Colon Cancer, Version 2.2018

Risk Stratification and Adjuvant Treatment for Stage III Colon Cancer

Nonmetastatic colon cancer is generally treated with curative intent by colectomy and, in some cases, adjuvant chemotherapy.¹⁰ Population and institutional studies have shown that adjuvant therapy may confer a survival advantage in some patients with resected colon cancer (eg, those with stage III or high-risk stage II disease).^11–14 Furthermore, randomized controlled trials have shown that the addition of oxaliplatin to these adjuvant regimens benefits some patients.^15–19 However, adjuvant treatment, especially with regimens containing oxaliplatin, is associated with considerable toxicity (notably chemotherapy-induced peripheral neuropathy),^16,20 and not all patients derive benefit. Consideration of disease stage and pathologic features, microsatellite instability (MSI) status, possible efficacy and toxicity profiles associated with treatment choice, and patient age, comorbidities, and preferences aid in decision-making regarding the use of adjuvant therapy for patients

The IDEA collaboration investigated whether a shortened duration of adjuvant therapy would be a feasible way to avoid or lessen toxicities associated with oxaliplatin-containing adjuvant therapy in some patients with locoregional colon cancer, without impairing oncologic outcomes. IDEA included >12,000 patients in an international effort that pooled data from 6 concurrently conducted randomized phase III trials to assess the noninferiority of 3 months compared with 6 months of adjuvant FOLFOX or CapeOX in patients with stage III colon cancer.⁸ Median follow-up was 39 months. Importantly, grade ≥3 neurotoxicity rates were lower in the 3 months’ versus 6 months’ treatment arms (3% vs 16% for FOLFOX; 3% vs 9% for CapeOX; P<.0001), as were grade 2 neurotoxicity rates (14% vs 32% for FOLFOX; 12% vs 36% for CapeOX; P<.0001). Rates of grades 2 and 3/4 diarrhea were also lower with the shorter duration of therapy (P<.0001 for FOLFOX; P=.01 for CapeOX). The primary end point of 3-year disease-free survival (DFS) did not meet the pre-specified cutoff for noninferiority, despite the small absolute difference of 0.9% (74.6% for 3 months vs 75.5% for 6 months; hazard ratio [HR], 1.07; 95% CI, 1.00–1.15), which is of questionable clinical significance. Notable differences in 3-year DFS were seen between FOLFOX and CapeOX and between patients with T1–3N1 (low-risk) versus T4 or N2 (high-risk) disease. Specifically, in the T1–3N1 subset, DFS with 3 months of CapeOX was noninferior to that with 6 months of CapeOX (HR, 0.85; 95% CI, 0.71–1.01), whereas noninferiority could not be proven for 3 versus 6 months of FOLFOX (HR, 1.10; 95% CI, 0.96–1.26). In the T4 or N2 subset, DFS with 3 months of FOLFOX was inferior to that of 6 months with FOLFOX (HR, 1.20; 95% CI, 1.07– 1.35), whereas noninferiority could not be proven for 3 versus 6 months of CapeOX (HR, 1.02; 95% CI, 0.89–1.17).

The panel discussed several details of the trial and how they believed the results should be translated into guideline recommendations. First, the panel discussed what might account for the unexpected differences seen in DFS between FOLFOX and CapeOX. Oxaliplatin doses are different between the regimens (85 mg/m² every 2 weeks for mFOLFOX6; 130 mg/m² every 3 weeks for CapeOX), which might partially account for the observed difference in outcomes. Although total oxaliplatin exposure is similar, more oxaliplatin is received in the first 4 weeks with CapeOX (260 mg/m²) compared with FOLFOX (170 mg/m²). Also, the panel noted that capecitabine may result in more continuous fluorouracil exposure than 5-FU/leucovorin. One panel member noted that some sites used FOLFOX4; however, panel consensus was that this difference would only account for differences in toxicity, not in efficacy. Treatment compliance was similar between the CapeOX and FOLFOX arms (percent reaching planned last cycle was 90% for 3-month FOLFOX, 86% for 3-month CapeOX, 71% for 6-month FOLFOX, and 65% for 6-month CapeOX), and therefore compliance is unlikely to account for the differences between the regimens. Importantly, the panel noted that no significant differences have been seen between CapeOX and FOLFOX in randomized adjuvant or metastatic studies.^26,27 Therefore, the panel discussed the possibility that selection bias may account for the difference in DFS seen between FOLFOX and CapeOX in IDEA. The use of FOLFOX versus CapeOX was by physician’s choice, not by randomization; choice of regimen was used as a stratification factor for randomization to 3 or 6 months of treatment duration. The proportion of patients treated with CapeOX varied from 0% to 75% between trial sites.⁸ In the US/Canadian study that used only mFOLFOX6, in fact, 3 months of FOLFOX achieved noninferiority to 6 months of FOLFOX, with an HR for DFS of 1.10. However, in the French trial,⁸ most patients (90%) received FOLFOX, and the HR for DFS between 3 and 6 months of FOLFOX was the worst of all trial sites at 1.27. Though the reasons for the different outcomes based on regimen choice are not completely clear, the panel agreed that the data are convincing and that the difference cannot be ignored.

