Introduction

Chemotherapy for colorectal cancer (CRC) was essentially limited to the antimetabolite 5-fluorouracil from the development of this drug in the late 1950s through the mid 1990s. Over the past 15 years, six additional drugs (irinotecan, oxaliplatin, capecitabine, cetuximab, bevacizumab, and panitumumab) have received US regulatory approval for the management of this disease. While each of these agents has provided some benefit to some patients, overall improvements in patient outcomes have tended to be modest, and toxicities of these newer agents can at times be considerable. Proper treatment of patients with metastatic CRC today therefore requires careful consideration of the risks and benefits of all available agents, consideration of the merits of combination versus sequential single-agent regimens, and development of individual strategies to optimize treatment for each patient. This article reviews and discusses recent data on chemotherapy plus targeted agents and molecular markers in CRC.

Cytotoxic Agents

Despite years of attempts to develop a drug better than the fluoropyrimidine 5FU, this agent remains the focus of most CRC treatment strategies. Capecitabine, an oral prodrug that is converted to 5FU, has similar efficacy to the parent compound. Most colorectal treatment paradigms utilize combinations of infusional 5FU with either oxaliplatin (FOLFOX) or irinotecan (FOLFIRI). Some clinicians prefer to substitute capecitabine for infusional 5FU, and available data would suggest that this is an acceptable option with oxaliplatin, while the data with irinotecan are less supportive. The two regimens, FOLFOX and FOLFIRI, have been shown to have similar efficacy in first-line treatment of metastatic CRC [1, 2]. In the first randomized study, 109 patients were allocated to FOLFIRI and 111 patients received FOLFOX followed by planned crossover to the other regimen at the time of progression. The median survival was 21.5 months versus 20.6 months (P = 0.99), respectively, and the median second progression-free survival was 14.2 months versus 10.9 months (P = 0.64). Frontline FOLFOX achieved a 54% response rate while FOLFIRI achieved a 56% response rate [1]. In the second large study, 360 treatment-naïve patients were randomly assigned either FOLFIRI or FOLFOX. The median overall survival was 14 months for FOLFIRI versus 15 months for FOLFOX (P = 0.28), and the response rates were 34% and 36%, respectively (P = 0.60) [2]. The most notable differences between regimens in both studies were in the toxicity profiles wherein grade 3 and 4 mucositis, nausea/vomiting, and grade 2 alopecia were more common with FOLFIRI, and grade 3 and 4 neutropenia and neurotoxicity were more frequent with FOLFOX. Thus, the choice of the initial treatment regimen is largely based on the toxicity profile.

Of note, oxaliplatin has minimal activity as a single agent, and so if FOLFIRI is used first, then FOLFOX will be needed in a subsequent line. The same is not true of irinotecan, however, and the common practice of following FOLFOX by FOLFIRI is not supported by data. Use of single-agent irinotecan as a second-line treatment is therefore an acceptable alternative.

Capecitabine has been extensively compared to 5FU in combination treatment regimens in randomized phase 3 trials, and several studies have demonstrated the noninferiority of CapeOx compared to FOLFOX [3–5]. The combination of capecitabine with irinotecan (CAPIRI), however, has been less successful. In European Organization for Research and Treatment of Cancer study 40015, enrollment was suspended after 85 patients due to toxicities including grade 3 and 4 diarrhea (37% vs 13% in CAPIRI vs FOLFIRI, respectively), as well as six treatment-related deaths with CAPIRI versus two in the FOLFIRI arm. The progression-free survival (PFS) and overall survival were worse with CAPIRI than with FOLFIRI [6]. In the BICC-C study, CAPIRI was associated with worse PFS compared with FOLFIRI, as well as higher rates of toxicities including vomiting and diarrhea [7]. As such, it is difficult to support the routine use of the CAPIRI regimen.

Sequential Versus Combination Chemotherapy in First-Line Treatment

Several trials have attempted to determine whether upfront combination chemotherapy offers a clear advantage over administering the agents sequentially. One study known as the FOCUS trial randomly assigned 2,135 patients to three potential treatment arms: 1) sequential single-agent 5-FU/leucovorin followed by single-agent irinotecan; 2) single-agent 5-FU/leucovorin followed by combinations with either FOLFOX or FOLFIRI; or 3) first-line combination therapy with FOLFIRI or FOLFOX and the opposite regimen upon disease progression. There were no significant survival differences between the three arms [8].

