Survival in Young-Onset Metastatic Colorectal Cancer: Findings From Cancer and Leukemia Group B (Alliance)/SWOG 80405

Abstract

Background

The incidence of young-onset colorectal cancer (yoCRC) is increasing. It is unknown if there are survival differences between young and older patients with metastatic colorectal cancer (mCRC).

Methods

We studied the association of age with survival in 2326 mCRC patients enrolled in the Cancer and Leukemia Group B and SWOG 80405 trial, a multicenter, randomized trial of first-line chemotherapy plus biologics. The primary and secondary outcomes of this study were overall survival (OS) and progression-free survival (PFS), respectively, which were assessed by Kaplan-Meier method and compared among younger vs older patients with the log-rank test. Hazard ratios (HRs) and 95% confidence intervals (CIs) were calculated based on Cox proportional hazards modeling, adjusting for known prognostic variables. All statistical tests were 2-sided.

Results

Of 2326 eligible subjects, 514 (22.1%) were younger than age 50 years at study entry (yoCRC cohort). The median age of yoCRC patients was 44.3 vs 62.5 years in patients aged 50 years and older. There was no statistically significant difference in OS between yoCRC vs older-onset patients (median = 27.07 vs 26.12 months; adjusted HR = 0.98, 95% CI = 0.88 to 1.10; P = .78). The median PFS was also similar in yoCRC vs older patients (10.87 vs 10.55 months) with an adjusted hazard ratio of 1.02 (95% CI = 0.92 to 1.13; P = .67). Patients younger than age 35 years had the shortest OS with median OS of 21.95 vs 26.12 months in older-onset patients with an adjusted hazard ratio of 1.08 (95% CI = 0.81 to 1.44; P_trend = .93).

Conclusion

In this large study of mCRC patients, there were no statistically significant differences in survival between patients with yoCRC and CRC patients aged 50 years and older.

The overall incidence of colorectal cancer (CRC) has decreased by about 2% to 3% annually in recent years (1). This is because of various factors, including lifestyle changes and increased adherence to screening recommendations for average-risk individuals starting at age 50 years (2). Unfortunately, the reduction in CRC in the 50 years and older screened population has been partially offset by a concomitant increased incidence of CRC in younger patients (1). From 2000 to 2013, there was an approximately 22% increase in incidence of young-onset CRC (yoCRC), defined as CRC in individuals younger than age 50 years (1,3). Less than 10% of all CRC cases are attributed to known hereditary syndromes, though in patients younger than age 50 years, hereditary syndromes are found in up to 20% of patients (4,5). Beyond this subset, it is unclear if there are unique molecular features underlying this changing demographic. Conflicting data exist about whether metastatic yoCRC patients have better or worse prognosis compared with older patients.

Colorectal cancer is projected to become the second-leading cancer and the leading cause of cancer death in patients ages 20-49 years by the year 2040 (6). It is important to understand survival in this population. To investigate the prognosis of a large population of yoCRC patients with metastatic CRC (mCRC), we studied the association of age at study entry with overall survival (OS) and progression-free survival (PFS) among more than 2000 patients with mCRC enrolled in the Cancer and Leukemia Group B (CALGB) and SWOG 80405 (Alliance) trial. CALGB is now part of the Alliance for Clinical Trials in Oncology.

Methods

Study Population

This study population was derived from patients enrolled in the CALGB/SWOG 80405 trial (7). This multicenter, randomized phase III trial compared the efficacy of chemotherapy (fluorouracil, leucovorin, and oxaliplatin [mFOLFOX6] or fluorouracil, leucovorin, and irinotecan [FOLFIRI]) in combination with biologics (cetuximab and/or bevacizumab) for initial treatment of mCRC. The CALGB/SWOG 80405 trial included patients from multiple academic and community medical centers in the United States and Canada. Institutional review board approval was received at all participating centers. Written informed consent was obtained for all participants. Subjects were enrolled between October 2005 and March 2012. In October 2008, after demonstration of lack of efficacy of anti–epidermal growth factor receptor therapy in KRAS-mutant CRC, the trial restricted enrollment to patients with KRAS wild-type (WT) tumors (8). The CALBG/SWOG 80405 trial methods and results have previously been published (7).