Another point that the NCCN Panel discussed was whether DFS data formed a sufficient basis for guideline recommendations. The panel was unified in its opinion that, although DFS effects have been shown to be larger than overall survival (OS) effects in other trials, the almost complete elimination of oxaliplatin toxicity with a shorter treatment duration justified changes to the guidelines at this time.

The data were sufficient to warrant division of stage III colon cancer into a low-risk (T1-3N1) and a high-risk group (T4N1-2 or TanyN2; see COL-3, page 361. For the low-risk group, the recommended duration of adjuvant therapy is 3 months if CapeOX is chosen, because noninferiority for DFS was proven in IDEA and patients should be spared the increased toxicity, cost, and inconvenience of longer therapy. If FOLFOX is chosen for these patients, then 3 to 6 months of FOLFOX is recommended. The panel believes the shorter duration can be considered to reduce toxicity, cost, and inconvenience, but noninferiority could not be proven for 3 versus 6 months of FOLFOX in this subset. For patients with high-risk stage III colon cancer, 3 to 6 months of adjuvant therapy is recommended if CapeOX is chosen; noninferiority of the shorter duration was not proven in this subset of patients, but the panel believes the shorter duration can be considered to minimize toxicity, cost, and inconvenience. Furthermore, DFS with 3 months of therapy was noninferior to DFS with 6 months for the entire cohort (TanyN1-2) who received CapeOX (HR, 0.95; 95% CI, 0.85–1.06).⁸ Finally, FOLFOX, if chosen, should be given for a full 6 months in the high-risk group because 3 months was shown to be inferior to 6 months for DFS. With these updated and nuanced recommendations, careful discussion and review of the available data between the patient and clinician are important to guide the choice of approach.

The panel then discussed some concerns about the dosing of capecitabine in CapeOX, which they believe is critical if the decision is to give 3 months of CapeOX. Many oncologists, especially in the United States, routinely reduce the starting dose of capecitabine, but the panel strongly believes that the starting dose used in IDEA⁸ (1,000 mg/m² twice daily) should be used because it is the only validated dose. The panel also discussed that, in practice, oxaliplatin is dropped from FOLFOX and CapeOX after 6 to 10 doses to reduce the risk of severe neurotoxicity. Therefore, it may be reasonable to complete 3 months of CapeOX for patients with high-risk stage III colon cancer and then drop oxaliplatin and continue capecitabine. However, the panel was divided on whether to recommend this approach.

Because of the methodologies and results of IDEA, some of the findings are open to interpretation. However, it will be difficult to generate data that are any more definitive than those from the IDEA analysis. Therefore, the NCCN Guidelines list FOLFOX and CapeOX as equally preferred options without strongly recommending one over the other. Efficacy is similar, FOLFOX is less expensive for US patients, and CapeOX is more convenient for most patients (especially 3 months of CapeOX vs 6 months of FOLFOX). Both regimens are preferred over the other options for adjuvant treatment of patients with stage III colon cancer—6 months of either capecitabine or 5-FU/leucovorin—which are generally only recommended for patients with stage III colon cancer if they cannot tolerate oxaliplatin.

Despite the large number of patients in IDEA, the results were not as definitive as the panel would require for these new recommendations to be listed as category 1. Therefore, these shorter durations are included in the guidelines as category 2A recommendations. The 6-month durations of FOLFOX or CapeOX for patients with stage III colon cancer remain as category 1 recommendations based on older trials.^15–19

Finally, the panel also briefly discussed the limited data from IDEA on stage II colon cancer or stage II/III rectal cancer. Only one site included patients with rectal cancer, none of whom had preoperative chemotherapy, and only 2 sites included patients with stage II colon cancer, one of which included mostly those with high-risk stage II. Overall, the panel believed there are not enough data from IDEA to draw any conclusions for patients with rectal or stage II colon cancer.

