Field Synopsis of Environmental and Genetic Risk Factors of Sporadic Early-Onset Colorectal Cancer and Advanced Adenoma

This article is featured in Highlights of This Issue, p. 995

Colorectal cancer is the third most common cancer in terms of incidence and the second leading cause of cancer-related death in the world (1). Although the incidence of colorectal cancer in populations of all ages in many countries has remained stable or decreased largely due to increased and more effective colorectal cancer screening (2), the incidence of early-onset colorectal cancer (EOCRC) has been rising worldwide, which might be attributed to a birth cohort effect after 1950 (3).

In general, EOCRC refers to cases diagnosed before the age of 50 and this cutoff is based on recommendations for colorectal cancer screening in high-income countries (4). It has been suggested that EOCRC differs from late-onset colorectal cancer (LOCRC) in terms of several factors. For example, prior studies have shown that EOCRCs are more commonly left-sided and present with rectal bleeding and abdominal pain (5). Furthermore, the histologic characteristics of EOCRC are more likely to be mucinous and signet-ring histology with reasons unknown (6). Generally, EOCRC is more likely to be diagnosed at advanced stage and has poorer cell differentiation at diagnosis (7, 8). Even though the 5-year cancer-specific survival of EOCRC and LOCRC is somewhat comparable, EOCRC is associated with a higher risk of tumor metastasis and recurrence (6, 9).

The etiology and reasons for the increasing incidence of EOCRC are unclear and are likely to be multifactorial, including interactions between genetic and environmental risk factors. For example, pathogenic germline variants have been demonstrated to be related to hereditary EOCRC risk (10). Although hereditary syndromes play a crucial role in EOCRC risk (9, 11), most cases are sporadic without identifiable cause and hereditary syndromes are unlikely to fully account for the increasing incidence of sporadic EOCRC (12). In recent years, multiple studies have demonstrated strong associations between obesity in children, adolescents, or young adults and risk of sporadic EOCRC in several countries including Australia, the USA, and Germany (13, 14). Other risk factors for EOCRC include unhealthy lifestyle factors such as an unhealthy diet (e.g., high intake of processed meat, low fiber diet; ref. 15), physical inactivity (16), and smoking (17).

Given the increasing incidence of colorectal cancer in younger adults in whom colorectal cancer screening is generally not recommended, synthesizing evidence on key risk factors for EOCRC may be relevant for enhanced tailored primary prevention and personalized screening for colorectal cancer in this population of special interest. Some previous meta-analyses have partly addressed this knowledge gap (18, 19), but they mostly focused on environmental risk factors or did not consider EOCRAs which are the precursors of most colorectal cancers. In this study, we aim to search, appraise, and synthesize available epidemiologic evidence on the associations between environmental or genetic factors and risk of EOCRC and EOCRA.

The protocol for this study was registered in PROSPERO (registration number CRD42021269993), and the study was reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines.

We conducted a comprehensive search in MEDLINE (1946–) and EMBASE (1974–) databases from inception to February 9, 2023. All identified publications went through a two-step parallel review (performed by R. Zhang and N. Yang) based on predefined selection criteria. Further details of the search strategy and selection criteria can be found in Supplementary Method. EOCRC was defined as colorectal cancer cases diagnosed before the age of 50 years (20). EOCRA was defined as advanced colorectal adenomas diagnosed before the age of 50 years. Advanced adenomas included an adenoma ≥1 cm or adenoma with high-grade dysplasia or villous/tubulovillous histologic elements (21).

For all the eligible studies, we extracted information on basic study characteristics including author, year of publication, country, study design, the definition of early-onset, the number of cases (and controls), sample size, type of risk factors examined, the reported risk estimates (ORs, RRs, or HRs) and corresponding 95% CIs, as well as the covariates that were adjusted for. In case a study had multiple control groups, we extracted estimates that were based on healthy controls as the control group. Also, in case a study had different levels of statistical adjustment, estimates from the most comprehensively adjusted model were extracted. Data extraction was conducted independently by three investigators (R. Zhang, N. Yang and Y. Zhou). All the extracted information was checked for accuracy by two other investigators (X. Zhou and D. Boakye).