Subjects were defined as having yoCRC if they were younger than age 50 years at the time of study entry. The choice of age 50 years was driven by the historical American Cancer Society and US Preventive Services Task Force (USPSTF) recommendation for the starting age for routine CRC screening (9,10). The USPSTF recently updated its guidelines for CRC screening and now recommends beginning screening at age 45 years (11,12). In addition to reporting our primary findings using age 50 years as a cut-point, as has been defined in other recent publications studying yoCRC, we also performed analyses exploring subsets of even younger patients (13–19).

Study Endpoints

The primary endpoint of our study was OS, defined as time from study entry until death from any cause. Secondary outcomes included PFS, defined as time from study entry until progression of disease or death from any cause (whichever was first); objective response rate (ORR), defined as proportion of patients demonstrating either a complete or partial response per Response Evaluation Criteria in Solid Tumors (RECIST) 1.0; and safety, assessed using the National Cancer Institute Common Toxicity Criteria version 3.0 (20,21).

Molecular Markers

Microsatellite instability (MSI) vs microsatellite stability was assessed using polymerase chain reaction or next-generation sequencing as previously described (22,23). To detect mutations in APC, BRAF, KRAS, and TP53, tumor DNA was extracted from formalin-fixed, paraffin-embedded tumor tissues using QIAamp DNA kits (QIAGEN, Hilden, Germany) and detected using allele-specific polymerase chain reaction (23,24).

Statistical Analysis

Though primary analysis for the trial was conducted on only 1137 patients who had proven KRAS WT CRC, our study includes the 2326 of 2334 patients who provided consent to full data use (see Figure  1) (7).

Figure 1.

Study cohort diagram. Of the 3058 patients with colorectal cancer registered through the Cancer and Leukemia Group B/SWOG 80405 trial, 724 were excluded, and 8 patients withdrew consent for complete data use in this study. The remaining 2326 patients were included in this study analysis.

Baseline characteristics according to age group were compared using the Wilcoxon rank sum test for continuous variables and χ² test for categorical variables (25,26). Median OS and PFS were calculated based on the Kaplan-Meier method and compared between young vs older patients using the log-rank test (27,28). Hazard ratios (HRs) and 95% confidence interval (CI) calculations were based on Cox proportional hazards modeling (29). In addition to analyzing survival in 2 age groups (younger than 50 years vs 50 years and older), we also subdivided younger patients into smaller categories (younger than 35, 35-39.9, 40-44.9, 45-49.9 years) for further analysis. We tested for linear trend by using age categories as an ordinal variable in a proportional hazards model. The assumption of proportional hazards was assessed graphically and using time varying covariates; the assumption was met. The hazard ratios were adjusted a priori for covariates known to be confounding or prognostic variables, including sex (male vs female), race (White vs Black vs other), Eastern Cooperative Oncology Group (ECOG) performance status (0 vs 1 to 2), primary tumor location (right and transverse colon vs left colon vs unknown), primary tumor unresected (no vs yes), prior radiation (no vs yes), prior adjuvant chemotherapy (no vs yes), KRAS status (WT vs mutant vs unknown), self-reported diabetes (no vs yes), body mass index in kg/m² (<21 vs 21 to <25 vs 25 to <30 vs 30 to <35 vs ≥35), chemotherapy backbone (FOLFIRI vs mFOLFOX6), and assigned trial treatment arm (bevacizumab vs cetuximab vs dual-antibody therapy). Any missing values for covariates, except KRAS status and primary tumor location, were recoded into the majority category if the proportion of missing data was less than 5%. If the proportion of missing data for a variable was more than 5%, these missing values were coded in a separate category. Analysis of missing data related to diabetes has been previously described (30). We studied the frequency of dose reduction—defined, for a given drug, as the proportion of patients who had at least 1 dose reduction for the drug out of the number of patients who received that drug. In addition, relative dose intensity was analyzed—defined for each drug as the ratio of the sum of dose received to the sum of dose planned.

We performed subgroup analyses to explore the association of age with relevant clinicopathologic variables. Tests for statistical interaction were conducted using multivariable Cox proportional-hazards model with a cross-product term of young-patient indicator and the covariate of interest (29). Correction for multiple hypotheses was not performed, as all subgroup analyses were considered exploratory and hypothesis generating; conclusions were not made based on these analyses.

Data were collected by the Alliance Statistics and Data Center. Data quality was ensured through review by the Alliance Statistics and Data Center per Alliance policies. Statistical analyses were performed using SAS software (SAS/STAT User’s Guide, Version 9.4, SAS Institute, Cary, NC) on the study database frozen on January 18, 2018. P values in this study were 2-sided, and a P value less than .05 was considered statistically significant.

