which FOLFOX7 (high-dose oxaliplatin and 5FU infusion) was given for 6 cycles followed by maintenance therapy with infusional 5-FU/LV (without oxaliplatin) for 12 cycles, and finally reintroduction of FOLFOX7.⁶³ Median PFS (9.0 vs. 8.7 months; HR, 1.06; P = .47) and OS (19.3 vs. 21.2 months; HR, 0.93; P = .49) did not differ between treatments, but grade 3/4 toxicities, including sensory neuropathy, occurred less often in the experimental arm. In a subsequent study, chemotherapy was completely discontinued after 6 cycles of FOLFOX7, but it had a negative impact on the duration of disease control and on PFS.⁶⁴ However in two trials conducted by the Medical Research Council in the UK, the one by Maughan et al in 2003 and the COIN trial (which, in addition to investigating the combination of cetuximab with oxaliplatin-based therapy, also included a third arm in which chemotherapy was given intermittently every 12 weeks until the next progression without maintenance) showed that there did not seem to be meaningful OS detriment to full treatment-free intervals (Table 436, 63, 64, 65, 66, 67).65, 66

It therefore appears that interruptions may not necessarily compromise efficacy. In clinical practice, oncologists may not proactively plan an intermittent strategy but may be compelled to do so once toxicity develops, particularly neurotoxicity or hand-foot syndrome. The results given earlier indicate that it may be feasible to step down from a complex regimen to a simpler single-drug strategy when toxicity is problematic provided that some form of maintenance therapy is offered. Appropriate maintenance regimens can include a fluoropyrimidine and/or a targeted agent. In the MACRO (Maintenance in Colorectal Cancer) trial, maintenance bevacizumab started after 6 cycles of XELOX plus bevacizumab was not inferior to continuing the same combination regimen until progression (median PFS, 9.7 vs. 10.4 months; HR, 1.11; median OS, 21.7 vs. 23.4 months; HR, 1.04).⁶⁷ The maintenance bevacizumab arm had less grade 3/4 paresthesia, hand-foot syndrome, and fatigue but more diarrhea and hypertension. Along similar lines, NORDIC VII (another trial with a third, intermittent chemotherapy plus cetuximab maintenance arm in addition to the arms comparing the addition of cetuximab to FLOX), showed similar OS with the administration of continuous chemotherapy with or without cetuximab and the administration of intermittent chemotherapy, with breaks every 16 weeks until next progression, accompanied by continuous cetuximab.³⁶

Although intermittent regimens may be useful to manage toxicities associated with chemotherapy, certain toxicities related to targeted therapies may also play a role in treatment decisions. EGFR inhibitors cause rash in the vast majority of patients; this toxicity is rarely the cause of clinically relevant complications, but it has a particularly strong impact on patients' self-perception and quality of life. Management protocols, many of them based on simple skin care interventions are available,⁶⁸ and prophylactic interventions may help ameliorate the symptoms of rash.⁶⁹ In the case of bevacizumab, hypertension and venous or arterial thromboembolic events (VTEs or ATEs) constitute probably the most relevant toxicities in routine clinical practice. Pooled analyses, however, indicate that once the variability on treatment exposure is taken into account, bevacizumab does not specifically increase the risk for VTEs in patients with mCRC (N = 2573), and the risk for ATEs, although numerically greater with bevacizumab, can be considered acceptable.70, 71

Given the average age at diagnosis and the growing survival times in patients with mCRC, the elderly may be becoming a population of special interest. Elderly patients benefit from cytotoxic chemotherapy (5-FU, capecitabine, and oxaliplatin or irinotecan) to a degree similar to that in younger patients, although the risk for certain toxicities—particularly neutropenia, oxaliplatin-related neurotoxicity, and irinotecan-related diarrhea—may be greater for them.⁷²

Age does not seem to preclude benefit from first-line targeted therapies, as shown in most pivotal trials of bevacizumab20, 22 or cetuximab;30, 34, 35 in the case of panitumumab, the subgroup > 65 years in the PRIME trial appeared to benefit less, but the retrospective nature of the analysis and the lack of similar observations in other panitumumab studies makes this an inconclusive finding.³² Neither cetuximab nor panitumumab poses age-specific safety problems, and extensive data from bevacizumab registries and observational cohort studies have also shown a consistently tolerable profile, regardless of age.73, 74, 75 These data would therefore support treatment approaches for fit elderly patients similar to that for their younger counterparts and, in general, management strategies that are tailored to the performance or clinical status of individual patients rather than considering age a treatment determinant in itself.