A list of genetic variants to be summarized using meta-analysis was generated from the included studies, which reported genetic associations with EOCRC. Additional genotypic data on these genetic variants from the UK Biobank were incorporated to fully examine their associations using a nested case–control design with two control groups, healthy population and LOCRC (colorectal cancer cases diagnosed at age >50 years), which were chosen to minimize the risk of false-positive findings. Details of the UK Biobank study and similar approaches to EOCRC analysis in the UK Biobank have been published previously (22, 23).

We assessed the quality of case–control studies and cohort studies using the Newcastle-Ottawa Quality Assessment Scales. The NIH Quality Assessment Tool was used for cross-sectional studies. Two authors (N. Yang and R. Zhang) rated the quality of the studies independently.

We conducted meta-analyses for environmental risk factors with at least two primary studies. For each risk factor, we estimated several metrics, including (i) the summary effect and 95% CIs using the random-effects model (DerSimonian Laird method), (ii) the heterogeneity among studies using the Q statistic and I² metric, (ii) the 95% prediction interval (PI), (iii) the presence of small study effects using the Egger regression asymmetry test, and (iv) the excess significance test. For genetic variants, pooled ORs and 95% CIs were calculated for their associations with EOCRC compared with healthy controls and LOCRC patients, respectively, using allelic, recessive, and dominant genetic models. The Q statistic and I² metric were calculated to quantify heterogeneity. Funnel plot analysis with an Egger test was conducted to test for the small study effect. Statistical power was estimated by the Power and Sample Size Program (24). Bayesian false-discovery probability (BFDP) was calculated to assess the credibility of the observed associations. All statistical analyses were performed using the “metafor” packages in R (version 4.0.2). For the studies that could not be included in the meta-analysis, we synthesized the evidence thematically and reported the results narratively.

The evidence credibility was assessed according to the criteria reported in Supplementary Method. Observational associations were categorized into four categories considering several metrics as described previously (14). For genetic variants (24), we assessed the credibility of genetic association by using the BFDP (25) and the Venice criteria (26).

Researchers can request for the data used from the UK Biobank (www.ukbiobank.ac.uk/).

The manuscript did not previously publish or be under consideration for publication elsewhere. And the ethical approval was granted for the UK Biobank by the North West-Haydock Research Ethics Committee (REC reference: 16/NW/0274).

Figure 1 shows the results of the literature search. A total of 61 studies were included, of which 50 articles examined environmental risk factors and 11 articles investigated genetic risk factors. Of them, 44 articles provided enough data to perform meta-analyses. Supplementary Table S1 shows the basic characteristics of the included studies, including 27 case–control studies, 25 cohort studies, and 9 cross-sectional studies. A total of 47 studies were on EOCRC, whereas 14 were on EOCRA. We classified the examined risk factors into six categories: sociodemographic factors (n = 8), anthropometric factors (n = 19), personal medical history or family history (n = 34), medication use (n = 11), lifestyle factors (n = 28), and genetic factors (n = 11). The details of the quality assessment are summarized in Supplementary Table S2. The main results could be found in the graphical abstract (Supplementary Fig. S1).

We identified 11 studies investigating the association between sociodemographic factors and the risk of EOCRC or EOCRA. We defined high-level education as education up to college level or higher (with low-level education as the reference group) in the meta-analysis. Four studies and two studies examining the association between education level and risk of EOCRC and EOCRA, respectively, were included in the meta-analysis. The results of the meta-analysis demonstrated no significant association between education levels and the risk of EOCRC or EOCRA. A meta-analysis of three studies exploring the association between race or ethnicity and risk of EOCRC reported that Caucasian (vs. non-Caucasian; pooled OR = 1.59; 95% CI, 1.36–1.85) and African-American individuals (vs. non-African-American; pooled OR = 1.18; 95% CI, 1.04–1.35) had higher odds of EOCRC (Figs. 2 and 3). In a narrative synthesis of the studies that were not included in the meta-analysis, living with a spouse and colonoscopy screening were associated with a lower risk of EOCRA. Asians and Hispanics had a lower risk of EOCRA than Whites. As for EOCRC, 1–2 times CT scan was associated with a lower risk of EOCRC, whereas farming as occupation (vs. nonfarmers) was associated with an increased risk of EOCRC (Supplementary Tables S3 and S4).