Results

Of 2326 eligible subjects, 514 (22.1%) were younger than age 50 years at study entry and were included in the yoCRC cohort (Table  1). Among yoCRC patients, the median age was 44.3  (interquartile range [IQR] = 39.3-47.6) years. Among patients aged 50 years and older, the median age was 62.5 (IQR = 56.4-69.7) years. There was no statistically significant difference in sex or body mass index between age groups. Patients with yoCRC were more physically active, with median 6.9 metabolic equivalent task hours compared with 2.9 metabolic equivalent task hours in older-onset patients (P < .001). Fewer patients with yoCRC had diabetes (7.4% vs 18.9%; P < .001). Younger patients had better performance status, though this was not statistically significant (61.9% with ECOG 0 vs 57.6%; P = .08). A higher percentage of yoCRC patients had left-sided primary tumors (62.5% vs 56.0%; P = .005). Though there were more White than Black patients in both age groups, there was a statistically significantly higher percentage of Black patients in the yoCRC cohort compared with patients aged 50 years and older (16.1% vs 10.8%; P < .001). Baseline characteristics for the youngest patients, younger than age 35 years, are included in Supplementary Table 1 (available online).

Table 1.

Baseline characteristics in young-onset vs older-onset metastatic colorectal cancer

Characteristic	Age younger than 50 years	Age 50 years and older	Total	P
	(n = 514)	(n = 1812)	(n = 2326)
Age, median (Q1-Q3)	44.3 (39.3-47.6)	62.5 (56.4-69.7)	59.1 (51.2-67.6)	<.001a
Sex, No. (%)				.73b
Female	218 (42.4)	753 (41.6)	971 (41.7)
Male	296 (57.6)	1059 (58.4)	1355 (58.3)
Race, No. (%)				<.001b
Black	83 (16.1)	195 (10.8)	278 (12.0)
White	388 (75.5)	1508 (83.2)	1896 (81.5)
Other	32 (1.38)	65 (2.8)	97 (4.2)
Missing	11 (0.47)	44 (1.89)	55 (2.4)
ECOG performance status, No.c (%)				.08b
ECOG 0	318 (61.9)	1043 (57.6)	1361 (58.5)
ECOG 1,2	196 (38.1)	769 (42.4)	965 (41.5)
Median BMI (Q1-Q3), kg/m2d	26.9 (23.5-31.6)	27.1 (23.8-31.1)	27.1 (23.7-31.2)	.98a
Median physical activity (Q1-Q3), MET-he	6.9 (1.8-22.5)	2.9 (0.4-10.0)	3.4 (0.6-12.9)	<.001a
Diabetes, No. (%)				<.001b
No	476 (92.6)	1470 (81.1)	1946 (83.7)
Yes	38 (7.4)	342 (18.9)	380 (16.3)
Prior adjuvant chemotherapy, No. (%)				.01b
No	458 (89.1)	1534 (84.7)	1992 (85.6)
Yes	56 (10.9)	278 (15.3)	334 (14.4)
Primary tumor unresected, No. (%)				.06b
No	343 (66.7)	1289 (71.1)	1632 (70.2)
Yes	132 (25.7)	399 (22.0)	531 (22.8)
Missing	39 (7.6)	124 (6.8)	163 (7.0)
Prior radiation, No. (%)				.96b
No	469 (91.2)	1652 (91.2)	2121 (91.2)
Yes	45 (8.8)	160 (8.8)	205 (8.8)
Protocol chemotherapy, No. (%)				.47b
FOLFIRI	111 (21.6)	419 (23.1)	530 (22.8)
mFOLFOX6	403 (78.4)	1393 (76.9)	1796 (77.2)
Assigned treatment arm, No. (%)				.37b
Bevacizumab	209 (40.7)	688 (38.0)	897 (38.6)
Cetuximab	198 (38.5)	699 (38.6)	897 (38.6)
Bevacizumab + cetuximab	107 (20.8)	425 (23.5)	532 (22.9)
Primary tumor location, No. (%)				.005b
Left	321 (62.5)	1015 (56.0)	1336 (57.4)
Right and transverse colon	147 (28.6)	638 (35.2)	785 (33.7)
Missing	46 (8.9)	159 (8.8)	205 (8.8)
KRAS mutation status, No. (%)				.44b
Wild-type	280 (54.5)	990 (54.6)	1270 (54.6)
Mutant	93 (18.1)	365 (20.1)	458 (19.7)
Missing	141 (27.4)	457 (25.2)	598 (25.7)
RAS mutation status, No. (%)				.64b
Wild-type	138 (26.8)	514 (28.4)	652 (28.0)
Mutant	102 (19.8)	407 (22.5)	509 (21.9)
Missing	274 (53.3)	891 (49.2)	1165 (50.1)
TP53 mutation status, No. (%)				.31b
Wild-type	112 (21.8)	474 (26.2)	586 (25.2)
Mutant	57 (11.1)	200 (11.0)	257 (11.0)
Missing	345 (67.1)	1138 (62.8)	1483 (63.8)
APC mutation status, No. (%)				.16b
Wild-type	135 (26.3)	569 (31.4)	704 (30.3)
Mutant	34 (6.6)	105 (5.8)	139 (6.0)
Missing	345 (67.1)	1138 (62.8)	1483 (63.8)
BRAF mutation status, No. (%)				.05b
Wild-type	155 (30.2)	576 (31.8)	731 (31.4)
Mutant	18 (3.5)	112 (6.2)	130 (5.6)
Missing	341 (66.3)	1124 (62.0)	1465 (63.0)
MSI status, No. (%)				.36b
MSS/MSI-L	161 (31.3)	617 (34.1)	778 (33.4)
MSI-H	8 (1.6)	44 (2.4)	52 (2.2)
Missing	345 (67.1)	1151 (63.5)	1496 (64.3)

