The results of the literature search are shown in Fig. S1. In total, 15,556 records were identified in MEDLINE and in EMBASE. After the exclusion of duplicates, the titles and abstracts from a total of 15163 publications were screened, and out of these, 393 full-text articles were retrieved and assessed for inclusion. Of these, 362 records were excluded due to study design or publication type (abstract, letter, note, commentary, review, or meta-analysis), lack of risk estimates, or irrelevant outcome, exposure, and data. In total, we included 31 publications from 15 unique cohort studies on the associations between metabolic obesity phenotypes and cancer risk that met all the inclusion criteria for the meta-analysis [15,16,17,18,19,20,21,22,23,24, 28,29,30,31,32,33,34, 40, 41, 45,46,47,48,49,50,51,52,53,54,55,56]. Of the included articles, 15 publications were from Asia, 9 publications from North America, and 7 publications from Europe. All studies were conducted in adult subjects, the age at baseline ranged from 20 to 79 years for the publications that provided an age range. Sample sizes varied widely among publications. The characteristics of the eligible studies are presented in Tables S1–S16.

Out of the 31 publications assessed using the ROBINS-I tool, 22 were evaluated as being at low risk of bias and 9 were considered at moderate risk of bias (Table 1). In this study, risk of bias due to confounding and outcome assessment was low since almost all studies included in this review provided risk estimates that adjusted for potential confounders, and cancer cases were clinically confirmed. However, selection bias was identified at a moderate level in some studies, particularly those conducted among older populations or postmenopausal women. Bias due to exposure assessment was considered moderate in certain studies that used measurement of a single biomarker with a study-specific cut-point to define metabolic health categories, which may not reflect a complete and objective definition of metabolic healthy status. Nevertheless, none of the studies were classified as being at serious risk of bias or at critical risk of bias.

The definition of metabolic health status varied across the studies (Table 2). In most studies, the MU phenotype was defined as having either 1, 2, or 3 metabolic syndrome components based on the criteria of the Adult Treatment Panel III, which includes (1) elevated blood pressure or the use of antihypertensive medication at baseline, (2) hypertriglyceridemia or current use of lipid-lowering medication at baseline, (3) low HDL cholesterol, (4) hyperglycaemia or use of medications for diabetes at baseline, and (5) abdominal obesity. In other studies, elevated levels of glucose or insulin levels, HOMA-IR, C-peptide levels, or C-reactive protein alone was used to define MU phenotype.

A total of 27 publications were included in the analysis of MUOW/OB versus MHNW phenotype and risk of overall and site-specific cancers [15,16,17,18,19,20,21,22,23,24, 28, 30,31,32,33,34, 40, 41, 45, 46, 48,49,50, 53,54,55,56]. There was high certainty of evidence that individuals classified as MUOW/OB were at a higher risk of overall cancer compared to MHNW individuals (SRR = 1.21, 95% CI = 1.02–1.44, n = 3) and there was little evidence of heterogeneity (I2 = 19%) and no evidence of publication bias (Egger’s test: P = 0.91) (Fig. 1). Compared with MHNW individuals, obesity-related cancer risk was higher among those classified as MUOW/OB (SRR = 1.42, 95% CI = 1.15–1.74, I2 = 96%, n = 3, very low certainty of evidence). Specifically, MUOW/OB individuals were at higher risk of cancers of the postmenopausal breast (SRR = 1.32, 95% CI = 1.17–1.48, I2 = 91%, n = 7, low certainty), colorectum (SRR = 1.24, 95% CI = 1.16–1.31, I2 = 53%, n = 6, moderate certainty), colon (SRR = 1.35, 95% CI = 1.27–1.43, I2 = 0%, n = 3, high certainty), rectum (SRR = 1.18, 95% CI = 1.10–1.27, I2 = 0%, n = 2, high certainty), pancreas (SRR = 1.35, 95% CI = 1.24–1.47, I2 = 0%, n = 3, high certainty), gallbladder (SRR = 1.42, 95% CI = 1.17–1.73, I2 = 0%, n = 2, high certainty), bladder (SRR = 1.36, 95% CI = 1.19–1.56, I2 = 48%, n = 2, high certainty), and stomach (SRR = 1.50, 95% CI = 1.12–2.01, I2 = 32%, n = 2, high certainty). In addition, we found that MUOW/OB phenotype was strongly and positively associated with risks of endometrial (SRR = 2.31, 95% CI = 2.08–2.57, I2 = 75%, n = 4, high certainty), thyroid (SRR = 1.42, 95% CI = 1.29–1.57, I2 = 75%, n = 4, moderate certainty), kidney (SRR = 1.71, 95% CI = 1.40–2.10, I2 = 80%, n = 3, low certainty), and liver (SRR = 1.81, 95% CI = 1.36–2.42, I2 = 67%, n = 2, moderate certainty) cancers. Conversely, individuals with the MUOW/OB were not at statistically significant increased risk of ovarian (SRR = 1.08, 95% CI = 0.97–1.20, I2 = 0%, n = 2, moderate certainty) or prostate (SRR = 1.05, 95% CI = 0.74–1.48, I2 = 99%, n = 2, very low certainty) cancers, or myeloma (SRR = 1.06, 95% CI = 0.85–1.31, I2 = 61%, n = 2, low certainty) compared to MHNW individuals, whereas MUOW/OB individuals had a lower risk of premenopausal breast cancer (RR = 0.71, 95% CI = 0.52–0.97, n = 1).