We identified 19 articles examining the relationship between anthropometric factors and EOCRC or EOCRA risk. Regarding EOCRC, the meta-analysis of 9 studies showed a positive but nonstatistically significant association between current obesity and EOCRC (pooled OR = 1.39; 95% CI, 0.99–1.94). Restricting the analysis to cohort studies showed a stronger association (pooled OR = 1.97; 95% CI, 1.38–2.83). Meta-analysis of three studies showed a 37% increased EOCRC risk among participants who were overweight in adolescence (pooled OR = 1.37; 95% CI, 1.15–1.63) and the association was stable when restricting the analysis to cohort studies (pooled OR = 1.41; 95% CI, 1.17–1.71). No significant association was found between obesity in adolescence and EOCRC risk, whereas restricting the analysis to two cohort studies yielded a significant positive association (pooled OR = 1.48; 95% CI, 1.11–1.96). High waist circumference was defined as ≥90 cm for men and ≥80 cm for women (normal circumference as the reference group) in our study. The result of meta-analysis bespeaks high waist circumference was associated with an increased risk of EOCRC (pooled OR = 1.17; 95% CI, 1.01–1.34; Fig. 2). A total of 6 articles explored the relationship between overweight and EOCRA and meta-analysis showed a positive association (pooled OR = 1.33; 95% CI, 1.16–1.51; Fig. 3).

In the narrative synthesis of other studies, one prospective cohort study reported that weight gain of more than 40 kg since age 18 (vs. loss or gain <5.0 kg) was associated with a higher risk of EOCRC (RR = 2.15; 95% CI, 1.01–4.55). By contrast, one retrospective cohort study showed that weight loss (vs. stable weight) was associated with over 7-fold increased risk of EOCRC (OR = 7.43; 95% CI, 6.77–8.15). One nested case–control study suggested that a waist-to-hip ratio of 0.73–0.78 (vs. <0.72) was associated with a lower risk of EOCRC (OR = 0.45; 95% CI, 0.22–0.92). One cohort study showed that body surface area (per m²) was associated with EOCRC risk (RR = 3.40; 95% CI, 3.30–3.50; Supplementary Table S3). Regarding EOCRA, one cross-sectional study showed a positive association between abdominal obesity and EOCRA risk (OR = 1.28, 95% CI: 1.05–1.57; Supplementary Table S4).

We identified 34 studies examining the relationship between personal medical history or family history of colorectal cancer among first-degree relatives with EOCRC or EOCRA risk. In terms of EOCRC, 8 studies exploring the association between family history of colorectal cancer and EOCRC risk were included in the meta-analysis. Results showed a strong positive association between family history of colorectal cancer and EOCRC risk (pooled OR = 5.81; 95% CI, 2.91–11.61). The association became stronger when restricting the analysis to cohort studies (three studies, pooled OR = 6.64; 95% CI, 1.98–22.25). Meta-analysis of two studies showed a 51% increased EOCRC risk among participants with metabolic syndrome (two studies, pooled OR = 1.51; 95% CI, 1.05–2.19; Fig. 2). Regarding EOCRA, four risk factors were identified in meta-analyses: family history of colorectal cancer (seven studies, pooled OR = 1.31, 95% CI, 1.14–1.50), hypertension (six studies, pooled OR = 1.22; 95% CI, 1.05–1.41), hyperlipidemia (four studies, pooled OR = 1.34; 95% CI, 1.01–1.79) and metabolic syndrome (three studies, pooled OR = 1.37; 95% CI, 1.15–1.64). However, when restricting the analysis to cohort studies, no significant association was found between family history of colorectal cancer and EOCRA risk (Fig. 3).

In a narrative synthesis of other studies, one retrospective study reported that abdominal pain (OR = 4.73; 95% CI, 4.49–4.98), rectal pain (OR = 7.48; 95% CI, 6.42–8.72), altered bowel function (OR = 5.51; 95% CI, 5.19–5.85), rectal bleeding (OR = 9.83; 95% CI, 9.12–10.60), and colitis (OR = 4.10; 95% CI, 3.79–4.43) were positively associated with EOCRC risk. Furthermore, patients with iron-deficiency anemia (HR = 10.81; 95% CI, 8.15–14.33), hematochezia (HR = 10.66; 95% CI, 8.76–12.97), and chronic kidney disease (HR = 3.70; 95% CI, 1.83–7.49) were had a significantly higher risk of developing EOCRC. Another retrospective study reported that those with allergy or asthma (OR = 0.62; 95% CI, 0.39–0.98), hyperthyroidism (OR = 0.67; 95% CI, 0.48–0.94), and with higher parity (HR = 10.81; 95% CI, 8.15–14.33) had a lower risk of EOCRC (Supplementary Table S3). Regarding EOCRA, pelvic irradiation (OR = 12.8; 95%CI, 1.33–122) and anemia (OR = 3.11; 95% CI, 1.32–7.34) were reported as risk factors, whereas previous use of screening sigmoidoscopy, colonoscopy, or barium enema (OR = 0.26; 95% CI, 0.07–0.98) was associated with lower risk of EOCRA (Supplementary Table S4).

