We identified multiple studies analysing obesity-related outcomes: 16 cohort studies, two cross-sectional and six case-control studies.

Effect of obesity at diagnosis on eoCRC risk. Forest plot of a meta-analysis of studies analysing the association of obesity at cancer diagnosis and eoCRC (full study names in references). Each study (stratified by age where provided) shows multivariate-adjusted risk ratios (MVRR), number of cases and individuals at risk. The horizontal lines indicate confidence intervals. Boxes indicate estimates of adjusted risk ratios. The size of the boxes inversely correlates with the width of the confidence intervals. The last row shows the pooled adjusted risk ratio. eoCRC Obese: study population with early-onset colorectal cancer and a BMI superior or equal to 30 kg/m2. Total Obese: study population with a BMI superior or equal to 30 kg/m2. eoCRC non-obese: study population with early-onset colorectal cancer and a BMI inferior to 25 kg/m2. Total non-obese: study population with a BMI inferior to 25 kg/m2. *Pooled risk ratio estimate from subgroups **: estimation of population numbers by dividing person-years by the median time of follow-up ***Prospective open cohort study only providing person-years as quantifiable size of study population

Effect of overweight at diagnosis on eoCRC risk. Forest plot of a meta-analysis of studies analysing the association of overweight at cancer diagnosis and eoCRC (full study names in references). Each study (stratified by age where provided) shows multivariate-adjusted risk ratios (MVRR), number of cases and individuals at risk. The horizontal lines indicate confidence intervals. Boxes indicate estimates of adjusted risk ratios. The size of the boxes inversely correlates with the width of the confidence intervals. The last row shows the pooled adjusted risk ratio. eoCRC Overweight: study population with early-onset colorectal cancer and a BMI between 25 and 29.99 kg/m2. Total Overweight: study population with a BMI between 25 and 29.99 kg/m2. eoCRC non-overweight: study population with early-onset colorectal cancer and a BMI inferior to 25 kg/m2. Total non-overweight: study population with a BMI inferior to 25 kg/m2. *Pooled risk ratio estimate from subgroups **: estimation of population numbers by dividing person-years by the median time of follow-up ***Prospective open cohort study only providing person-years as quantifiable size of study population

Most studies investigated the association of obesity at CRC diagnosis with eoCRC. The pooled adjusted risk ratio shows a statistically significant 1.45-fold risk of eoCRC in patients with obesity compared with controls (Summary RR 1.45 [1.00–2.10.00.10]) with extremely high heterogeneity and substantial between-study variance (Fig. 1) [2, 18,19,20,21, 23, 26, 30,31,32,33,34,35]. Being overweight at onset was not statistically associated with risk of eoCRC (Summary RR 1.02 [0.71–1.46]) (Fig. 2) [18,19,20, 26, 31]. This association shows high heterogeneity with moderate variance. One study found obesity at the time of colonoscopy to be a protective factor (multi-variable adjusted OR (MVOR) 0.69 [0.55–0.86]). However, this study included only US veterans(mostly male) and did not include any data on family history of CRC [20].Compared with loCRC, eoCRC patients were significantly less likely to be overweight (MVOR 0.56 [0.41–0.76]) or obese (MVOR 0.66 [0.48–0.90]) [36].

Four studies investigated obesity before cancer diagnosis with eoCRC. Risk of eoCRC was associated with A BMI above the 85th percentile in late adolescence colon cancer for male individuals (Hazard Ratio (HR) 1.53 [1.17–2.00.17.00]); B rectal cancer for male individuals (HR 1.09 [0.68–1.73]); C BMI ≥ 30 kg/m² at age 20 and significantly at age 30 (MVOR 2.56 [1.18–5.44], 2.06 [1.25–3.40]); D each 5-unit increase of the BMI at 20 and significantly at 30 years (MVOR 1.44 [1.18–1.75], 1.36 [1.15–1.61]) with diverging results across studies; and E an obese phenotype 10 years before CRC diagnosis (MVOR 1.88 [1.30–2.73]) [4, 16, 17, 37,38,39].

Two studies examined weight gain since adulthood in female-only cohorts. One found no association, whereas the other reported that weight gain of more than 40 kg (MVRR) 2.15 [1.01–4.55]) and each 5 kg-weight increase since adolescence (MVRR 1.09 [1.02–1.16]) were associated with higher eoCRC risk among women [18, 19].

One study assessed longitudinal obesity and abdominal obesity trajectories and found that maintained overweight over a 2-year period (MVHR 1.09 [1.03–1.16]), and constant abdominal obesity (defined by the Korean Society for the Study of Obesity [40]) (multivariate-adjusted HR (MVHR) 1.18 [1.09–1.29]) were independent risk factors. Neither an increase nor a decrease in BMI or waist circumference during this time frame showed additional influence on eoCRC risk [22].

Only one study reported a lower prevalence of overweight and obesity in eoCRC compared to loCRC (27.8% vs. 37.9% and 30.6% vs. 36.4%), but these differences were not statistically significant [41].