Treatment of RAS Wild-Type/BRAF Mutation–Positive Metastatic CRC

In the 2017 version of the NCCN Guidelines, patients with unresectable, advanced, or metastatic CRC (mCRC) were managed with a continuum of care that included 20 first-line systemic treatment options, 33 second-line options, and 13 options for subsequent therapies in up to 7 lines of treatment.^28,29 A growing list of factors are considered when choosing therapies for each patient, including the goals of treatment, type and timing of prior therapy, different efficacy and toxicity profiles of the regimens, KRAS and NRAS mutational tumor status, and patient comorbidities and preferences. MSI status and location of the primary tumor were recently added as additional considerations.^30–34 Although survival for patients with advanced CRC has improved dramatically over the past decades, the most recently reported 5-year survival rate for patients with stage IV CRC was only 14%.^1,35 Moreover, patients with BRAF-mutated mCRC have a particularly poor prognosis. BRAF V600E mutations are present in approximately 8% of mCRC cases and are associated with a more aggressive biology, shorter OS, and decreased response to chemotherapy compared with BRAF wild-type tumors.³⁶ Thus, additional treatment options are still needed in mCRC, especially for the BRAF mutation–positive subset of patients.

BRAF is downstream of EGFR and RAS in the MAPK signaling pathway. The BRAF V600E mutation results in constitutive MAPK signaling that encourages cellular proliferation.³⁷ Vemurafenib selectively inhibits the V600E-mutated form of the BRAF kinase, thereby reducing aberrant MAPK signaling.³⁸ It has an FDA indication for treatment of patients with BRAF V600E–mutated, unresectable, or metastatic melanoma.³⁹ However, vemurafenib monotherapy has shown limited activity in mCRC.⁴⁰ Preclinical data suggest that BRAF V600E inhibition alone is ineffective because feedback activation of EGFR occurs.^41,42 Therefore, blockage of EGFR and BRAF V600E together has been speculated to be more effective than BRAF V600E inhibition alone. In fact, preclinical studies show a synergistic effect of such dual inhibition.^41,42 However, vemurafenib in combination with cetuximab- or panitumumab-based therapy has been ineffective in early clinical trials.⁴³ Results from preclinical and early clinical studies suggest that the addition of irinotecan to BRAF and EGFR inhibition may improve antitumor activity.^44,45

The combination of vemurafenib, cetuximab, and irinotecan was thus tested in patients with BRAF V600E–mutated mCRC in the recent phase II SWOG S1406 trial. In this trial, 99 patients with BRAF-mutant, RAS wild-type tumors who received 1 or 2 prior regimens were randomized to irinotecan and cetuximab with or without vemurafenib.⁹ The NCCN Panel reviewed updated results of this trial that were presented at the 2017 ASCO annual meeting. The primary end point of median progression-free survival (PFS) was improved in the vemurafenib arm (4.3 vs 2.0 months; HR, 0.48; 95% CI, 0.31–0.75; P=.001). Response was also improved, with response rates (only partial responses [PRs] were seen) of 4% vs 16% (P=.08) and disease control rates (partial responses + stable disease) of 22% vs 67% (P=.001). Grade 3/4 adverse events (AEs) that were higher in the vemurafenib arm included neutropenia (33% vs 7%), anemia (13% vs 0%), and nausea (20% vs 2%). Crossover was allowed for the cetuximab/irinotecan group, and 48% of those patients received vemurafenib at time of progression. Of those who crossed over, the partial response rate was 17% and the stable disease rate was 55%, for a disease control rate of 72%. The HR for the secondary end point of OS, which would be attenuated by the crossover, was 0.73 (95% CI, 0.45–1.17; P=.19).