Another study, known as the CAIRO trial, randomly allocated patients to sequential capecitabine followed by irinotecan, followed by CAPOX, versus first-line CAPIRI followed by second-line CAPOX. The combination-treatment arm demonstrated an improvement in PFS and overall response rate versus first-line capecitabine alone; however, use of the concurrent versus sequential strategies did not significantly impact overall survival [9].

Duration of Therapy

With the improvement of PFS seen with combination regimens, the question of duration of treatment has been raised. The first study to address the issue of planned interruption of a working treatment was the OPTIMOX study. This trial randomized 620 patients to FOLFOX4 until disease progression versus FOLFOX7 (higher-dose oxaliplatin and omission of bolus 5FU) for 12 weeks, followed by planned interruption of the oxaliplatin, and continued treatment with 5FU/leucovorin alone, with planned re-introduction of oxaliplatin, either at 6 months, or at time of disease progression, which ever came sooner [10]. In this trial, there were no differences in response rates, durations of disease control. or overall survival, validating the “stop and go” approach. The incidence of neurotoxicity was significantly reduced in the “stop and go” arm compared with the continuous FOLFOX arm. Subsequently, the OPTIMOX-2 study investigated early planned interruption of all treatment, with the OPTIMOX-1 arm compared to full cessation of all treatment at 3 months, followed by reintroduction of treatment upon progression. The fully interrupted arm had an inferior overall survival, suggesting that early planned stopping of all therapy in all patients is not optimal [11]. However, data do support that patients enjoying a strong early response that is durable beyond 6 months may be good candidates for later full chemotherapy interruptions, or “holidays.”

Anti-Vascular Endothelial Growth Factor Therapy in Metastatic CRC

Bevacizumab is a monoclonal antibody that binds to vascular endothelial growth factor (VEGF). While the initial expectation was that this agent would function as an anti-angiogenic agent, the absence of meaningful single-agent activity, and the potentiation, rather than interference with, cytotoxic agents, argues against this agent working via anti-angiogenesis. Thus, while the target of bevacizumab is clear, the mechanism of antitumor activity remains the subject of some debate. One hypothesis is that bevacizumab impacts vascular flow and/or permeability, thus facilitating delivery of chemotherapy to the tumor.

The contribution of bevacizumab to front-line chemotherapy in metastatic CRC was first definitively demonstrated in a phase 3 trial reported by Hurwitz et al. [12], in which patients received front-line bolus 5FU, irinotecan and leucovorin (IFL), the standard regimen at the time, and then were randomized to concurrent bevacizumab or placebo. Median overall survival was increased by 4.7 months (P < 0.001) in the IFL plus bevacizumab arm. Improvements in response rate and PFS were also statistically significant [12]. The most serious side effects noted were the risk of gastrointestinal perforation (1.5%) and a small increase in arterial thrombotic events over the rate seen with chemotherapy alone. Hypertension was also increased, with 11% of patients developing grade 3 hypertension.

By the time the Hurwitz trial was reported, however, IFL was no longer routinely used, and FOLFOX (oxaliplatin/leucovorin/5FU) had become the most commonly used standard front-line regimen. As a result, the combination of FOLFOX plus bevacizumab was widely adopted as a standard practice, and despite the lack of direct supporting evidence, this became the most commonly used first-line regimen in the United States. Subsequently, the large phase 3 trial, NO16966, was expanded to evaluate bevacizumab in the front-line setting with either FOLFOX or CAPOX (capecitabine/oxaliplatin) in a placebo-controlled 2 × 2 design [13•]. A total of 1,400 patients were randomly assigned to CAPOX versus FOLFOX and then to bevacizumab versus placebo. While the improvement in PFS with bevacizumab to front-line oxaliplatin-based chemotherapy was statistically significant, this improvement was only 1.4 months, as compared with the 4.4-month median PFS improvement in the initial IFL trial discussed above. Furthermore, in this oxaliplatin-based trial, the overall survival was also only improved by 1.4 months, with the P value approaching but not reaching statistical significance. In addition, the response rates were absolutely identical in the bevacizumab-containing and non-bevacizumab-containing arms.

There have been no randomized trials reported to date comparing FOLFOX-bevacizumab with FOLFIRI-bevacizumab, and none are likely to be performed. An extrapolation from the available data, however, would suggest that the choice of front-line treatment with either FOLFOX-bevacizumab or FOLFIRI-bevacizumab is acceptable, and the choice between the two treatment regimens can be made based on the patient’s preference of one side effect profile versus the other. The choice of oxaliplatin without a biologic is also an acceptable consideration, especially in patients who have a relative contraindication to bevacizumab, as the contribution of bevacizumab does not appear to be as substantial with oxaliplatin as with irinotecan. It should be noted that in the United States at the time of this writing, irinotecan is available as a generic agent, while oxaliplatin is not; thus, FOLFIRI is considerably less expensive than FOLFOX at the present time.