^a

P values were calculated using the 2-sided Wilcoxon rank sum test for continuous variables. BMI = body mass index; FOLFIRI = leucovorin, fluorouracil, and irinotecan; MET-h = metabolic equivalent task hours; mFOLFOX6 = leucovorin, fluorouracil, and oxaliplatin; MSI-H = microsatellite-instability-high; MSI-L = microsatellite instability-low; MSS = microsatellite stable; Q = quartile.

^b

P values were calculated using the 2-sided χ² test for categorical variables with missing category excluded.

^c

Eastern Cooperative Oncology Group (ECOG) performance status (PS) of 0 indicates patient is fully active without restriction; PS of 1 indicates patient is restricted in physically strenuous activity but remains ambulatory and able to perform light or sedentary work; PS of 2 indicates patient is ambulatory and able to perform self-care but unable to perform work.

^d

Number of missing = 3 for BMI.

^e

Number of missing = 976 for physical activity.

The median follow-up time was 5.98  (95% CI = 5.82 to 6.21) years. There were no statistically significant differences in OS or PFS in yoCRC patients vs older-onset patients (Table  2). The median OS was 27.07 months in yoCRC patients vs 26.12 months in patients aged 50 years and older with an adjusted hazard ratio of 0.98 (95% CI = 0.88 to 1.10; P = .78). The median PFS in yoCRC patients was 10.87 months vs 10.55 months in patients aged 50 years and older, with adjusted hazard ratio of 1.02 (95% CI = 0.92 to 1.13; P = .67). The ORR was similar between age groups with ORR of 57.8% in yoCRC patients vs 55.7% in patients aged 50 years and older (P = .40).

Table 2.

Hazard ratios for overall survival and progression-free survival by age (n = 2326)

Outcome and analysis	Age younger than 50 years	Age 50 years and older	P
OS
No. of events/No. of patients	416/514	1557/1812
Median OS (95% CI), mo	27.07 (25.04 to 30.06)	26.12 (24.94 to 27.30)	.12a
Unadjusted HR (95% CI)	0.92 (0.82 to 1.02)	Referent	.12b
Multivariable aHR (95% CI)c	0.98 (0.88 to 1.10)	Referent	.78b
PFS
No. of events/No. of patients	473/514	1700/1812
Median PFS (95% CI), mo	10.87 (9.99 to 11.50)	10.55 (10.12 to 10.94)	.67a
Unadjusted HR (95% CI)	0.98 (0.88 to 1.08)	Referent	.67b
Multivariable aHR (95% CI)c	1.02 (0.92 to 1.13)	Referent	.67b
ORR, No. (%)	297 (57.8)	1009 (55.7)	.40d

^a

P values and associated median overall survival (OS) and progression-free survival (PFS) were calculated using the Kaplan-Meier method. All tests were 2-sided. aHR = adjusted hazard ratio; CI = confidence interval; FOLFIRI = leucovorin, fluorouracil, and irinotecan; HR = hazard ratio; mFOLFOX6 = leucovorin, fluorouracil, and oxaliplatin.