We identified 11 studies investigating the association between medication use and EOCRC or EOCRA risk. For EOCRC, it was demonstrated that ever use of aspirin or nonsteroidal anti-inflammatory drugs (NSAIDs) had inversely but nonstatistically significant association with EOCRC risk (pooled OR = 0.89; 95% CI, 0.71–1.11) in the meta-analysis of three studies. On the contrary, the meta-analysis of three studies found a positive but nonstatistically significant association between antibiotic use and EOCRC (pooled OR = 1.17; 95% CI, 0.97–1.42). For EOCRA, no significant associations were found between the use of aspirin/NSAIDs (five studies, pooled OR = 1.18; 95% CI, 0.90–1.55) or statin (two studies, pooled OR = 1.20; 95% CI, 0.86–1.67) and risk of EOCRA (Figs. 2 and 3).

In a narrative synthesis of other studies, one retrospective study reported that tetracyclines use (OR = 1.15; 95% CI, 1.02–1.29) and quinolones use (OR = 1.52; 95% CI, 1.29–1.78) were positively associated with EOCRC risk (Supplementary Tables S3).

A total of 28 studies explored the association between lifestyle factors (i.e., cigarette smoking, alcohol consumption, dietary supplement use, and physical activity) and EOCRC or EOCRA risk.

In our meta-analyses of all studies combined, cigarette smoking (seven studies, pooled OR = 1.62; 95% CI, 1.26–2.07), alcohol consumption (six studies, pooled OR = 1.49; 95% CI, 1.28–1.74), sedentary lifestyle (three studies, pooled OR = 1.42; 95% CI, 1.00–2.01), sugary beverage intake (two studies, pooled OR = 2.58; 95% CI, 1.61–4.13), and red meat intake (three studies, pooled OR = 1.12; 95% CI, 1.07–1.17) were significantly associated with EOCRC risk. The association with smoking became stronger when the analysis was restricted to cohort studies (three studies, pooled OR = 2.34; 95% CI, 1.99–2.76). By contrast, vitamin D intake (three studies, pooled OR = 0.70; 95% CI, 0.59–0.95), folate intake (three studies, pooled OR = 0.77; 95% CI, 0.60–0.99), and calcium intake (three studies, pooled OR = 0.82; 95% CI, 0.68–1.00) were inversely associated with EOCRC risk. The association with fruits and vegetables was based on three studies, all using the food frequency questionnaire to assess fruit and vegetable intake. Even though fruit and vegetable intake were inversely associated with EOCRC risk, the associations were not statistically significant (Fig. 2). Concerning EOCRA, smoking (six studies, pooled OR = 1.56; 95% CI, 1.27–1.92) was associated with an increased risk of EOCRA and the association was stronger when the analysis was restricted to cohort studies (two studies, pooled OR = 2.38; 95% CI, 1.41–4.02; Fig. 3).

In a narrative synthesis of the studies that were not included in the meta-analysis, a significantly higher risk of EOCRC or EOCRA was associated with a high westernized dietary pattern score, high sulfur microbial diet score, and western dietary pattern. In contrast, β-carotene supplements, vitamin C supplements, vitamin E supplements, prudent dietary pattern, dietary approaches to stopping hypertension, alternative Mediterranean dietary pattern, alternative healthy eating index-2010 (AHEI-2010), and Chinese food pagoda were all associated with reduced risk of EOCRC or EOCRA. On the contrary, higher sulfur microbial diet score and western dietary pattern were associated with increased risk of EOCRC or EOCRA (Supplementary Tables S3 and S4).