Other Anti-VEGF Agents

It should be noted that thus far no anti-VEGF agent other than bevacizumab has shown efficacy in CRC. Sunitinib, a VEGF tyrosine kinase inhibitor, failed to show single-agent activity in a cohort of chemotherapy-refractory CRC patients [14]. Recently, a wire service reported that a phase 3 trial of FOLFIRI plus/minus sunitinib had been closed by the data safety monitoring board for futility. Thus, at this time, there is no justification for the routine use of commercially available VEGF tyrosine kinase inhibitors in CRC.

Anti-Epidermal Growth Factor Receptor Therapy in Metastatic CRC

Cetuximab and panitumumab are monoclonal antibodies that block the ligand-binding site of the epidermal growth factor receptor (EGFR), thereby blocking this intracellular signaling [15–17]. Panitumumab is a fully human anti-EGFR (IgG2), and cetuximab is a chimeric human–mouse IgG1 monoclonal antibody.

The first phase 2 trial was conducted in the United States, and combined cetuximab with irinotecan in 120 patients with metastatic CRC refractory to both 5FU and irinotecan and reported a response rate of 22.5% [18]. A second US trial showed a response rate of 10.5% in 57 patients treated with single-agent cetuximab [19].

A confirmatory trial, designated the BOND trial, enrolled 329 patients with metastatic CRC who were previously treated with irinotecan and oxaliplatin [20]. Patients were randomly assigned to continue irinotecan and add cetuximab versus receive cetuximab alone. The results of the BOND trial were nearly identical to the two previous US studies, with a 22.9% response rate for the combination of cetuximab plus irinotecan, and a 10.8% response rate for single-agent cetuximab. The irinotecan plus cetuximab combination produced a median time to progression of 4.1 months, compared with 1.5 months with cetuximab alone. While this apparent reversal of irinotecan resistance is not well understood, preclinical evidence indicates that this action may, at least in part, result from the ability of blockade of EGFR signal to impair activation of anti-apoptotic pathways such as BCL-2 [21].

Panitumumab has also demonstrated similar anti tumor activity, with phase 2 trials reporting response rates in the range of 8% to 10% [22, 23]. A phase 3 trial that randomized patients with chemotherapy-refractory CRC to either panitumumab plus best supportive care or best supportive care alone found that panitumumab modestly improved PFS. There was little change in the median PFS, with panitumumab resulting in a median of 8 weeks, versus 7.3 weeks for best supportive care alone. The overall response rate to panitumumab was 10% [24].

Anti-EGFR Therapy in Front-Line Regimens in Metastatic CRC

The results of anti-EGFR agents in the refractory setting led to enthusiasm for exploring their incorporation into first-line regimens. The CRYSTAL trial randomized 1,217 patients to front-line FOLFIRI plus or minus cetuximab. The results of this trial, while technically positive, were overall disappointing, with an increase in PFS of only 27 days. This very modest improvement did achieve statistical significance (P = 0.048), albeit with increased toxicities, including increased grade 3/4 diarrhea and skin rash [25].

The OPUS trial, a randomized phase 2 study, evaluated the addition of cetuximab to front-line FOLFOX in 337 patients [26]. The overall response rate was 45.6% in the cetuximab-containing arm, versus 35.7% with FOLFOX alone. The PFS for the overall group was not statistically significantly improved. The side effect profile was similar to previously mentioned studies, with the most common grade 3/4 adverse effects being neutropenia, diarrhea, leukopenia, and skin rash (in the cetuximab-containing arm only).

Concurrent Anti-VEGF Plus Anti-EGFR Therapies in Metastatic CRC

Preclinical models have demonstrated synergy between anti-EGFR and anti-VEGF agents [27]. A National Cancer Institute-sponsored randomized phase 2 trial, known as the BOND 2 trial, evaluated the feasibility, safety, and efficacy of concurrent administration of cetuximab and bevacizumab in patients with irinotecan-refractory stage IV CRC, who were naïve to anti-VEGF and anti-EGFR therapy [28]. Treatment with the two antibodies alone yielded a 20% response rate, while the use of the two antibodies plus irinotecan resulted in a 37% response rate. The toxicity pattern of the combination of these agents was similar to that that would have been expected from the single agents. The results of the trial indicated that it was reasonable from a safety standpoint to proceed with larger-scale trials of combined anti-EGFR and anti-VEGF therapy. Several large trials were therefore undertaken to evaluate the combination of anti-EGFR and anti-VEGF monoclonal antibodies with front-line chemotherapy.

