Abstract
Background:
Stratification of patients with stage III colon cancer into low (T1–3N1) and high (T4 and/or N2) risk groups is used to guide the duration of adjuvant chemotherapy. We determined the relative contribution of clinical and molecular features to survival by risk group.
Materials & Methods:
Stage III colon cancer (N=5,337) patients from 2 adjuvant trials of FOLFOX ± cetuximab [N0147 (Alliance), PETACC-8] were risk grouped, then subgrouped by clinical features and molecular variables [KRAS and BRAF/mismatch repair (MMR) combined variable]. Distributions of DFS, OS, and survival after recurrence (SAR) were estimated. In multivariable Cox models, backward elimination was performed for analysis of candidate predictors of outcomes. Relative contributions of model-selected variables to outcomes by risk group were calculated using χ2.
Results:
In high vs low risk tumors, significantly poorer OS was observed for right sidedness and for mutant KRAS, BRAFV600E/pMMR or BRAFV600E/dMMR subgroups. Poor prognosis of mutant KRAS was similar among risk groups. BRAF/MMR and sidedness were associated with poorer OS and SAR in high but not low risk tumors. Specifically, BRAFV600E/pMMR (OS: HR=1.75; 95% CI: 1.36–2.24; Padj<.0001) and right- vs left-sidedness were associated with significantly poorer DFS, OS (HR=1.56; 95% CI: 1.31–1.83; Padj<.0001), and SAR (HR=1.64; 95% CI: 1.37–1.95; Padj<.0001). Age, gender and KRAS were the top three relative contributors to DFS and OS among low risk tumors; sidedness ranked first for DFS and OS and second to BRAF/MMR for SAR among high risk tumors.
Conclusion:
Sidedness and BRAF/MMR contributed the most to survival outcomes among high risk tumors and should be interpreted in the context of risk group.
Keywords: Colon cancer, stage III, adjuvant therapy, risk groups, microsatellite instability, deficient mismatch repair, prognosis, recurrence
INTRODUCTION
Despite adjuvant chemotherapy, nearly one-third of patients with stage III colon cancer will develop tumor recurrence[1–3] and most of these patients eventually die of their disease. An international adjuvant study known as IDEA (International Duration Evaluation of Adjuvant Therapy) evaluated the noninferiority of 3 months compared to the standard 6 months of adjuvant fluoropyrimidine plus oxaliplatin in patients with stage III colon cancer[4]. While noninferiority of 3 versus 6 months of therapy was not confirmed in the overall population, a post hoc analysis of patients with low risk (T1–3N1) cancers revealed that 3 months of therapy was noninferior to 6 months and among those classified as high risk (T4, N2, or both), 6-months of therapy was superior to 3 months[4]. Based upon these data, T and N risk grouping is routinely used by clinicians and endorsed by ESMO[5] and NCCN [6] guidelines, to determine the recommended duration of adjuvant chemotherapy for stage III patients. To date, however, the relative contribution of clinical and molecular features to patient survival in low and high risk groups of stage III patients has not been studied. Improving patient outcomes in the adjuvant setting will require development of prediction models that incorporate clinical and biological data into T and N staging to guide precision oncology approaches.
Using pooled data from two phase III trials of adjuvant fluoropyrimidine plus oxaliplatin chemotherapy (NCCTG N0147, PETACC-8) where patient outcomes were similar by treatment arm[1, 3], we categorized stage III patients into low risk (T1–3 and N1) and high risk (T4, N2 or both) groups [4]. Stratified by risk groups, we examined the relationship of clinical and molecular variables to survival outcomes, and then utilized Cox model selection procedures to identify candidate predictors of disease-free survival (DFS), overall survival (OS), and survival after recurrence (SAR). Importantly, insight into tumor biology and metastatic potential can be gained by evaluation of SAR. Data indicate that the impact of the BRAFV600E point mutation on patient prognosis in colon cancer is dependent on the status of the DNA mismatch repair (MMR) system[7]. Accordingly, we examined BRAF [wild type (WT) or mutant V600E) and MMR as a combined variable (BRAF/MMR) with four possible combinations (BRAF WT/pMMR, BRAF WT/dMMR, BRAFV600E/pMMR, BRAFV600E/dMMR). Right- vs left-sided colon cancers are enriched in BRAFV600E and microsatellite instability (MSI) [due to deficient MMR (dMMR)][8, 9] that can contribute to differences in prognosis by primary tumor sidedness. Information gained from this study may inform patient management including decision-making in the adjuvant setting.
PATIENTS AND METHODS
Patient population
Patients with resected stage III colon cancers (N=5,337) had participated in two trials of adjuvant FOLFOX ± cetuximab [North Central Cancer Treatment Group (NCCTG) N0147[1] (Alliance) and PETACC-8[3]] where outcomes were similar by study arm that enabled data pooling. The study population includes patients with available data for all biomarkers that were analyzed in prospectively collected tissues.
Molecular Analysis
DNA mismatch repair (MMR) status was determined in tumor tissue by analysis of MMR protein (MLH1, MSH2, and MSH6) by immunohistochemistry (IHC). If IHC was indeterminate [10], microsatellite instability (MSI) testing was performed. Tumors were classified as dMMR if lthere was loss of one or more MMR proteins or if tumors exhibited high-level MSI.