A total of 11 studies explored the associations between 62 genetic variants and EOCRC risk (Supplementary Table S5). Using genetic data from the UK Biobank, we examined the associations between these reported genetic variants and EOCRC risk (N_case = 618) in three genetic models (Supplementary Tables S6 and S7). Of them, only two variants rs4939827 (located within SMAD7, OR = 0.78; 95% CI,0.69–0.89, P = 1.59×10⁻⁴) and rs961253 (intergenic variant, OR = 1.24; 95% CI,1.09–1.41, P = 9.45×10⁻⁴) showed nominally significant associations with EOCRC in the allelic model. When meta-analysis was conducted to synthesize data from published studies and those from the UK Biobank, no statistically significant associations with EOCRC risk were observed in any of the three genetic models after correction for multiple testing. The summary crude ORs and 95% CI for the allelic, dominant, and recessive models are presented in Table 1.

We applied our evidence classification criteria to grade the level of evidence from the included studies (Figs. 2 and 3). Based on the metrics of evidence grading, no environmental factor presented convincing evidence; six factors for EOCRC and four factors for EOCRA presented highly suggestive evidence. The remaining eight and two statistically significant factors presented weak evidence for EOCRC and EOCRA, respectively. For genetic factors, none of the associations had suggestive evidence and all of them were “non-significant.” The detailed summary statistics of the highly suggestive risk factors for EOCRC and EOCRA are shown in Figs. 4 and 5.

In this study, we conducted a systematic review and meta-analysis of observational studies to comprehensively assess the role of genetic and environmental factors in EOCRC and EOCRA risk. A total of 120 environmental factors and 62 genetic factors were thoroughly investigated in this study. Our meta-analysis demonstrated that 12 factors (current overweight, overweight in adolescence, high waist circumference, smoking, alcohol, sugary beverage intake, sedentary behavior, red meat intake, family history of colorectal cancer, hypertension, hyperlipidemia, and metabolic syndrome) were associated with increased risk of EOCRC or EOCRA. By contrast, higher levels of reported intake of calcium, folate, and vitamin D were associated with a reduced risk of EOCRC. Recent data suggest that the risk factors of EOCRC are similar to those for colorectal cancer in the general population. However, most studies investigating risk factors for EOCRC have focused on traditional risk factors for colorectal cancer in general. Future studies should determine whether the changing patterns of traditional risk factors and/or increasing prevalence of less studied risk factors are contributing to the apparent rising incidence of EOCRC.

In our study, Caucasian and African-Americans had higher odds of EOCRC, while Asians and Hispanics had a lower risk of EOCRA than Whites. This finding is generally consistent with a population-based study (27) of elderly Medicare enrollees which reported the interval colorectal cancer risk was 31% higher in Blacks compared with Whites while was lower among Asians. Given the limited studies investigating ethnicity and race as risk factors for EOCRC, future studies should be addressing this important question for a more comprehensive conclusion.

Family history of colorectal cancer in at least one first-degree relative is an established risk factor for colorectal cancer in the general population (28) and our meta-analysis indicated that family history of colorectal cancer is also a strong risk factor for both EOCRC and EOCRA. The association could be related to the high prevalence of the mutation in the high-penetrance cancer-susceptibility genes. Given this, the National Comprehensive Cancer Network recommends genetic risk counseling and evaluation for EOCRC patients (12). Individuals with family history of colorectal cancer are regarded as high risk, and it is recommended that they undergo colonoscopy screening more frequently or at an earlier age than the general population (29). Metabolic syndrome was identified as a risk factor for EOCRC in our meta-analysis, but only a few studies were included in this analysis. Aspirin has been shown in previous studies to have the potential for the chemoprevention of colorectal cancer, and our study also demonstrated that its use is inversely associated with EOCRC risk (30). A previous RCT study has shown that regular use of aspirin at or after age 70 years is associated with a lower risk of colorectal cancer (31). Given that the evidence regarding the inverse association with colorectal cancer risk has mostly been examined in the general population, further studies focusing on populations younger than 50 years are required for a more comprehensive conclusion on the potential for aspirin to reduce EOCRC risk. Oral antibiotic is known to impact the gut microbiome and long-term use is probably a risk factor not only for colorectal cancer but also for colorectal adenomas (32, 33). Even though there are limited studies on its impact, oral antibiotics may confer an impetus on EOCRC risk. Further studies are needed to confirm this link.