The PACCE trial was the first large-scale clinical trial to report the results of front-line use of chemotherapy plus combined anti-EGFR and anti-VEGF agents [29]. In this trial, 823 patients were assigned to FOLFOX-bevacizumab and 200 patients to FOLFIRI-bevacizumab, and then these patients were randomized to either receive or not receive concurrent panitumumab as well. The trial completed planned accrual, but was then terminated after a pre-planned efficacy analysis at 231 events showed a shorter PFS for the panitumumab-containing arm when compared with the control. The most recently reported, updated results show median PFS in the FOLFOX-bevacizumab plus panitumumab arm was 9.5 months, versus 11 months in the FOLFOX-bevacizumab control. The median overall survival was trivially (15 days) shorter in the panitumumab-containing arm. No statistically significant response benefit was seen with the addition of panitumumab. The incidence of grade 3/4 adverse events was also greater in the panitumumab arm. These included skin toxicity (39% vs 2%), diarrhea (24% vs 13%), and infection (19% vs 10%).

More recently, the CAIRO-2 trial, a phase 3 study, evaluated the first-line use of capecitabine, oxaliplatin, and bevacizumab (CAPEOX-BEV) with or without cetuximab in 730 metastatic CRC patients [30]. The results of this trial, unfortunately, confirmed the negative results of the PACCE trial. Of note, not only did the addition of cetuximab not improve outcome, but there was actually a mild detriment in outcome with the addition of cetuximab. The median PFS with the addition of cetuximab was 9.8 months versus 10.7 months for CAPEOX-BEV alone, while the response rates were 40.6% and 43.9%, respectively. There was no statistically significant difference in median overall survival.

Finally, the CALGB/SWOG 80405 trial is currently ongoing. This trial was designed to allow treating physicians to elect to use either FOLFOX or FOLFIRI, and then patients are randomly assigned to receive concurrent cetuximab, bevacizumab, or the combination of cetuximab plus bevacizumab. Enrollment in this trial is continuing for patients with KRAS wild-type tumors only (see below). This trial is being monitored regularly by the Data Safety Monitoring Board. Given the negative findings of the PACCE and CAIRO-2 trials, front-line use of combined anti-VEGF and anti-EGFR antibody therapy with chemotherapy cannot be recommended, and outside of a clinical trial, should not be done.

Predictive Biomarkers for Anti-EGFR Therapy

KRAS

KRAS is a key protein involved in the signaling cascade downstream of the EGFR on the cell surface to the nucleus. Recent data have indicated that mutations in exon 2 of KRAS confer resistance to cetuximab and panitumumab [31, 32]. Amado et al. [33•] examined 427 tumor specimens of patients treated with panitumumab monotherapy for KRAS status. The results were analogous to the cetuximab data.

Data from the CRYSTAL, OPUS, and CAIRO 2 trials have also substantiated these findings. In the CRYSTAL trial, KRAS mutational status was analyzed for 540 patients. The study was reanalyzed for treatment efficacy based on stratification by KRAS wild-type and KRAS-mutated tumors. Correlative analysis showed a statistically significant difference in favor of the cetuximab-containing arm in the 348 KRAS wild-type patients. There was a modest improvement in overall response (59% vs 43%, P = 0.0025) and PFS (9.9 months vs 8.7 months, P = 0.017), compared with patients receiving FOLFIRI alone [34]. There was no significant difference in PFS or response in the 192 patients with mutated KRAS, and PFS trended toward inferior with the addition of cetuximab (Table 1). The benefit of the addition of front-line cetuximab was completely limited to the KRAS wild-type population; this benefit, even in this selected population, was at best, modest. The addition of cetuximab to front-line FOLFIRI in the KRAS wild-type patients only led to an improvement in median PFS of 1.2 months, or approximately 37 days.