Tumor tissue was analyzed for mutations in KRAS (codon 12 or 13 in exon 2) or the BRAFV600E point mutation (exon 15) genes, as described previously[11, 12]. Written informed consents were obtained from patients at study entry, and the study was approved by the Mayo Clinic Institutional Review Board.
Statistical analyses
Patients were risk grouped using T and N data and the associations of the study variables with patient disease-free survival (DFS), overall survival (OS), and survival after recurrence (SAR) were analyzed univariately and in multivariable Cox models. DFS was defined as the time from the date of random assignment to recurrence or death due to all causes, whichever occurred first. Survival after recurrence (SAR) was defined as the time from recurrence to death from any cause. 5-year survival rates were determined based on Kaplan-Meier estimates. Cox models were used for evaluating associations between outcomes and clinical/molecular factors, stratifying by treatment arm. Study variables included in the initial model were age, sidedness, histologic grade, gender, lymph nodes, performance status and KRAS and BRAF/MMR (Tables 1, 2). After determination of the optimal functional form (continuous or categorical) of clinical and molecular variables, a backward elimination selection was performed to identify the important independent prognostic factors. The proportional hazards assumption was confirmed by examination of the Schoenfeld residuals plot[13]. The relative contributions of model selected variables to outcomes by risk group was calculated using the χ2 from Harrell’s rms R package (version 3.2.3; http://biostat.mc.vanderbilt.edu/rms) based on multivariable models. In an exploratory analysis, Cox models for survival outcomes were utilized to examine interaction between sidedness and treatment arm among WT KRAS tumors. The relative contribution of each factor represents the percentage of chi-square made up of the total for the model. Two-sided P values are reported; P <.05 was considered statistically significant and was not adjusted for multiple comparisons. Analyses were performed using SAS software (version 9.4; SAS Institute Inc.). Data were frozen as of 8/5/15. Data collection and statistical analyses were performed by the Alliance Statistics and Data Center.
Table 1.
Association of BRAF/MMR and Clinicopathological Features in Low and High Risk Groups of Stage III Colon Cancer Patients.
| Low Risk (T1–3 N1) | |||||
|---|---|---|---|---|---|
| BRAF WT/pMMR (n=1845) | BRAF WT/dMMR (n=133) | BRAF MT/pMMR (n=105) | BRAF MT/dMMR (n=108) | p-value | |
| Age, median (range) | 59 (19–82) | 55 (28–80) | 62 (28–80) | 66 (43–86) | <0.0001 |
| Right (%) | 682 (37%) | 106 (80%) | 72 (69%) | 99 (92%) | |
| High (%) | 240 (13%) | 53 (40%) | 27 (25%) | 55 (51%) | |
| Female (%) | 808 (44%) | 60 (45%) | 50 (48%) | 79 (73%) | |
| LNs, median (range) | 2 (1–3) | 1 (1–3) | 1 (1–3) | 2 (1–3) | 0.5705 |
| 1 | 359 (20%) | 23 (18%) | 20 (20%) | 28 (26%) | |
| High Risk (T4 and/or N2) | |||||
| BRAF WT/pMMR (n=1596) | BRAF WT/dMMR (n=128) | BRAF MT/pMMR (n=183) | BRAF MT/dMMR (n=103) | p-value | |
| Age, median (range) | 58 (19–85) | 51 (23–80) | 61 (27–81) | 67 (45–84) | <0.0001 |
| Right (%) | 612 (39%) | 90 (72%) | 134 (73%) | 97 (94%) | |
| High (%) | 371 (23%) | 69 (55%) | 86 (47%) | 59 (58%) | |
| Female (%) | 685 (43%) | 53 (41%) | 100 (55%) | 81 (79%) | |
| LNs, median (range) | 6 (1–33) | 5 (1–31) | 7 (1–51) | 5 (1–22) | <0.0001 |
| 1 | 357 (23%) | 25 (20%) | 47 (26%) | 32 (32%) | |
WT: wild-type; MT: mutant. LN: lymph nodes
Table 2.