^b

P values for hazard ratios were calculated in corresponding Cox model. All tests were 2-sided.

^c

Adjusted with Cox proportional hazards analysis for patient sex (male vs female), race (White vs Black vs other), Eastern Cooperative Oncology Group performance status (0 vs 1 to 2), primary tumor location (right and transverse colon vs left colon vs unknown), primary tumor unresected (no vs yes), prior radiation (no vs yes), prior adjuvant chemotherapy (no vs yes), KRAS mutation status (wild-type vs mutant vs unknown), diabetes (no vs yes, as reported in a diet and lifestyle questionnaire), body mass index (BMI) at study entry (<21 vs 21 to <25 vs 25 to <30 vs 30 to <35 vs ≥35 kg/m²; 3 patients with missing BMI were recoded into the majority category in 25 to <30 kg/m²), protocol chemotherapy received (FOLFIRI vs mFOLFOX6), and arm of trial (bevacizumab vs cetuximab vs dual-antibody therapy).

^d

P value for the objective response rate (ORR) is based on 2-sided χ² test.

The youngest patients, younger than age 35 years, had a shorter median OS of 21.95 (95% CI = 17.15 to 32.82) months vs 26.12  (95% CI = 24.94 to 27.30) months in patients aged 50 years and older, with adjusted hazard ratio of 1.08 (95% CI = 0.81-1.44; P_trend = .93) (Table  3). Patients younger than age 35 years had a shorter median PFS of 9.33 (95% CI = 7.00 to 11.96) months vs 10.55 (95% CI = 10.12 to 10.94) months in older-onset CRC patients with adjusted hazard ratio of 1.22 (95% CI = 0.93 to 1.59; P_trend = .68). Although sample size was limited, there were no statistically significant differences in survival for patients younger than age 35 years by subgroups of race, with adjusted hazard ratio for OS of 0.80 (95% CI = 0.36 to 1.77; P = .58) and adjusted hazard ratio for PFS of 1.05 (95% CI = 0.52 to 2.14; P = .88) in White vs non-White patients.

Table 3.

Hazard ratios for overall survival and progression-free survival by age group of young-onset metastatic colorectal cancer patients (n = 2326).

Outcome and analysis	Age group, y					Pa
	<35	35-39.9	40-44.9	45-49.9	≥50
OS
No. of events/patients	49/59	61/84	119/145	187/226	1557/1812
Median OS (95% CI), mo	21.95 (17.15 to 32.82)	26.05 (22.54 to 38.70)	26.74 (22.57 to 31.64)	28.65 (25.04 to 31.97)	26.12 (24.94 to 27.30)	.53b
Unadjusted HR (95% CI)	1.02 (0.77 to 1.36)	0.87 (0.68 to 1.13)	0.93 (0.77 to 1.12)	0.90 (0.78 to 1.05)	Referent	.27a
Multivariable aHR (95% CI)c	1.08 (0.81 to 1.44)	0.91 (0.70 to 1.18)	1.01 (0.83 to 1.22)	0.97 (0.84 to 1.13)	Referent	.93a
PFS
No. of events/patients	56/59	71/84	133/145	213/226	1700/1812
Median PFS (95% CI), mo	9.33 (7.00 to 11.96)	10.35 (8.28 to 14.26)	11.37 (9.66 to 11.99)	10.91(9.50 to 12.32)	10.55 (10.12 to 10.94)	.48b
Unadjusted HR (95% CI)	1.17 (0.90 to 1.53)	0.90 (0.71 to 1.14)	0.90 (0.76 to 1.08)	1.02 (0.88 to 1.17)	Referent	.79a
Multivariable aHR (95% CI)c	1.22 (0.93 to 1.59)	0.94 (0.74 to 1.19)	0.94 (0.79 to 1.13)	1.07 (0.92 to 1.23)	Referent	.68a
ORR, No. (%)	34 (57.6)	44 (52.4)	88 (60.7)	131 (58.0)	1009 (55.7)	.70d

^a

P values and associated median OS and PFS were calculated using the 2-sided log-rank test. aHR = adjusted hazard ratio; CI = confidence interval; FOLFIRI = leucovorin, fluorouracil, and irinotecan; HR = hazard ratio; mFOLFOX6 = leucovorin, fluorouracil, and oxaliplatin; min = minimum; OS = overall survival; PFS = progression-free survival;.

^b

P values for corresponding Cox model were calculated with ordinal age categories to reflect any linear trend. All tests were 2-sided.