Table 1 The OPUS and CRYSTAL trials
Full size table

An analysis of the OPUS trial for KRAS mutation has also been reported. Of the 337 patients enrolled in the initial trial, 233 tumor specimens were analyzed for KRAS mutation status [35]. The addition of cetuximab to FOLFOX4 showed a benefit in patients with wild-type KRAS in response rate (61% vs 37%, P = 0.0163), with a minimal improvement in PFS from 7.2 to 7.7 months. Not only was there no benefit in KRAS-mutated patients, there appeared to be a detriment when cetuximab was added. The addition of cetuximab led to a decrease in response rate from 49% to 33% and a decrease in PFS from 8.6 months to 5.5 months (P = 0.02).

In the CAIRO-2 trial, of the 528 patients who had tissue available for KRAS analysis, the subgroup who were KRAS wild-type showed no difference in PFS with or without the addition of cetuximab to capecitabine, oxaliplatin, and bevacizumab, while the subgroup who were KRAS-mutated again showed a detriment in PFS with the addition of cetuximab (10.7 months vs 9.4 months, P = −0.01) [30].

The data from these studies support the concept that KRAS testing should be done routinely on tumors of all stage IV CRC patients, and that use of cetuximab or panitumumab should be restricted to patients with wild-type KRAS tumors. Unfortunately, therapeutic options beyond fluoropyrimidines, irinotecan, and oxaliplatin for patients with KRAS-mutated tumors do not yet exist. This does not justify exposure of these patients to the risk and expense of anti-EGFR agents, given the futility of such a maneuver and the potential for harm.

Furthermore, the conclusions that can be drawn from the CRYSTAL, CAIRO, and OPUS studies discussed above are that although there is a statistically significant benefit in response rate, the overall clinical benefits appear modest at best and the morbidity associated with the acneform skin rash must be considered.

BRAF and PIK3CA

Although KRAS mutations have been shown to incur resistance to anti-EGFR therapies, they account for only 30% to 40% of nonresponsive patients. More recent data have implicated the serine-threonine kinase, BRAF V600E mutation in resistance to EGFR-targeted therapy. BRAF is the principal downstream effector of KRAS and is mutated in approximately 10% of colorectal carcinomas. Interestingly, the KRAS and BRAF mutations are mutually exclusive. If a tumor is KRAS-mutated then it is BRAF wild-type and vice versa. In a retrospective analysis by Di Nicolantonio et al. [36•], of 113 patients with metastatic CRC who received cetuximab or panitumumab, none of the BRAF patients responded to treatment and none of the responders had BRAF-mutated tumors. In the study, the response rate of patients with KRAS and BRAF wild-type tumors was 32%, which concurred with previous studies. The patients who were wild-type for both mutations also had a longer PFS (P = 0.001) and overall survival (P < 0.001) [36•]. These findings are important in the treatment decision making process, as potentially more than 10% of KRAS wild-type patients will have mutated BRAF, thus rendering them ineligible for EGFR-targeted therapies, as their probability of benefitting is low and treatment would only expose them to unnecessary toxicities.

Recent studies have also demonstrated resistance to anti-EGFR agents in patients with mutations in PIK3CA. The PIK3CA gene encodes a lipid kinase that also regulates the signaling cascade downstream of the EGFR receptor. The frequency of PIK3CA-mutated colorectal tumors is approximately 15% to 20%. In a recent retrospective analysis of 110 patients treated with either cetuximab or panitumumab, 15 PIK3CA mutations were observed and all were associated with clinical resistance to anti-EGFR therapy. Furthermore, the mutations were consistent with a worse clinical outcome in terms of PFS (P = 0.035) [37].

Conclusions

The pathways of cell signaling that promote CRC tumor growth have proven to be more complicated than was initially believed. There are multiple redundancies that ensure tumor proliferation when a targeted agent blocks a single pathway. The combination of anti-VEGF therapy with traditional cytotoxic therapies has shown modest improvement in PFS in metastatic CRC; however, no such benefit has been demonstrated in the adjuvant setting to date.

The knowledge that treatment of patients with mutated KRAS with anti-EGFR agents does not benefit, but rather adds detriment in terms of toxicities, represents an important step forward in our understanding of targeted agents. Screening for KRAS mutations should be performed routinely in all stage IV CRC patients prior to initiating anti-EGFR therapy, and patients with KRAS-mutated tumors should not receive these agents at any point in their care. Likewise, new data have shown that patients with wild-type KRAS and mutated BRAF may not benefit either from anti-EGFR therapy. As has been demonstrated by the markers KRAS and BRAF in the EGFR pathway, new markers delineating which patients would benefit from certain therapies are crucial to improving the progress and outcome of CRC treatment.