Univariate Analysis of Clinical and Molecular Features with Patient Survival in Low and High Risk Stage III Colon Cancers
| Variable | DFS | OS | SAR | ||||||
|---|---|---|---|---|---|---|---|---|---|
| Total | Hazard Ratio (95% CI) | P-value | Total | Hazard Ratio (95% CI) | P-value | Total | Hazard Ratio (95% CI) | P-value | |
| Low Risk | |||||||||
| Age | 2770 | 1.02 (1.01–1.03) | <.00011 | 2770 | 1.04 (1.03–1.05) | <.00011 | 473 | 1.02 (1.00–1.03) | 0.00701 |
| Right | 1170 | 1.02 (0.87–1.20) | 0.78931 | 1170 | 1.25 (1.03–1.52) | 0.02481 | 200 | 1.57 (1.24–2.00) | 0.00021 |
| MT/ dMMR | 108 | 0.62 (0.38–1.02) | 0.06171 | 108 | 0.79 (0.45–1.37) | 0.39931 | 10 | 2.05 (0.96–4.36) | 0.06351 |
| Mutated | 726 | 1.40 (1.18–1.66) | 0.00011 | 726 | 1.44 (1.17–1.76) | 0.00051 | 165 | 1.18 (0.92–1.50) | 0.18611 |
| 1 | 515 | 1.36 (1.12–1.65) | 0.00161 | 515 | 1.63 (1.30–2.04) | <.00011 | 99 | 1.51 (1.14–2.00) | 0.00361 |
| Female | 1243 | 0.71 (0.60–0.83) | <.00011 | 1243 | 0.64 (0.53–0.79) | <.00011 | 191 | 0.95 (0.74–1.21) | 0.67341 |
| High | 459 | 1.03 (0.83–1.28) | 0.78691 | 459 | 1.14 (0.88–1.47) | 0.32011 | 85 | 1.52 (1.14–2.04) | 0.00471 |
| LNs | 1455 | 0.99 (0.98–1.00) | 0.02561 | 1455 | 0.99 (0.97–1.00) | 0.05951 | 266 | 1.00 (0.98–1.01) | 0.56151 |
| High Risk | |||||||||
| Age | 2565 | 1.01 (1.00–1.01) | 0.00401 | 2565 | 1.02 (1.01–1.02) | <.00011 | 988 | 1.01 (1.01–1.02) | 0.00061 |
| Right | 1171 | 1.32 (1.18–1.49) | <.00011 | 1171 | 1.69 (1.47–1.94) | <.00011 | 487 | 1.74 (1.50–2.02) | <.00011 |
| MT/ dMMR | 103 | 1.19 (0.89–1.60) | 0.24861 | 103 | 1.54 (1.12–2.11) | 0.00731 | 36 | 2.23 (1.55–3.21) | <.00011 |
| Mutated | 650 | 1.26 (1.10–1.43) | 0.00051 | 650 | 1.31 (1.13–1.51) | 0.00031 | 298 | 1.06 (0.91–1.24) | 0.42881 |
| 1 | 545 | 1.21 (1.05–1.39) | 0.00831 | 545 | 1.35 (1.15–1.58) | 0.00021 | 221 | 1.23 (1.03–1.45) | 0.01991 |
| Female | 1177 | 0.94 (0.83–1.06) | 0.29761 | 1177 | 0.91 (0.79–1.04) | 0.18041 | 450 | 0.94 (0.81–1.09) | 0.43101 |
| High | 722 | 1.27 (1.12–1.45) | 0.00021 | 722 | 1.50 (1.30–1.73) | <.00011 | 303 | 1.56 (1.34–1.82) | <.00011 |
| LNs | 1323 | 1.00 (0.99–1.01) | 0.96331 | 1323 | 1.00 (0.99–1.01) | 0.84601 | 529 | 1.00 (0.99–1.01) | 0.96231 |
Covariate Wald p-value;
Type 3 Wald p-value. LNs: lymph nodes
Age and LNs are continuous variables (1 unit increase)
RESULTS
Study Population Stratified by Risk Group
Patients with stage III colon cancer (N=5,337) from two phase III adjuvant trials were divided into low risk (T1–3N1) [N=2770 (51.9%)] and high risk (any T4 and/or N2) [N=2565 (48.1%)] groups. Among low risk patients, 726 (31.3 %) tumors harbored mutant KRAS, 224 (9.6%) had BRAFV600E, and 253 (9.1%) showed dMMR. Corresponding numbers in the high risk group were 650 (30.3%) for KRAS, 296 (13.8%) for BRAFV600E, and 239 (9.3%) for dMMR. A flow chart of the study population is shown in Figure 1. Given that the association of BRAF with prognosis is ideally interpreted in the context of MMR status, we evaluated a BRAF/MMR combined variable. For BRAF/MMR, numbers for low and high risk subgroups are shown in Table 1. Median patient follow-up was 83.4 months.
Figure 1:
Flow chart of the study population of stage III colon cancer patients from two adjuvant chemotherapy trials by T (tumor), N (lymph node) risk group. Tumor molecular characteristics within each risk group are shown. dMMR, deficient mismatch repair; pMMR, proficient mismatch repair. *Missing data, unable to subgroup (n=442); **Missing MMR status (n=422); ***Missing data, unable to subgroup (n=424); ****Missing MMR data (n=428).
In both risk groups, patients with BRAFV600E/dMMR tumors were signficantly older and more likely female consistent with a sporadic origin[9], and tumors showed high grade histology and a predilection for the right colon (Table 1). Patients with WT BRAF/dMMR tumors were younger and more likely to be male (Table 1), as were patients with mutant KRAS /dMMR tumors (Supplemental Table 1) independent of risk group. BRAFV600E/pMMR tumors were also significantly associated with high grade histology and right vs left sidedness compared to vs WT BRAF/pMMR tumors in both risk groups (Table 1). KRAS mutant/pMMR tumors were more likely to have low grade histology and were evenly distributed by primary site (Supplemental Table 1). Among patients with high risk tumors, an increased number of positive regional lymph nodes was seen in patients with BRAFV600E/pMMR tumors compared to the other subgroups (Table 1).
Association of molecular and clinical variables with clinical outcome by risk group
To enhance the interpretation of clinical and molecular features in stage III tumors, we analyzed their relationship to clinical outcome variables by risk group. In a univariate analysis, primary tumor sidedness was prognostic for OS and SAR among patients with low risk tumors and for DFS, OS and SAR among those with high risk tumors (Table 2). Specifically, patients with right- vs left-sided tumors had significantly poorer survival. Interestingly, women had significantly better DFS and OS than did men that was exclusive to low risk tumors (Table 2). Data for age and performance status (PS) are shown in Table 2. In both low and high risk groups, patients with BRAFV600E/pMMR tumors had significantly shorter OS and SAR compared to those with WT BRAF/pMMR tumors (Table 2). Among high risk tumors, those that were BRAFV600E/dMMR also showed significantly poorer OS (HR=1.54; 95% CI: 1.12–2.11; P=.0073) and SAR (HR=2.23; 95% CI: 1.55–3.21; P<.0001) (Table 2). These findings for BRAF/MMR in high risk patients are shown in Kaplan-Meier plots (Figure 2A, B). Mutant vs WT KRAS was associated with significantly shorter DFS and OS, but not SAR, independent of risk group (Table 2).