BC breast cancer, CI confidence interval, MHNW metabolically healthy normal weight, MHOW/OB metabolically healthy overweight or obese, MUNW metabolically unhealthy normal weight, MUOW/OB metabolically unhealthy overweight or obese, SRR summary relative risks. I2 is a measure of the proportion of the heterogeneity attributed to between-study variation rather than due to chance.

When performing separate analyses for overweight and obesity, we found that the MUOW phenotype was associated with increased risks of seven cancers including postmenopausal breast (SRR = 1.12, 95% CI = 1.00–1.26, I2 = 60%, n = 3), colorectal (SRR = 1.20, 95% CI = 1.12–1.27, I2 = 0%, n = 3), endometrial (SRR = 1.58, 95% CI = 1.29–1.95, I2 = 54%, n = 3), pancreas (SRR = 1.27, 95% CI = 1.12–1.44, I2 = 0%, n = 2), kidney (SRR = 1.57, 95% CI = 1.39–1.78, I2 = 0%, n = 2), gallbladder (SRR = 1.33, 95% CI = 1.03–1.72, I2 = 0%, n = 2), and liver (1.50, 95% CI = 1.20–1.86, I2 = 9%, n = 2) cancers (Fig. 2). The MUOB phenotype was positively associated with the risk of eight cancers: postmenopausal breast (SRR = 1.29, 95% CI = 1.05–1.60, I2 = 88%, n = 3), colorectal (SRR = 1.29, 95% CI = 1.12–1.48, I2 = 73%, n = 3), endometrial (SRR = 3.42, 95% CI = 2.57–4.56, I2 = 80%, n = 3), pancreas (SRR = 1.51, 95% CI = 1.28–1.80, I2 = 16%, n = 2), kidney (SRR = 2.43, 95% CI = 2.10–2.82, I2 = 12%, n = 2), gallbladder (SRR = 1.57, 95% CI = 1.16–2.13, I2 = 0%, n = 2), liver (SRR = 2.35, 95% CI = 1.63–3.39, I2 = 58%, n = 2) and ovarian (SRR = 1.19, 95% CI = 1.02–1.40, I2 = 0%, n = 2) cancers (Fig. 2).

BC breast cancer, CI confidence interval, MHNW metabolically healthy normal weight, MHOB metabolically healthy obese, MHOW metabolically healthy overweight, MUNW metabolically unhealthy normal weight, MUOB metabolically unhealthy obese, MUOW metabolically unhealthy overweight, SRR summary relative risks. I2 is a measure of the proportion of the heterogeneity attributed to between-study variation rather than due to chance.