^c

Adjusted with Cox proportional hazards analysis for patient sex (male vs female), race (White vs Black vs other), Eastern Cooperative Oncology Group performance status (0 vs 1 to 2), primary tumor location (right and transverse colon vs left colon vs unknown), primary tumor unresected (no vs yes), prior radiation (no vs yes), prior adjuvant chemotherapy (no vs yes), KRAS mutation status (wild-type vs mutant vs unknown), diabetes (no vs yes, as reported in a diet and lifestyle questionnaire), body mass index (BMI) at study entry (<21 vs 21 to <25 vs 25 to <30 vs 30 to <35 vs ≥35 kg/m²; 3 patients with missing BMI were recoded into the majority category in 25 to <30 kg/m²), protocol chemotherapy received (FOLFIRI vs mFOLFOX6), and arm of trial (bevacizumab vs cetuximab vs dual-antibody therapy).

^d

P value for objective response rate (ORR) is based on a 2-sided χ²test.

Overall, the frequency of common oncogenic mutations in tumors from patients with yoCRC were similar to those aged 50 years and older. MSI status was missing for 1496 (64.3%) patients. Of patients with available data, there was no statistically significant difference in MSI status, with 8 (1.6%) MSI-high patients in the yoCRC cohort and 44 (2.4%) MSI-high patients in the older-onset cohort (P = .36). Additionally, there was no statistically significant difference in frequency of mutations in KRAS, BRAF, RAS, TP53, or APC. There was a slightly lower incidence of BRAF mutations in patients with yoCRC (3.5% vs 6.2%; P = .05). Regarding treatment, fewer yoCRC patients had received prior adjuvant chemotherapy (10.9% vs 15.3%; P = .01). There were no statistically significant differences in frequency of yoCRC patients who received prior radiation or had their primary tumor resected. In addition, there were no statistically significant differences in the assigned trial treatment arm or backbone chemotherapy received between the yoCRC and older-onset CRC cohorts. There were no statistically significant differences in the frequency of dose reduction by age (Supplementary Table 2, available online). Younger patients overall received higher doses of chemotherapy, as measured by median relative dose intensity, particularly for 5-fluorouracil (0.80 vs 0.75; P < .001) and oxaliplatin (0.75 vs 0.68; P < .001) (Supplementary Table 3, available online). Overall, yoCRC patients were less likely to experience adverse events than older patients (30.7% vs 43.8%; P < .001) (Supplementary Table 4, available online).

In subgroup analyses, there were no statistically significant differences in the relationship between young age and OS or PFS by sex, race, performance status, primary tumor sidedness, chemotherapy received, or assigned treatment arm (Figure  2). Our data showed no statistically significant differences in OS or PFS between yoCRC patients who were obese, overweight, normal weight, and underweight. Younger age was associated with improved OS for patients with KRAS WT mCRC (HR = 0.85, 95% CI = 0.73 to 0.99; P = .04), BRAF-mutant mCRC (HR = 0.39, 95% CI = 0.21 to 0.72; P = .002), and APC-mutant mCRC (HR = 0.56; 95% CI = 0.35 to 0.90; P = .02). Similarly, young age was associated with better PFS in patients with BRAF-mutant mCRC (HR = 0.61, 95% CI = 0.36 to 1.04; P = .07) and APC-mutant mCRC (HR = 0.74, 95% CI = 0.49 to 1.14; P = .17).

Figure 2.

Subgroup analyses of overall survival and progression-free survival in young-onset patients vs patients aged 50 years and older. Multivariable adjusted hazard ratios (HRs) and 95% confidence intervals (CIs) for overall survival and progression-free survival, comparing young-onset colorectal cancer patients with patients aged 50 years and older. Adjusted with Cox proportional hazards analysis for patient sex (male vs female), race (White vs Black vs other), Eastern Cooperative Oncology Group (ECOG) performance status (0 vs 1 to 2), primary tumor location (right and transverse colon vs left colon vs unknown), primary tumor unresected (no vs yes), prior radiation (no vs yes), prior adjuvant chemotherapy (no vs yes), KRAS mutation status (wild-type vs mutant vs unknown), diabetes (no vs yes, as reported in a diet and lifestyle questionnaire), body mass index (BMI) at study entry (see below), protocol chemotherapy received (leucovorin, fluorouracil, and irinotecan [FOLFIRI] vs leucovorin, fluorouracil, and oxaliplatin [mFOLFOX6]), and arm of trial assigned (bevacizumab vs cetuximab vs dual-antibody therapy). BMI categories were defined as obese (BMI ≥30 kg/m²), overweight (BMI = 25 to <30 kg/m²), normal weight (BMI = 18.5 to <25 kg/m²), and underweight (BMI <18.5 kg/m²). Three patients with missing BMI were recoded into the majority category in 25 to less than 30 kg/m². Error bars represent 95% confidence intervals. All statistical tests were 2-sided.