Figure 2:
Association of BRAF/MMR with overall survival (OS) and survival after recurrence (SAR) in patients with stage III colon cancer. All patients were treated with FOLFOX-based adjuvant chemotherapy. (A), Kaplan-Meier (KM) plot of OS for the BRAF/MMR combined variable in patients with high (T4 and/or N2) risk tumors. (B) K-M plot of SAR is shown for BRAF/MMR among patients with stage III tumors. P values are derived from the stratified logrank test. HR, hazard ratio; CI, confidence interval; deficient (d) MMR; proficient (p) pMMR; WT, wild type.
By multivariable analysis, high vs. low risk patients had significantly poorer DFS (HR=2.42; 95% CI: 2.16–2.71; Padj<.0001), OS [HR =2.61; 95% CI: 2.29–2.98; Padj<.0001], and SAR [HR =1.54; 95% CI: 1.32–1.81; Padj<.0001] independent of covariates. At 5-years of follow-up, only 57.2% (95% CI: 55.1–59.5) of high risk patients were alive and disease-free compared to 79.2% (95% CI: 77.5–81.0) of low risk patients. Significantly poorer OS was found for patients with high vs low risk tumors of the right colon (HR=2.84; 95% CI: 2.36–3.41; Padj <0.0001) and left-colon (HR=2.35; 95% CI: 1.94–2.85; Padj <0.0001), and those harboring BRAFV600E (HR=3.08; 95% CI: 2.12–4.48; Padj <0.0001), mutant KRAS (HR=2.43; 95% CI: 1.97–3.00; Padj <0.0001), proficient MMR (HR=2.46; 95% CI: 2.14–2.83; Padj <0.0001) or deficient MMR (HR=4.18; 95% CI: 2.69–6.49; Padj <0.0001[Figure 3A–F].
Figure 3:
Overall survival (OS) by KRAS, BRAF and DNA mismatch repair (MMR) status in low (T1–3N1) versus high (T4 and/or N2) risk stage III colon cancer patients.
The multivariable association of clinical and molecular features and with patient survival by risk group is shown in Table 3. We performed model selection procedures in low and high risk groups, separately, where initial models included variables of age, gender, primary tumor sidedness, PS, BRAF/MMR and KRAS. Since all associations with outcome variables (DFS, OS, SAR) by risk group showed P <0.10, we sought to determine which variables independently contributed the most to prognosis by performing backwards elimination utilizing Cox models. This procedure was repeated for each of the three survival outcomes. The multivariable association of primary tumor sidedness with DFS and OS was limited to high risk patients. Specifically, high risk patients with right- vs left- sided tumors had significantly worse DFS (HR=1.29; 95% CI: 1.15–1.46, Padj <.0001) and OS (HR=1.56; 95% CI: 1.31–1.83, Padj <.0001) independent of covariates (Table 3). For SAR, significantly shorter SAR was observed for right- vs left-sided tumors in both risk groups (low risk: HR=1.76, 95% CI: 1.33–2.32, Padj <.0001; high risk: HR=1.64, 95% CI: 1.37–1.95, Padj <.0001). Female vs male gender was associated with significantly better DFS and OS (HR= 0.65, 95% CI: 0.53–0.80; Padj <.0001) in low risk but not high risk patients; gender was eliminated from the model for SAR (Table 3). PS was significantly associated with all outcome variables independent of risk group. Age was significantly prognostic only among low risk patients.
Table 3.