After the panel reviewed these data, one member pointed out that results of the safety lead-in phase of the BEACON CRC trial (ClinicalTrials.gov Identifier: NCT02928224) were to be presented at the ESMO 2017 Congress. This international phase III trial is comparing (1) encorafenib with cetuximab plus or minus binimetinib versus (2) investigator choice of irinotecan/cetuximab or FOLFIRI/cetuximab in patients with BRAF V600E–mutant mCRC whose disease has progressed on 1 or 2 prior metastatic regimens. Encorafenib is another inhibitor of mutant BRAF, and binimetinib inhibits MEK, which is downstream from BRAF in the MAPK pathway.^37,46 The combination of these 2 drugs has been shown to improve PFS in BRAF-mutant melanoma compared with encorafenib monotherapy.⁴⁷ Although the BEACON CRC data had not been presented at the time of panel discussion, several panel members were aware that the results would be better than those seen in SWOG S1406. In fact, the now-presented data show that the triplet regimen (encorafenib/cetuximab/binimetinib) was fairly well tolerated in 30 patients, with dose-limiting toxicities in 5 patients (cetuximab infusion reaction, n=2; grade 2 retinopathy, n=2; grade 2 decreased ejection fraction, n=1). Of the 30 patients, 19 (63%) experienced a grade 3 or 4 AE, including fatigue (grade 3 in 4 patients), urinary tract infection (grade 3 in 3 patients), increased aspartate aminotransferase (grade 3 in 2 patients; grade 4 in 1 patient), and increased blood creatine phosphokinase (grade 3 in 3 patients). The most common AEs of any grade were diarrhea (77%), dermatitis acneiform (67%), nausea (63%), and fatigue (63%).^48,49 In 29 patients with BRAF V600E–mutated mCRC, the overall response rate (3 complete responses + 11 partial responses) was 48%. All remaining patients had stable disease, for a disease control rate of 100%. The preliminary estimated median PFS was 8 months. One panel member reasoned that vemurafenib/cetuximab/irinotecan should not be added based on the SWOG S1406 results, but rather the panel should wait for BEACON CRC results and FDA approval of encorafenib and binimetinib. Although the panel members eagerly await more mature data from BEACON CRC and forthcoming FDA approvals, the vast majority of the panel felt that the data from SWOG S1406 were strong enough to warrant the addition of vemurafenib/cetuximab/irinotecan for patients with BRAF V600E–mutated mCRC at this time. They believed that this subset of patients is in great need of additional treatment options and saw no justification for waiting for better regimens.

The panel then discussed whether to recommend vemurafenib/panitumumab/irinotecan for these patients by extrapolation from the SWOG S1406 data. In general, the panel believes that panitumumab and cetuximab are interchangeable, and some clinic formularies may only list one or the other agent. Furthermore, panel members pointed out that some patients may experience a severe reaction to cetuximab. Safety of vemurafenib/panitumumab/irinotecan has been shown in a case report of a patient with BRAF-mutant cholangiocarcinoma.⁵⁰ Therefore, the panel voted to add this regimen as an additional option for patients with BRAF V600E–mutant mCRC.

Next, the panel discussed whether dabrafenib could be substituted for vemurafenib by extrapolation of SWOG S1406 data. However, they do not believe that dabrafenib and vemurafenib are interchangeable. Dabrafenib was studied in patients with BRAF V600E–mutant mCRC in a case report and a few early clinical trials.^51–54 The largest was a single-arm trial that included 43 patients with BRAF V600E–mutant mCRC who received dabrafenib with trametinib.⁵¹ Trametinib is another MEK inhibitor with an FDA indication in BRAF-mutated melanoma and non–small cell lung cancer in combination with dabrafenib.⁵⁵ The complete response rate was 2% (1 patient), partial response rate was 9% (4 patients), and stable disease rate was 56% (24 patients). Grade 3 AEs occurred in 58% of patients, with nausea, pyrexia, and fatigue being the most common events of any grade. Four patients (9%) discontinued treatment because of AEs. To the best of the panel’s knowledge, however, dabrafenib has not been studied in combination with irinotecan or an EGFR inhibitor, and therefore no safety data for the proposed combination exist. Therefore, the panel declined to add dabrafenib/(cetuximab or panitumumab)/irinotecan to the continuum of care for patients with BRAF-mutant mCRC. The panel noted that trametinib/dabrafenib/panitumumab is currently being studied in an open-label phase I/II trial of patients with BRAF V600E–mutated mCRC (ClinicalTrials.gov identifier: NCT01750918).

Overall, the panel is gratified by the addition of new treatment options for patients with BRAF-mutated mCRC that has progressed on 1 or 2 previous metastatic regimens (see COL-11 and COL-D 2 of 10, pages 362 and 364, respectively, as examples). They are hopeful that ongoing trials will lead to even better options for this difficult-to-treat subset of patients. With the addition of vemurafenib/(cetuximab or panitumumab)/irinotecan to the mCRC continuum of care, the BRAF status of the tumor is now another factor that must be considered when choosing therapies for patients with mCRC. BRAF status can be determined using any methodology, including next-generation sequencing, which can be used to simultaneously test for MSI and mutations in RAS (COL-B 4 of 5, page 363). The panel further notes that BRAF and RAS mutations are mutually exclusive. Therefore, although the guidelines only indicate that the new regimens are for “BRAF V600E mutation–positive” disease, these patients’ tumors contain wild-type RAS.