Discussion

We found no differences in OS or PFS between mCRC patients younger than age 50 years and mCRC patients aged 50 years and older who participated in a large, randomized trial. To our knowledge, this is the first clinical trial cohort analysis of survival outcomes in a large sample size of younger vs older patients with mCRC. Our findings add to a recent analysis of patients with stage III CRC, which showed that prognosis was not different in young vs older patients after adjusting for molecular markers (31). Previous data on mortality among yoCRC patients are sparse and conflicting (32–43). Some of this discrepancy may be related to differences in baseline characteristics, specifically race. Prior data suggest mortality rates are rising more steeply in young White patients than young Black patients (44). However, overall, Black patients with yoCRC have worse survival than White patients (45). Some studies show this may be because of access to care, whereas other studies suggest that differences in somatic mutation profile and tumor mutation burden may play a role (46–48). Although our study has a higher percentage of Black patients in the yoCRC cohort, we did not find statistically significant differences in OS or PFS between Black and White yoCRC patients.

The lack of difference in survival between yoCRC and older patients in this study is particularly interesting given the more favorable baseline characteristics, higher treatment dose intensity, and decreased incidence of adverse events in the yoCRC cohort. Younger patients had better performance status, higher levels of physical activity, less diabetes, and a higher proportion of left-sided primary tumors (49–51). The lack of association between age and survival was consistent across subgroups of assigned treatment arm and primary tumor location.

Other studies have also shown that younger patients receive more aggressive treatment than older patients with a similar stage of CRC (38,39,43,52,53). Younger patients are more likely to receive cancer-directed therapy, and among those who receive chemotherapy, younger patients are more likely to receive multi-agent treatment regimens (34,39). Prior studies suggest that younger CRC patients are more likely to undergo primary tumor resection and radiation therapy (39,43). Although our data did not show statistically significant differences in rate of primary tumor resection or prior radiation, this may reflect an overall late stage at diagnosis of patients in this trial.

Younger patients are more likely to present with advanced-stage disease (18,32,34,36,54,55). This is supported by our data, which show that fewer younger patients received prior adjuvant therapy, suggesting they were more likely to be diagnosed with de novo metastatic disease. The increased likelihood of metastatic disease at presentation is partly attributable to delays in diagnosis, and there is conflicting data on whether this is because of more aggressive biology (4,56). Young-onset CRC has previously been associated with mucinous and signet ring features, as well as poorly differentiated histology, whereas some recent data show no difference in tumor histopathology (54,56). Tumors from yoCRC patients have shown increased hypomethylation of long interspersed nucleotide element-1, which has been associated with poor prognosis (54,57). Additionally, RNA sequencing data from The Cancer Genome Atlas identified increased expression of the PEG10 gene—which is thought to affect cell proliferation and apoptosis—in younger CRC patients, though the prognostic implications of this are unclear (58,59).

It is interesting that the youngest patients in this trial, younger than age 35 years, had the shortest OS and PFS. This difference was not statistically significant, and future confirmation will require a larger sample size. However, this observation is consistent with existing retrospective data suggesting that very young patients have worse outcomes (60,61). Tumors from very young patients may have different and distinct molecular changes. For example, some data showed patients younger than age 30 years had fewer APC mutations and MAPK pathway mutations, had more ATM mutations, and were more likely to have signet ring histology than patients ages 30-50 years (62). In addition, other data from patients younger than age 40 years show fewer TP53 and CTNNB1 mutations (63). It is unknown if these differences impact survival. Nevertheless, survival of very young CRC patients warrants further investigation, particularly because the rate of rise of yoCRC has been steepest in the youngest age groups (64).

Our data show a lower prevalence of BRAF mutation in the yoCRC cohort, which is consistent with prior studies (62,65). Although BRAF mutation is known to confer an overall poor prognosis in mCRC, our data show improved OS for young patients with BRAF-mutant mCRC compared with older patients with BRAF-mutant mCRC (66,67). It is unknown whether these patients received treatment with BRAF-directed therapy after participation in this trial.