Multivariable Cox Models with Backward Elimination in Low and High Risk Stage III Colon Cancers
| Low Risk (T1–3, N1) | ||||||
|---|---|---|---|---|---|---|
| DFS | OS | SAR | ||||
| Variable | HR (95% CI) | p-value | HR (95% CI) | p-value | HR (95% CI) | p-value |
| Age | 1.02 (1.01, 1.03) | <.0001 | 1.03 (1.02, 1.04) | <.0001 | 1.02 (1.00, 1.03) | .0462 |
| Female | 0.71 (0.60, 0.84) | 0.65 (0.53, 0.80) | <.0001 | |||
| Mutated (MT) | 1.38 (1.16, 1.64) | .0003 | 1.41 (1.14, 1.73) | .0012 | ||
| 1 | 1.31 (1.08, 1.59) | .0059 | 1.53 (1.22, 1.92) | .0002 | 1.66 (1.22, 2.28) | .0014 |
| dMMR & MT | 1.91 (0.89, 4.09) | .0964 | ||||
| Right | 1.76 (1.33, 2.32) | |||||
| High Risk (T4 and/or N2) | ||||||
| DFS | OS | SAR | ||||
| Variable | HR (95% CI) | p-value | HR (95% CI) | p-value | HR (95% CI) | p-value |
| Mutated (MT) | 1.25 (1.10 1.43) | 1.43 (1.20, 1.70) | <.0001 | 1.23 (1.02, 1.48) | ||
| 1 | 1.22 (1.06, 1.402 | .0063 | 1.34 (1.12, 1.59) | .0010 | 1.17 (0.98, 1.42) | .0913 |
| dMMR & MT | 1.30 (0.92, 1.83) | .1325 | 1.80 (1.21, 2.67) | .0035 | ||
| Right | 1.29 (1.15, 1.46) | 1.56 (1.31, 1.83) | 1.64 (1.37, 1.95) | |||
Among patients with low risk tumors, BRAF/MMR was eliminated from the models for DFS and for OS. Patients with low risk tumors that were BRAFV600E/pMMR had significantly shorter SAR (HR=1.82; 95% CI: 1.11–2.97, Padj =.017) (Table 3). Similarly, low risk BRAFV600E/dMMR tumors had shorter SAR (HR=1.91; 95% CI: 0.89–4.08, Padj =.096) that did not achieve statistical significance due to small patient numbers. Mutant KRAS was associated with significantly shorter DFS and OS (HR=1.41; 95% CI: 1.14–1.73, Padj =.0012), but not SAR among low risk tumors (Table 3). Among high risk tumors, BRAF/MMR was prognostic for OS and SAR. Compared to those with WT BRAF, BRAFV600E/pMMR tumors had significantly shorter OS (HR=1.75; 95% CI: 1.36–2.24, Padj <.0001) which did not achieve statistical significance for BRAFV600E/dMMR tumors (Table 3). For the endpoint of SAR, however, high risk BRAFV600E/pMMR tumors (HR=2.62; 95% CI: 2.04–3.42, Padj <.0001) and BRAFV600E/dMMR (HR=1.80; 95% CI: 1.21–2.67, Padj =.0035) tumors each had significantly shorter SAR (Table 3). Mutant KRAS was associated with significantly shorter DFS, OS (OS: HR=1.43; 95% CI: 1.20–1.70, Padj <.0001), and SAR.
Relative Contribution to Patient Survival
We determined the relative contribution (in percent) of each clinical and molecular feature to patient survival in low and high risk groups. Among low risk tumors, sidedness was a top three contributor only to SAR (Figure 4). However, sidedness was the highest contributor to DFS (42.3%) and OS (40%) among high risk tumors, and was second (37.4%) only to BRAF/MMR (53.1%) as a contributor to SAR. Therefore, sidedness ranked first for as a contributor to DFS and OS among high risk patients and ranked second for SAR (Figure 4B, D, and F). The contribution of BRAF/MMR to SAR increased nearly 2-fold in high vs low (53.1% vs. 28.7%) risk patients (Figure 4E, F). Age and gender were the top two contributors to DFS and OS among low, but not high risk tumors. Mutant KRAS had a similar relative contribution to DFS and OS in low and high risk patients whereby it ranked third for DFS and was fourth for OS among both low and high risk tumors, respectively (Figure 4). KRAS did not contribute appreciably to SAR.
Figure 4:
Relative contribution (in percentage) of clinical and molecular tumor features to the prediction of disease-free survival (DFS)[top], overall survival (OS)[middle], and survival after recurrence (SAR) [bottom] among all stage III patients with low risk (T1–3N1)[left panel] or high risk (T4 and/or N2) [right panel] cancers. MMR, mismatch repair; PS, performance status; primary tumor site, left vs right.
Primary tumor sidedness and risk group
Patients with right- vs left-sided tumors in both risk groups were older and their tumors were significantly more likely to be high grade and to harbor mutant KRAS and any BRAF/dMMR (data not shown). Those with right-sided tumors were also more likely to be female among low, but not high risk patients. None of the other variables differed signficantly by tumor site. Significantly poorer DFS, OS and SAR were found for patients with high vs low risk tumors of both the right colon (DFS: HR=2.63; 95% CI: 2.23–3.10; Padj <0.0001); SAR: HR=1.48; 95% CI: 1.19–1.84; Padj =0.0003) and the left-colon (DFS: HR=2.23; 95% CI: 1.91–2.61; Padj <0.0001); SAR: HR=1.59; 95% CI: 1.26–2.00; Padj <0.0001).
We performed an exploratory analysis of patient survival by tumor sidedness and study treatment arm in stage III colon cancers. Of note, the addition of cetuximab to FOLFOX failed to improve the primary endpoint of patient DFS in the two adjuvant trials included in this report[1, 3]. A multivariate test of interaction between sidedness and treatment arm for survival outcomes did not achieve statistical significance. Furthermore, similar DFS and OS were observed by treatment arm in both right- and left-sided cancers with WT KRAS (data not shown).