When comparing MUNW individuals to MHNW individuals, 30 publications were identified [15,16,17,18,19,20,21,22,23,24, 28,29,30,31,32,33,34, 40, 41, 45,46,47,48,49,50, 52,53,54,55,56]. We observed that MUNW individuals had a modestly higher risk of obesity-related cancers (SRR = 1.11, 95% CI = 1.00–1.24, I2 = 70%, n = 3, moderate certainty) compared with MHNW (Fig. 1). More specifically, individuals classified as MUNW had increased risks of eight cancers: postmenopausal breast (SRR = 1.08, 95% CI = 1.03–1.14, I2 = 30%, n = 8 studies, high certainty), colorectal (SRR = 1.14, 95% CI = 1.06–1.23, I2 = 45%, n = 6, high certainty), colon (SRR = 1.12, 95% CI = 1.06–1.17, I2 = 0%, n = 5, high certainty), rectal (SRR = 1.13, 95% CI = 1.00–1.29, I2 = 58%, n = 3, moderate certainty), endometrial (SRR = 1.12, 95% CI = 1.02–1.24, I2 = 0%, n = 4, high certainty), thyroid (SRR = 1.06, 95% CI = 1.03–1.10, I2 = 0%, n = 4, moderate certainty), pancreatic (SRR = 1.35, 95% CI = 1.17–1.56, I2 = 30%, n = 3, high certainty), kidney (SRR = 1.29, 95% CI = 1.13–1.47, I2 = 15%, n = 3, high certainty), liver (SRR = 1.33, 95% CI = 1.05–1.69, I2 = 0%, n = 2, high certainty) and bladder (SRR = 1.18, 95% CI = 1.14–1.23, I2 = 0%, n = 2, high certainty) cancers. However, associations between MUNW and overall (SRR = 1.09, 95% CI = 0.90–1.33, I2 = 0%, n = 3, moderate certainty), gallbladder (SRR = 1.23, 95% CI = 0.92–1.64, I2 = 0%, n = 2, moderate certainty), stomach (SRR = 1.11, 95% CI = 0.75–1.62, I2 = 0%, n = 2, moderate certainty), ovarian (SRR = 1.05, 95% CI = 0.91–1.21, I2 = 0%, n = 2, moderate certainty), prostate (SRR = 1.04, 95% CI = 0.85–1.27, I2 = 92%, n = 2, very low certainty), myeloma (SRR = 0.79, 95% CI = 0.45–1.41, I2 = 81%, n = 2, very low certainty), and premenopausal breast (SRR = 0.89, 95% CI = 0.48–1.65, moderate certainty) cancers were not statistically significant when compared to MHNW individuals.

In total, 28 publications were included in the analysis of MHOW/OB versus MHNW phenotype and cancer risk [15,16,17,18,19,20,21,22,23,24, 28, 30,31,32,33,34, 40, 41, 45, 46, 48,49,50,51, 53,54,55,56]. Compared to MHNW individuals, those with MHOW/OB were at elevated risk of obesity-related cancer (SRR = 1.17, 95% CI = 1.09–1.26, I2 = 64%, n = 3, low certainty) (Fig. 1). The MHOW/OB phenotype was associated with increased risks of nine cancers: postmenopausal breast (SRR = 1.19, 95% CI = 1.12–1.26, I2 = 56%, n = 7, moderate certainty), colorectal (SRR = 1.08, 95% CI = 1.00–1.17, I2 = 58%, n = 6, moderate certainty), endometrial (SRR = 1.71, 95% CI = 1.58–1.85, I2 = 31%, n = 4, high certainty), thyroid (SRR = 1.27, 95% CI = 1.18–1.37, I2 = 16%, n = 4, high certainty), pancreatic (SRR = 1.17, 95% CI = 1.10–1.37, I2 = 53%, n = 3, moderate certainty), kidney (SRR = 1.35, 95% CI = 1.23–1.49, I2 = 0%, n = 3, high certainty), liver (SRR = 1.25, 95% CI 1.04–1.51, I2 = 0%, n = 2, high certainty), gallbladder (SRR = 1.48, 95% CI = 1.19–1.85, I2 = 0%, n = 2, high certainty) and bladder cancers (SRR = 1.07, 95% CI = 1.02–1.12, I2 = 0%, n = 2, high certainty). However, no statistically sign