Patterns in genomic mutations among patients with early-onset colorectal cancer: an international, multicohort, observational study

Abstract

Background

The increasing incidence of early-onset colorectal cancer (age <50 years; EOCRC) shows a dramatic growing trend globally, while late-onset colorectal cancer (LOCRC) is gradually decreasing. We aimed to characterise the distinct mutational landscape of EOCRC in an effort to inform age-specific clinical management.

Methods

In this observational study, we analysed whole-exome sequencing and clinical-grade targeted sequencing data from seven cohorts (the Memorial Sloan Kettering Cancer Center [MSKCC] cohort [the USA], the Leiden University Medical Center cohort [the Netherlands], the Nigerian African Research Group for Oncology [ARGO] cohort [Nigeria], the Genomics Evidence Neoplasia Information Exchange [GENIE] Project [Canada, France, Spain, and the USA], the Sun Yat-sen University Cancer Center (SYSUCC) cohort [China], the Asan Medical Center cohort [South Korea], and the Fudan University Shanghai Cancer Center–Colorectal Cancer [FUSCC-CRC] cohort [China]) across Canada, China, France, Nigeria, South Korea, Spain, the Netherlands, and the USA. Eligible patients were aged 18 years or older with a confirmed diagnosis of colorectal adenocarcinoma or mucinous adenocarcinoma. Samples were categorised into hypermutated (tumour mutational burden [TMB] >15 mutations per megabase) and non-hypermutated (TMB ≤15 mutations per megabase) groups. We evaluated the TMB difference between EOCRC and LOCRC using gamma regression and compared genomic mutation data between EOCRC and LOCRC using multiple logistic regression and pathway enrichment analysis. The primary study objective was to compare genomic mutational patterns between EOCRC and LOCRC, stratified by TMB groups.

Findings

Between Jan 1 and Dec 31, 2024, 17 133 tumour samples from patients with colorectal cancer in eight countries were analysed (Canada [n=218], China [n=3009], France [n=62], Nigeria [n=64], South Korea [n=44], Spain [n=250], the Netherlands [n=281], and the USA [n=13 205]). Among 17 133 patients, 9452 (55·2%) were male, 7681 (44·8%) were female, 10 174 (59·4%) were White, 3904 (22·8%) were Asian or Pacific Islander, 983 (5·7%) were Black, and 4983 (29·1%) had EOCRC. In hypermutated colorectal cancer, EOCRC exhibited a significantly higher TMB compared with LOCRC (mean ratio 1·11 [95% CI 1·06–1·16]; p<0·0001). In non-hypermutated colorectal cancer, EOCRC showed a significantly lower TMB than LOCRC after adjusting for skewness (mean ratio 2·92 [95% CI 2·88–2·96]; p<0·0001). In hypermutated colorectal cancer, a total of 23 genes, including APC (EOCRC 464 [75·0%] of 619 vs LOCRC 891 [58·6%] of 1521; odds ratio [OR] 2·00 [95% CI 1·59–2·51]; adjusted p<0·0001), KRAS (EOCRC 331 [53·3%] of 621 vs LOCRC 488 [32·0%] of 1526; OR 2·35 [95% CI 1·91–2·89]; adjusted p<0·0001), and CTNNB1 (EOCRC 196 [31·6%] of 621 vs LOCRC 274 [18·0%] of 1526; OR 2·15 [95% CI 1·70–2·72]; adjusted p<0·0001), and TCF7L2 (EOCRC 294 [51·2%] of 574 vs LOCRC 489 [35·0%] of 1398; OR 2·01 [95% CI 1·62–2·50]; adjusted p<0·0001), displayed elevated mutation frequencies in EOCRC compared with LOCRC, whereas only BRAF (EOCRC 97 [15·6%] of 621 vs LOCRC 674 [44·2%] of 1526; OR 0·27 [95% CI 0·21–0·35]; adjusted p<0·0001) and RNF43 (EOCRC 225 [39·3%] of 573 vs LOCRC 778 [53·9%] of 1444; OR 0·61 [95% CI 0·49–0·76]; adjusted p=0·0015) were less frequently mutated in EOCRC than in LOCRC. In non-hypermutated colorectal cancer, only TP53 (EOCRC 3468 [79·5%] of 4362 vs LOCRC 7825 [73·7%] of 10 624; OR 1·37 [95% CI 1·25–1·50]; adjusted p<0·0001) showed a higher mutation frequency in EOCRC, while nine genes had lower mutation frequencies, including BRAF (EOCRC 274 [6·3%] of 4362 vs LOCRC 909 [8·6%] of 10 624; OR 0·70 [95% CI 0·60–0·81]; adjusted p=0·00024) and KRAS (EOCRC 1794 [41·1%] of 4362 vs LOCRC 4840 [45·6%] of 10624; OR 0·83 [95% CI 0·77–0·89]; adjusted p=0·00019).

Interpretation

Within hypermutated colorectal cancer, younger patients exhibited a higher mutational burden than older patients. Our study reveals an abnormal accumulation of distinct somatic mutations in hypermutated EOCRC, the pattern of which might be contributing to the alarming rise in the incidence of EOCRC over the past decades. Our results support the need for EOCRC-specific molecular profiling to guide clinical practice.

Funding

National Natural Science Foundation of China and the Shanghai Science and Technology Development Fund.