DISCUSSION
After publication of the IDEA study[4], T, N risk grouping has been widely adopted by oncologists for guiding the recommended duration of adjuvant chemotherapy in stage III colon cancer and has been endorsed by current guidelines. In this report, we sought to identify clinical and/or molecular feature(s) that can best predict patient survival by risk group [low risk (T1–3N1); high risk (T4, N2, or both)] using data from two phase 3 adjuvant trials that evaluated a fluoropyrimidine plus oxaliplatin. Risk grouping was shown to provide robust prognostic stratification in our cohort. We then compared the association of variables with survival between risk groups. By multivariable analysis, mutant KRAS was significantly associated with poorer DFS and OS independent of risk group, as was its relative contribution to these survival outcomes. In contrast, BRAFV600E/pMMR tumors had significantly poorer OS that was limited to high risk tumors, yet the association with SAR was significant in both risk groups. The association of primary tumor sidedness with DFS and OS was limited to high risk tumors, although significantly shorter SAR was observed for right- vs left-sided tumors in both low and high risk groups. Importantly, the relative contribution of sidedness was highest among variables examined for DFS and OS among high but not low risk tumors, and its contribution to SAR was only exceeded by BRAFV600E/pMMR among high risk patients. Differences in the biology of colon cancer based on sidedness has been shown by multi-omics including differentially expressed genes (TP53, KRAS, BRAFV600E, PIK3CA, SMAD4, CTNNB1, and PTEN[14]), miRNAs, and DNA methylation profiles [15, 16]. In addition, transcriptomically determined consensus molecular subtypes (CMS) were shown to vary by tumor sidedness with a decrease in CMS1 and CMS3 and an increase in CMS2 prevalence moving from the right to left colon[14]. The gut microbiome may also be a factor in that CRCs with a high abundance of Fusobacterium nucleatum DNA were found to increase in prevalence from rectum to cecum[17]. The DNA abundance of this anaerobic bacterium was inversely related to intratumoral CD3+ T-cell density[18] and was associated with significantly poorer prognosis in patients with CRC[19].
Prior studies have not adequately examined BRAF in the context of MMR status which can lead to misinterpretation of its prognostic impact and inconsistencies between studies[11, 20]. BRAF/MMR was prognostic among high risk patients for OS and was the primary driver of SAR as shown by its relative contribution which increased nearly 2-fold in high vs low risk tumors (28.7% to 53.1%). Accordingly, the adverse impact of oncogenic BRAFV600E on patient survival occurs primarily following tumor recurrence. An important observation was that high risk tumors harboring BRAFV600E/pMMR or BRAFV600E/dMMR each had significantly poorer SAR (vs reference). These data indicate that any ‘protective’ effect of dMMR seen in BRAFV600E tumors is lost after tumor recurrence. Further analysis revealed that poorer SAR seen for BRAFV600E/dMMR tumors was limited to patients who received cetuximab (Supplemental Table 2). These results for SAR support data in metastatic CRCs where BRAFV600E is associated with poor prognosis independent of MMR status[21, 22]. For patients wth metastatic CRCs harboring BRAFV600E, targeted therapy with encorafenib and cetuximab has been shown to improve outcomes and is now an FDA-approved treatment option[23]. Furthermore, use of an immune checkpoint inhibitor is an FDA-approved treatment for patients with metastatic CRCs with dMMR irrespective of BRAF status both as first-line[24] and as salvage therapy[25], and a phase 3 adjuvant trial is ongoing to evaluate an anti-PD-L1 antibody in patients with resected stage III colon cancers with dMMR[26].
We also observed that women had significantly better DFS and OS than did men that was confined to low risk tumors. Furthermore, gender and age were the top two contributors to DFS and OS among low, but not high risk tumors. In support of these findings, consistent results were obtained from analysis of time-to-recurrence for these variables (data not shown). Population-based data have confirmed a survival advantage of female compared to male CRC patients whose mechanism has been speculated to be related to the effect of sex hormones, either endogenous or through hormonal replacement therapy[27–29]. Not unexpectedly, PS was an important contributor to patient outcomes. In an exploratory analysis, we examined the potential predictive role of sidedness for benefit from cetuximab given data in metastatic CRC whereby patients with right-vs left-sided primary tumors have poorer survival when treated with anti-EGFR antibodies plus chemotherapy[30, 31]. However, an interaction test was not significant and similar patient survival was found for WT KRAS tumors of the right and left colon by treatment arm.
Strengths of our study include prospective collection of tumor tissues with analysis of molecular features, the randomized phase III trial designs, and the uniform treatment and rigorous patient follow-up. Study limitations include the retrospective pooled analysis and the absence of expanded RAS mutation assessment, although mutations in NRAS and HRAS occur in fewer than 5% of CRCs[32] such that their inclusion would be unlikely to have substantially altered our results. Patients received FOLFOX alone or combined with cetuximab and that similar outcomes were seen by study arm in both adjuvant trials[1, 3]. We acknowledge the potential for cetuximab to influence biomarker-related outcomes. However, analysis by study arm was performed and the main treatment-related finding was that BRAFV600E/dMMR tumors showed poorer SAR only among those who received cetuximab vs FOLFOX alone (Supplemental Table 2), yet cautious interpretation is in order given small patient numbers. The fact that all patients received adjuvant therapy precludes a predictive analysis. We emphasize that with the exception of MMR status, molecular profiling of patients with stage III colon cancer is not standard of care although such data have prognostic value, but have not been shown to be predictive for adjuvant treatment.
In conclusion, the prognostic impact and relative contribution of clinical and molecular features show important differences by IDEA study-defined risk grouping in patients with stage III colon cancer. Whereas the relative contribution of KRAS to survival was largely independent of risk group, the impact of BRAF/MMR on outcome was limited to OS and SAR, and was the primary driver of SAR especially among high risk patients. Importantly, worse SAR observed for BRAFV600E tumors was independent of MMR status which indicates loss of a ‘protective’ effect of dMMR at tumor recurrence. As with BRAF/MMR, the prognostic impact of primary tumor sidedness was primarily seen in high risk tumors where it was a top contributor to all survival outcomes in high risk, but not low risk patients. Taken together, these data indicate that the prognostic impact of BRAF/MMR and tumor sidedness depend on risk group and should be interpreted in this context. Our data serve to refine prognostication by risk group in stage III patients, and demonstrate the clinical utility of integrating molecular analysis with anatomic tumor staging.