Regarding lifestyle factors, compared with nondrinkers, alcohol drinking was identified as a common risk factor for both EOCRC and EOCRA in our meta-analysis. A previous study has found that alcohol consumption is associated with colorectal cancer risk in patients of all ages (34). Despite a lack of large studies and rigorous research, alcohol consumption is suspected to be associated with EOCRC risk (20). In one previous systematic review and meta-analysis, Sullivan and colleagues (18) also found alcohol as a risk factor for EOCRC. With the increasing alcohol consumption in many countries (35), future studies should assess the exact dose–response relationship between alcohol use and EOCRC/EOCRA risk.

Our study also demonstrated associations between sedentary behavior and EOCRC risk. The presence of a sedentary lifestyle was mostly defined as having any type of physical activity less than 1 hour per week (36). Sedentary behavior may result in energy imbalance (37) and progressively lead to being overweight or obese, a factor that was found to be associated with increased EOCRC/EOCRA risk in our study. Aside from obesity/overweight and sedentary behavior, high waist circumference was also identified as a risk factor for EOCRC in this meta-analysis. In a previous meta-analysis of six observational studies by Li and colleagues (19), obesity was also found to be associated with approximately 90% increased risk of EOCRC. The prevalence of obesity in the USA and many developed countries is increasing, especially among adolescents and young adults (38). The rising trend of obesity prevalence appears to correspond with the increased trend of EOCRC incidence. As an established risk factor for colorectal cancer in populations of all ages, the rising obesity rates among young adults and adolescents may partly explain the increasing incidence of EOCRC. In our study, being overweight in adolescence was also a risk factor for EOCRC. Liu and colleagues (39) conducted a cohort study exploring the association between weight change in adolescence and EOCRC and found that each 5-kg weight increase contributed to a 9% increased risk of EOCRC. Additionally, a nested case–control study (23) showed that having a higher weight or height than peers at the age of 10 was associated with an increased EOCRC risk. These findings suggest that early life exposures including being overweight among adolescents might be emerging risk factors for EOCRC. However, the number of studies examining these factors in relation to EOCRC risk was not sufficient for a meta-analysis. Future studies are needed to evaluate the association between early life exposures and EOCRC/EOCRA risk.

In our study, smoking was associated with an increased risk of EOCRC and EOCRA. Multiple studies have reported smoking as a risk factor for colorectal cancer among the general population (40). However, the prevalence of smoking is decreasing in many countries with the incidence of EOCRC rising (41). Furthermore, there remains debate about the impact of different times since starting smoking on the risk of colorectal cancer. Micronutrients such as calcium, vitamin D, and folate intake were inversely associated with EOCRC risk in our meta-analysis. This finding warrants further investigation to identify potential mechanisms explaining this association. Red meat intake and sugary beverage intake were considered risk factors for EOCRC in our study. Higher red or processed meat intake has been regarded as a risk factor for colorectal cancer in the general population (42, 43). Potential mechanisms explaining the associations between red and processed meat intake and colorectal cancer risk include gut epithelial damage and proliferation, DNA damage, and genotoxicity (44). Whether these mechanisms also hold for EOCRC is unclear and warrants further investigations. Previous studies (45, 46) have reported that sugar-sweetened beverage consumption was related to the incidence of colorectal cancer in general, which is consistent with our finding. However, studies exploring the potential mechanisms underpinning this association are still needed.

A major strength of our meta-analysis was the consideration of genetic factors and their associations with EOCRC risk. Two genetic variants (rs4939827 and rs961253) showed nominally significant associations with EOCRC in the allelic model using data from UK Biobank. Even though genetic polymorphisms alone may not necessarily indicate the presence of or susceptibility to colorectal cancer or colorectal adenoma, the presence of these and other risk factors of colorectal cancer (e.g., alcohol consumption, physical inactivity, and overweight) may require further clinical assessment and interventions, as a combination of these factors may have an additive and/or multiplicative effect on the risk of colorectal cancer in both the general population and younger adults (47, 48).