A South Korean study found that higher waist circumference was associated with higher incidence of eoCRC (male < 100 cm/female < 95 cm vs. male < 80 cm/female < 75 cm MVHR 1.28 [1.20–1.37]) [26].

In summary, the literature highlights an association between obesity and eoCRC risk, including progressive associations with early adulthood obesity. Long-term obesity (> 10 years pre-diagnosis) is associated with eoCRC risk, as are substantial weight gain since adolescence, persistent overweight, and abdominal obesity.

To assess T2D as a risk factor for eoCRC, eighteen studies (ten cohort, seven case-control, one cross-sectional study) were analysed.

Most studies analysing T2D-related outcomes investigated the association between T2D at CRC diagnosis and eoCRC-incidence [2, 3, 20, 21, 23, 25, 26, 30,31,32, 35, 38, 39, 42, 43].

Effect of type 2 diabetes at diagnosis on eoCRC risk. Forest plot of a meta-analysis of studies analysing the association of Type 2 diabetes (T2D) at cancer diagnosis and eoCRC (full study names in references). Each study (stratified by age where provided) shows multivariate-adjusted risk ratios (MVRR), number of cases and individuals at risk. The horizontal lines indicate confidence intervals. Boxes indicate estimates of adjusted risk ratios. The size of the boxes inversely correlates with the width of the confidence intervals. The last row shows the pooled adjusted risk ratio. Only five out of nine outcome-related studies have been used for the forest plot and meta-analysis due to some either using a different measure of association without possibility for conversion or their definition of outcome not being suited for a direct comparison. eoCRC T2D: study population with early-onset colorectal cancer and Type 2 Diabetes at diagnosis. Total T2D: study population with Type 2 Diabetes at diagnosis. eoCRC non T2D: study population with early-onset colorectal cancer without Type 2 Diabetes. Total non T2D: study population without Type 2 Diabetes at diagnosis. *Pooled risk ratio estimate from subgroups

A meta-analysis showed a non-significant trend between T2D at CRC diagnosis and eoCRC risk with a pooled RR of 1.13 [0.89–1.44]) (Fig. 3) with relatively consistent findings across studies (I2 = 67.50%, t2 = 0.02). Stratification by age and sex, however, revealed T2D as a risk factor in males, doubling risk in 40–49-year-olds (MVRR 2.00 [1.75–2.28]) and showing higher risks in younger patients (20–39-year-olds: MVRR 3.42 [2.85–5.37]). A U.S. case-control study indicated that controlling hyperglycemia and diabetic complications was protective, while uncontrolled or complicated T2D were associated with a higher eoCRC risk (MVRR 1.32 [1.11–1.56] and 1.50 [1.07–2.07]) [4]. However, this finding did not validate in several separate studies [20, 39, 43,44,45].

A Swedish nationwide cohort study showed a forward shift of 4 to 5 years in the 10-year cumulative risk of CRC for patients with T2D compared to those without. When T2D was combined with a family history of CRC, the risk advancement increased markedly to 12–21 years relative to the general population [46].

In the same cohort, age at T2D diagnosis influenced the risk of developing eoCRC. Individuals diagnosed with T2D at ages 30–39 or 40–49 had significantly higher standardized incidence ratios (SIR) for eoCRC compared to non-diabetic controls (SIR 1.6 [1.1–2.2] and 3.6 [2.8–4.5] respectively). Additionally, T2D diagnosis at any age was associated with an increased risk of eoCRC (SIR 1.9 [1.6–2.3]), while the association for loCRC was less pronounced [47].

Another study comparing two prospective cohorts found that glucose levels above 7 mM were more strongly associated with eoCRC than loCRC for male individuals(defined here as eoCRC < 55 years). They also demonstrated linear dose-response relationship between glucose levels and overall CRC risk [48].

Collectively, current evidence suggests positive association between T2D and eoCRC in several large observational studies, particularly among males aged 20–49, although the overall pooled estimate was not statistically significant. Early-onset T2D elevates eoCRC incidence (SIR up to 3.6), and when combined with a family history of CRC, the cumulative risk is advanced by 12–21 years. Elevated blood glucose levels (> 7 mM) further reinforce this association, showing a strong, dose-dependent association with eoCRC risk in males.

Eight studies examined the influence of HLD on eoCRC outcome (three cohort studies, four case-control studies, one cross-sectional study), mostly by focusing on the prevalence of HLD at the time of eoCRC diagnosis.