Supplementary Material
Highlights.
Sidedness KRAS & BRAFV600E related with poorer survival in high vs low risk tumors
Sidedness and BRAF/MMR contributed the most to survival among high risk tumors
Age gender & KRAS were the top relative contributors to survival in low risk tumors
FUNDING:
The study was supported by grants from the U.S. National Cancer Institute: R01 CA210509-01A1 (to FAS), Alliance for Clinical Trials in Oncology: U10CA180821, U10CA180882 and U24CA196171 (to FAS).
Footnotes
Publisher's Disclaimer: This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.
CONFLICT OF INTEREST:
The authors who have taken part in this study declared that they do not have any conflict of interest with respect to this manuscript.
TRIAL IDENTIFICATION NUMBERS:
Declaration of interests
The authors declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.
REFERENCES
- [1].Alberts SR, Sargent DJ, Nair S, Mahoney MR, Mooney M, Thibodeau SN, et al. Effect of oxaliplatin, fluorouracil, and leucovorin with or without cetuximab on survival among patients with resected stage III colon cancer: a randomized trial. JAMA. 2012;307:1383–93. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [2].Andre T, de Gramont A, Vernerey D, Chibaudel B, Bonnetain F, Tijeras-Raballand A, et al. Adjuvant Fluorouracil, Leucovorin, and Oxaliplatin in Stage II to III Colon Cancer: Updated 10-Year Survival and Outcomes According to BRAF Mutation and Mismatch Repair Status of the MOSAIC Study. J Clin Oncol 2015;33:4176–87. [DOI] [PubMed] [Google Scholar]
- [3].Taieb J, Tabernero J, Mini E, Subtil F, Folprecht G, Van Laethem JL, et al. Oxaliplatin, fluorouracil, and leucovorin with or without cetuximab in patients with resected stage III colon cancer (PETACC-8): an open-label, randomised phase 3 trial. The Lancet Oncology. 2014;15:862–73. [DOI] [PubMed] [Google Scholar]
- [4].Grothey A, Sobrero AF, Shields AF, Yoshino T, Paul J, Taieb J, et al. Duration of Adjuvant Chemotherapy for Stage III Colon Cancer. N Engl J Med. 2018;378:1177–88. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [5].Argiles G, Tabernero J, Labianca R, Hochhauser D, Salazar R, Iveson T, et al. Localised colon cancer: ESMO Clinical Practice Guidelines for diagnosis, treatment and follow-up. Ann Oncol. 2020;31:1291–305. [DOI] [PubMed] [Google Scholar]
- [6].Benson AB 3rd, Venook AP, Al-Hawary MM, Cederquist L, Chen YJ, Ciombor KK, et al. NCCN Guidelines Insights: Colon Cancer, Version 2.2018. J Natl Compr Canc Netw. 2018;16:359–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [7].Blaker H, Alwers E, Arnold A, Herpel E, Tagscherer KE, Roth W, et al. The Association Between Mutations in BRAF and Colorectal Cancer-Specific Survival Depends on Microsatellite Status and Tumor Stage. Clin Gastroenterol Hepatol. 2018. [DOI] [PubMed]
- [8].Domingo E, Laiho P, Ollikainen M, Pinto M, Wang L, French AJ, et al. BRAF screening as a low-cost effective strategy for simplifying HNPCC genetic testing. J Med Genet. 2004;41:664–8. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [9].Poynter JN, Siegmund KD, Weisenberger DJ, Long TI, Thibodeau SN, Lindor N, et al. Molecular characterization of MSI-H colorectal cancer by MLHI promoter methylation, immunohistochemistry, and mismatch repair germline mutation screening. Cancer Epidemiol Biomarkers Prev. 2008;17:3208–15. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [10].Taieb J, Zaanan A, Le Malicot K, Julie C, Blons H, Mineur L, et al. Prognostic Effect of BRAF and KRAS Mutations in Patients With Stage III Colon Cancer Treated With Leucovorin, Fluorouracil, and Oxaliplatin With or Without Cetuximab: A Post Hoc Analysis of the PETACC-8 Trial. JAMA Oncol. 2016;2:643–53. [DOI] [PubMed] [Google Scholar]
- [11].Sinicrope FA, Shi Q, Allegra CJ, Smyrk TC, Thibodeau SN, Goldberg RM, et al. Association of DNA Mismatch Repair and Mutations in BRAF and KRAS With Survival After Recurrence in Stage III Colon Cancers : A Secondary Analysis of 2 Randomized Clinical Trials. JAMA Oncol. 2017;3:472–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [12].Yoon HH, Tougeron D, Shi Q, Alberts SR, Mahoney MR, Nelson GD, et al. KRAS codon 12 and 13 mutations in relation to disease-free survival in BRAF-wild-type stage III colon cancers from an adjuvant chemotherapy trial (N0147 alliance). Clin Cancer Res. 2014;20:3033–43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [13].SCHOENFELD D. Partial residuals for the proportional hazards regression model. Biometrika. 1982;69:239–41. [Google Scholar]
- [14].Loree JM, Pereira AAL, Lam M, Willauer AN, Raghav K, Dasari A, et al. Classifying Colorectal Cancer by Tumor Location Rather than Sidedness Highlights a Continuum in Mutation Profiles and Consensus Molecular Subtypes. Clin Cancer Res. 2018;24:1062–72. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [15].El Jabbour T, Ross JS, Sheehan CE, Affolter KE, Geiersbach KB, Boguniewicz A, et al. PD-L1 protein expression in tumour cells and immune cells in mismatch repair protein-deficient and -proficient colorectal cancer: the foundation study using the SP142 antibody and whole section immunohistochemistry. J Clin Pathol. 2018;71:46–51. [DOI] [PubMed] [Google Scholar]