The starting age of colorectal cancer screening in most countries is more than 50 years. The 2021 American College of Gastroenterology colorectal cancer screening guidelines (49), however, suggested that colorectal cancer screening should be conducted in average-risk individuals between ages 45 and 49 years. Indeed, on the one hand, a recent study (50) suggested that starting colorectal cancer screening at the age of 45 years is likely to be more cost-effective and a greater benefit could be achieved by increasing participation rates for higher-risk individuals. On the other hand, it may be impractical or even not feasible to extend colorectal cancer screening to all young adults because of the health risk associated with colonoscopy (e.g., perforation, infection) and the financial costs of colonoscopy. Evaluating the key risk factors for EOCRC may nevertheless be useful for identifying high-risk groups, enabling the cost of screening to be reduced. The modifiable risk factors identified in our study could be useful for personalized colorectal cancer screening and prevention in adolescents and younger adults. Meanwhile, there is a need for future risk-scoring algorithms to consider these important risk factors and also quantify whether and to what extent these risk factors enhance the prediction of EOCRC beyond family history and other established risk factors. This would be valuable for enhanced prediction and risk stratification for EOCRC (51).

Our study has several strengths. To our knowledge, this field synopsis is the first to investigate both genetic and environmental risk factors of colorectal cancer in populations under the age of 50. Our study thus makes an important contribution to the limited evidence on risk factors for EOCRC, a topic that has gained much attention in the last few years. We searched multiple databases, and study selection and data extraction were done by multiple authors, minimizing the risk of study selection bias and data extraction errors. Our study also has limitations. First, there were a limited number of studies meeting our inclusion criteria and we included some articles that used differing definitions of both EOCRC and risk factors, which might contribute to the observed between-study heterogeneity in some of the associations. The limited number of studies precluded us from conducting a meta-regression analysis to rule out sources of heterogeneity. Second, different covariates were adjusted for in the included studies. Thus, the results of our meta-analysis may be affected by residual confounding due to unmeasured or less accurately measured factors. Lastly, publication bias (e.g., location bias, language bias, or selective outcome reporting) is a general potential weakness of almost all meta-analyses, including this study.

In conclusion, current evidence did not identify any bespoke risk factor of EOCRC likely due to the limited research investigating risk factors other than traditional risk factors for colorectal cancer in general. Future studies should determine whether the changing patterns of traditional risk factors and/or increasing prevalence of less studied risk factors (e.g., microbiome and early life exposure) are contributing to the apparent rising incidence of EOCRC. The potential for identifying risk factors to enhance the identification of at-risk groups for personalized EOCRC screening and prevention and for prediction of EOCRC risk aside from the family history of colorectal cancer should be comprehensively addressed by future studies.

D. Boakye reports other support from employment relationship outside the submitted work. A.C. Hamilton reports Bristol Myers Squibb—speaking services in connection with immunotherapy service development and practical management of immune-related toxicities, provided on April 25, 2023, at Belfast City Hospital. M.G. Dunlop reports grants from Cancer Research UK during the conduct of the study. E. Theodoratou reports grants from Cancer Research UK during the conduct of the study. No disclosures were reported by the other authors.

R. Zhang: Data curation, formal analysis, validation, investigation, writing–original draft. D. Boakye: Formal analysis, visualization, writing–original draft. N. Yang: Data curation, formal analysis, visualization. X. Zhou: Data curation, formal analysis, visualization, writing–original draft. Y. Zhou: Data curation, formal analysis, visualization. F. Jiang: Data curation, formal analysis, visualization. L. Yu: Formal analysis, visualization. L. Wang: Data curation, formal analysis, visualization, writing–original draft. J. Sun: Data curation, formal analysis, visualization. S. Yuan: Writing–review and editing. J. Chen: Writing–review and editing. A.C. Hamilton: Writing–review and editing. H.G. Coleman: Writing–review and editing. S.C. Larsson: Writing–review and editing. J. Little: Writing–review and editing. M.G. Dunlop: Writing–review and editing. E.L. Giovannucci: Writing–review and editing. E. Theodoratou: Writing–review and editing. X. Li: Conceptualization, supervision, funding acquisition, project administration, writing–review and editing.

X. Li was supported by the Natural Science Fund for Distinguished Young Scholars of Zhejiang Province (LR22H260001) and the National Nature Science Foundation of China (82204019). E. Theodoratou was supported by the CRUK Career Development Fellowship (C31250/A22804). This research was conducted using the UK Biobank study under Application Number 66354. We want to thank all UK Biobank participants and the management team for their participation and assistance.

The publication costs of this article were defrayed in part by the payment of publication fees. Therefore, and solely to indicate this fact, this article is hereby marked “advertisement” in accordance with 18 USC section 1734.