Effect of hyperlipidemia at diagnosis on eoCRC risk. Forest plot of a meta-analysis of studies analysing the association of hyperlipidemia (HLD) at cancer diagnosis and eoCRC (full study names in references). Each study (stratified by age where provided) shows multivariate-adjusted risk ratios (MVRR), number of cases and individuals at risk. The horizontal lines indicate confidence intervals. Boxes indicate estimates of adjusted risk ratios. The size of the boxes inversely correlates with the width of the confidence intervals. The last row shows the pooled adjusted risk ratio. One study is not represented due to it only presenting measures of association for individuals with no hyperlipidemia and eoCRC-incidence [15]. Another is not represented due to it only providing one measure of association for the influence of BMI superior to 30 kg/m2 or Type 2 diabetes or hyperlipidemia [32]. eoCRC HLD: study population with early-onset colorectal cancer and hyperlipidemia at diagnosis. Total HLD: study population with hyperlipidemia. eoCRC non HLD: study population with early-onset colorectal cancer and without hyperlipidemia at diagnosis. Total non HLD: study population without hyperlipidemia. *Pooled risk ratio estimate from subgroups. **HDL-c < 50 mg/dL in women, < 40 mg/dL in men. ***serum triglyceride > 150 mg/dL

Multivariate analysis showed conflicting results among studies, with a pooled risk ratio of 1.28 [0.65–2.51], extreme heterogeneity and substantial between-study variance. Larger U.S. and South Korean cohorts have observed a positive independent positive correlation between HLD and risk of eoCRC, especially for 20–39 year olds [26, 30, 34], whereas smaller studies could not confirm these results [15, 31], or argue contrary to the observation [49] (Fig. 4).

Concerning HTN, only its influence on eoCRC incidence has been investigated so far. In total, we identified nine studies that addressed HTN as a potential risk factor for eoCRC [21, 23, 25, 26, 30, 31, 34, 35, 45].

Effect of Arterial hypertension at diagnosis on eoCRC risk. Forest plot of a meta-analysis of studies analysing the association of arterial hypertension (HTN) at cancer diagnosis and eoCRC (full study names in references). Each study (stratified by age where provided) shows multivariate-adjusted risk ratios (MVRR), number of cases and individuals at risk. The horizontal lines indicate confidence intervals. Boxes indicate estimates of adjusted risk ratios. The size of the boxes inversely correlates with the width of the confidence intervals. The last row shows the pooled adjusted risk ratio. One study is not represented due to it using a different measure of association [21]. HTN: arterial hypertension. Stage 1 HTN: systolic pressure 130–139 mmHg or diastolic pressure 80–89 mmHg. Stage 2 HTN: systolic pressure superior to 140 mmHg or diastolic pressure superior to 90 mmHg. eoCRC HTN: study population with early-onset colorectal cancer and arterial hypertension. Total HTN: study population with arterial hypertension, eoCRC non HTN: study population with early-onset colorectal cancer and without arterial hypertension, Total non HTN: study population without arterial hypertension. *Pooled risk ratio estimate from subgroups

The meta-analysis describing hypertension at onset was not statistically significant (summarized risk ratio of 1.46 [0.86–2.48]), had nearly complete heterogeneity and substantial variance., Subgroup analysis demonstrated a strong association between HTN at diagnosis and eoCRC, especially in male patients aged 20–39 and 40–49 years (MVRR 3.43 [2.77–4.22] and 2.00 [1.85–2.15] respectively) [23, 30] (Fig. 5).

A Chinese cohort study found a stronger association in women (MVHR 2.32 [1.01–5.34]) than in men (1.69 [0.60–4.67]), but the authors did not provide a clear definition of arterial hypertension. However, the authors did not provide a clear definition of arterial hypertension [21]. Two U.S. case-control studies failed to replicate this association; in one study, adjustment for family history did not alter the null result [31, 35]. Overall, evidence for an association between hypertension and eoCRC risk is inconsistent and limited.

Five studies were identified investigating the association between metabolic syndrome or metabolic comorbidities and eoCRC.

Two studies analysed the influence of individual or combined metabolic disorders on eoCRC incidence [24, 50]. Having at least one metabolic comorbidity (overweight/obesity, T2D, HLD and/or HTN) was an independent risk factor for eoCRC incidence (HR 1.82 [1.66–2.00.66.00]). Notably, this risk was most pronounced in individuals with particularly early onset disease: 30–39 year-olds (incidence rate ratio (IRR) 1.83) versus those aged 40–49 (IRR 1.26) and 50–59 (IRR 1.24) [24, 50].

A Chinese study identified metabolic syndrome as an independent risk factor for both eoCRC (MVOR 1.25 [1.098–1.43]) and loCRC (MVOR 1.15 [1.07–1.27]). The risk increased with the number of metabolic comorbidities for eoCRC: one disorder (MVOR 1.09 [1.00–1.17.00.17]), two disorders (MVOR 1.12 [1.01–1.24]), and three disorders (MVOR 1.31 [1.13–1.51]); for loCRC, only three disorders showed an independent association (MVOR 1.22 [1.15–1.29]) [13]. Another study reported progressively larger independent associations per additional metabolic comorbidity for eoCRC (one disorder (MVOR 1.07 [1.01–1.13]), two disorders (MVOR 1.13 [1.06–1.21]), and three disorders (MVOR 1.25 [1.16–1.35], four disorders (MVOR 1.27 [1.15–1.41]), and five disorders (MVOR 1.50 [1.26–1.79]) [26].

In summary therefore, the literature indicates that MetS and its comorbidities arerisk factors for eoCRC, with risk increasing with the number of metabolic disorders and being particularly pronounced in earlier-onset (30–39-year-old)patients o