- [16].Yamauchi M, Morikawa T, Kuchiba A, Imamura Y, Qian ZR, Nishihara R, et al. Assessment of colorectal cancer molecular features along bowel subsites challenges the conception of distinct dichotomy of proximal versus distal colorectum. Gut. 2012;61:847–54. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [17].Mima K, Cao Y, Chan AT, Qian ZR, Nowak JA, Masugi Y, et al. Fusobacterium nucleatum in Colorectal Carcinoma Tissue According to Tumor Location. Clinical and translational gastroenterology. 2016;7:e200. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [18].Mima K, Sukawa Y, Nishihara R, Qian ZR, Yamauchi M, Inamura K, et al. Fusobacterium nucleatum and T Cells in Colorectal Carcinoma. JAMA Oncol. 2015;1:653–61. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [19].Mima K, Nishihara R, Qian ZR, Cao Y, Sukawa Y, Nowak JA, et al. Fusobacterium nucleatum in colorectal carcinoma tissue and patient prognosis. Gut. 2016;65:1973–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [20].Taieb J, Le Malicot K, Shi Q, Penault-Llorca F, Bouche O, Tabernero J, et al. Prognostic Value of BRAF and KRAS Mutations in MSI and MSS Stage III Colon Cancer. J Natl Cancer Inst. 2017;109. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [21].Tran B, Kopetz S, Tie J, Gibbs P, Jiang ZQ, Lieu CH, et al. Impact of BRAF mutation and microsatellite instability on the pattern of metastatic spread and prognosis in metastatic colorectal cancer. Cancer. 2011;117:4623–32. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [22].Yaeger R, Cercek A, Chou JF, Sylvester BE, Kemeny NE, Hechtman JF, et al. BRAF mutation predicts for poor outcomes after metastasectomy in patients with metastatic colorectal cancer. Cancer. 2014;120:2316–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [23].Kopetz S, Grothey A, Van Cutsem E, Yaeger R, Wasan H, Yoshino T, et al. LBA-006BEACON CRC: a randomized, 3-Arm, phase 3 study of encorafenib and cetuximab with or without binimetinib vs. choice of either irinotecan or FOLFIRI plus cetuximab in BRAF V600E–mutant metastatic colorectal cancer. Annals of Oncology. 2019;30. [Google Scholar]
- [24].Andre T, Shiu K-K, Kim TW, Jensen BV, Jensen LH, Punt CJA, et al. Pembrolizumab versus chemotherapy for microsatellite instability-high/mismatch repair deficient metastatic colorectal cancer: The phase 3 KEYNOTE-177 Study. Journal of Clinical Oncology. 2020;38:LBA4–LBA. [Google Scholar]
- [25].FDA grants accelerated approval to pembrolizumab for first tissue/site agnostic indication. https://wwwfdagov/Drugs/InformationOnDrugs/ApprovedDrugs/ucm560040htm. 2017.
- [26].Sinicrope FA, Ou F-S, Shi Q, Nixon AB, Mody K, Levasseur A, et al. Randomized trial of FOLFOX alone or combined with atezolizumab as adjuvant therapy for patients with stage III colon cancer and deficient DNA mismatch repair or microsatellite instability (ATOMIC, Alliance A021502). Journal of Clinical Oncology. 2017;35:TPS3630–TPS. [Google Scholar]
- [27].Majek O, Gondos A, Jansen L, Emrich K, Holleczek B, Katalinic A, et al. Sex differences in colorectal cancer survival: population-based analysis of 164,996 colorectal cancer patients in Germany. PLoS One. 2013;8:e68077. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [28].Schmuck R, Gerken M, Teegen EM, Krebs I, Klinkhammer-Schalke M, Aigner F, et al. Gender comparison of clinical, histopathological, therapeutic and outcome factors in 185,967 colon cancer patients. Langenbecks Arch Surg. 2020;405:71–80. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [29].Hendifar A, Yang D, Lenz F, Lurje G, Pohl A, Lenz C, et al. Gender disparities in metastatic colorectal cancer survival. Clin Cancer Res. 2009;15:6391–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [30].Holch JW, Ricard I, Stintzing S, Modest DP, Heinemann V. The relevance of primary tumour location in patients with metastatic colorectal cancer: A meta-analysis of first-line clinical trials. Eur J Cancer. 2017;70:87–98. [DOI] [PubMed] [Google Scholar]
- [31].Arnold D, Lueza B, Douillard JY, Peeters M, Lenz HJ, Venook A, et al. Prognostic and predictive value of primary tumour side in patients with RAS wild-type metastatic colorectal cancer treated with chemotherapy and EGFR directed antibodies in six randomized trials. Ann Oncol. 2017;28:1713–29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- [32].Cercek A, Braghiroli MI, Chou JF, Hechtman JF, Kemeny N, Saltz L, et al. Clinical Features and Outcomes of Patients with Colorectal Cancers Harboring NRAS Mutations. Clin Cancer Res. 2017;23:4753–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.