Existing data suggest that the frequency of mutations in KRAS and APC are lower in yoCRC (61,63). Our study did not find a difference in frequency of APC or KRAS mutation between age groups, although there was insufficient sample size to detect these differences. Our data show that yoCRC patients with APC-mutant tumors had a higher OS than older-onset patients with APC-mutant tumors. This may be related to the number or type of APC mutations in younger vs older patients, which our study did not assess (68). Additionally, we identified an increased OS for yoCRC patients with KRAS WT tumors compared with older patients with KRAS WT mCRC. As with all subgroup analyses, this finding needs further study and confirmation.

This study has several strengths. Because it is nested in a large clinical trial cohort, our investigation was conducted in a standardized patient population with regimented treatment procedures and follow-up per trial protocol. In addition, we studied a large sample size of mCRC patients and had available molecular and clinical data for correlative analysis. Limitations of this study include lack of several important data points, including age at diagnosis of CRC rather than age at study entry. If age at diagnosis was available, it is possible that some patients older than age 50 years at study entry would have instead been placed in the yoCRC category. Additionally, we do not have data on subsequent lines of therapy after progression on this trial, which may affect survival outcomes. Importantly, MSI status is missing for more than 60% of patients. Lastly, although the sample size was large, there were not enough patients to appropriately analyze the association of age with molecular data or the outcomes of patients younger than age 35 years by subgroups of clinically relevant variables.

In conclusion, our data show no survival benefit for patients with yoCRC despite their having traditionally favorable baseline characteristics such as better performance status, more physical activity, and more left-sided primary tumors. This may be because of diagnosis at more advanced stages, differences in underlying tumor biology as reflected in differences in molecular markers, or other, as yet unidentified factors increasing disease aggression in these patients. Additional investigation into the tumor biology, clinical characteristics, and optimal treatment of patients with yoCRC is essential.

Funding

This work was supported by the National Cancer Institute of the National Institutes of Health under award numbers U10CA180821 and U10CA180882 (to the Alliance for Clinical Trials in Oncology), U10CA180794, UG1CA189858, UG1CA233196, UG1CA233253, UG1CA233290, UG1CA233333, UG1CA233337, UG1CA180830, U10CA180888 (SWOG), and R01CA205406 (to KN); Department of Defense award number CA160344 (to KN).; the Project P Fund (to JAM and KN); and P30 CA008748 (Cancer Center Support Grant to EMOR). Also supported in part by funds from Bristol Myers Squibb, Genentech, Pfizer, and Sanofi.

Notes

Role of the funders: The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health. The funding sources had no role in study design, data collection, analysis or interpretation, or writing of the manuscript.

Disclosures: EMOR reported the following funds: Research Funding to Memorial Sloan Kettering: Genentech/Roche, Celgene/BMS, BioNTech, BioAtla, AstraZeneca, Arcus Consulting Role: Cytomx Therapeutics (DSMB), Rafael Therapeutics (DSMB), Sobi, Silenseed, Molecular Templates, Boehringer Ingelheim, BioNTech, Ipsen, Polaris, Merck, AstraZeneca, Bayer (spouse), Genentech-Roche (spouse), Celgene-BMS (spouse), Eisai (spouse).

JAM reports consulting/advising Array BioPharma. JAM reports receiving honoraria from Ignyta, institutional research funding from Boston Biomedical, has served as an advisor/consultant to COTA Healthcare, and served on a grant review panel for the National Comprehensive Cancer Network funded by Taiho Pharmaceutical. His research is supported by the Douglas Gray Woodruff Chair fund, the Guo Shu Shi Fund, Anonymous Family Fund for Innovations in Colorectal Cancer, Project P fund, and the George Stone Family Foundation.

KN reports receiving research funding from Pharmavite, Evergrande Group, Janssen, Revolution Medicines, Genentech/Roche, Gilead Sciences, Celgene, Trovagene, and Tarrex Biopharma; and consulting/advising for BiomX, X-Biotix Therapeutics, Seattle Genetics, Array Biopharma, Bayer, Lilly, Genentech/Roche, and Tarrex Biopharma. The other authors have no disclosures.

Author contributions: Conceptualization: ML-S, SZ, JAM, and KN; Formal analysis: ML-S, SZ, CM, KN; Funding acquisition: JAM, KN; Writing—original draft: ML-S, KN; Writing—review & editing: all authors. All authors approved the manuscript to be published.

Data Availability

The data underlying this article will be shared on reasonable request to the corresponding author.