Increased colorectal cancer risk in prediabetes: A meta-analysis

Abstract

BACKGROUND

Previous research yielded conflicting results regarding the association between prediabetes and colorectal cancer (CRC).

AIM

To systematically assess the incidence of CRC in individuals with prediabetes compared with individuals with normoglycemia via a meta-analysis.

METHODS

Relevant cohort studies were acquired by searching MEDLINE, Web of Science, and EMBASE. A random-effects model was applied to combine the findings after accounting for heterogeneity. Several subgroup analyses were conducted to assess the impact of study characteristics on the results.

RESULTS

Eleven cohort studies involving 4996352 participants, including 383917 (7.7%) with prediabetes at baseline, were analyzed in this meta-analysis. Over a mean follow-up period of 6.5 years, the combined findings revealed that individuals with prediabetes at baseline had a higher likelihood of developing CRC than those with normoglycemia [risk ratio (RR) = 1.18, 95% confidence interval = 1.11 to 1.25, P < 0.001] with low statistical heterogeneity (I² = 27%). Subgroup analyses indicated that the association between prediabetes and an increased risk of CRC was mainly observed in studies defining prediabetes using impaired fasting glucose (RR = 1.24) and slightly elevated hemoglobin A1c levels (RR = 1.18) but not in those that defined prediabetes using impaired glucose tolerance (RR = 1.06). Other study characteristics such as design, country, participant age and sex, the duration of follow-up, or adjustment for body mass index did not significantly impact the results (all P > 0.05).

CONCLUSION

People with prediabetes might have a higher likelihood of developing CRC than individuals with normoglycemia.

Key Words: Colorectal cancer; Prediabetes; Incidence; Risk factor; Meta-analysis

Core Tip: Previous research yielded conflicting results regarding the association between prediabetes and colorectal cancer (CRC). We conducted a meta-analysis to systematically assess the incidence of CRC in individuals with prediabetes compared with those with normoglycemia. A significant 18% increase in the occurrence of CRC among individuals with prediabetes compared with those with normoglycemia was observed. These results suggested that people with prediabetes have a higher likelihood of developing CRC than those with normoglycemia.

INTRODUCTION

Colorectal cancer (CRC) represents a formidable challenge to global public health, and its global incidence is rising[1,2]. The current treatment of CRC involves a comprehensive strategy of surgical resection, chemoradiotherapy, targeted therapy, and immunotherapy[3,4]. However, the prognosis of CRC remains unsatisfactory despite these treatments, particularly for patients with advanced malignancy[5]. These facts highlight the importance of the primary prevention of CRC. Although established risk factors such as age, family history, and lifestyle habits contribute significantly to CRC development[6], emerging research has underscored the potential role of hyperglycemia in its pathogenesis[7]. Continuous hyperglycemia leads to chronic inflammation, insulin resistance, and oxidative stress, which might play key roles in carcinogenesis[8,9]. Consistently, diabetes has been related to an increased incidence and mortality of CRC[10,11].

Recent advancements in the field of diabetology underlined the importance of prediabetes, a transitional phase between normal and diabetic states characterized by elevated blood glucose levels[12,13]. Prediabetes signifies impaired glucose regulation before an actual diabetes diagnosis, including conditions such as impaired fasting glucose (IFG), impaired glucose tolerance (IGT), and slightly elevated hemoglobin A1c (HbA1c) levels (5.7%-6.5%)[14]. Similarly as people with diabetes, individuals with prediabetes also face an increased risk of cardiovascular events[15]. Moreover, there is a suggested association between prediabetes and cancer risk, although this link might vary depending on the specific type of cancer[16]. Prediabetes has attracted attention as a metabolic state that could predispose individuals to CRC because of its characteristic high blood glucose levels[17]. Although there is growing recognition of the mechanisms connecting hyperglycemia with CRC, epidemiological evidence concerning their relationship remains inconclusive[18-28]. Specifically, several previous studies did not suggest a significant association between prediabetes and CRC[20,25,28], although most of the previous epidemiological studies suggested a significant association[18,19,21-24,26,27]. In view of the conflicting findings of the previous studies, the present study conducted a meta-analysis to compare the incidence of CRC between individuals diagnosed with prediabetes and those with normoglycemia by systematically integrating data from diverse cohort studies to clarify the extent of the association between hyperglycemia and CRC risk and investigate potential sources contributing to variation across previous studies.

MATERIALS AND METHODS

This study adhered to the guidelines outlined in PRISMA 2020[29,30] and the Cochrane Handbook for Systematic Reviews and Meta-analyses[31] encompassing study design, data collection, statistical analysis, and data interpretation. The protocol of the meta-analysis has been registered at POSRERO (No. CRD42024542485).

Literature search

To identify studies relevant to the aim of the meta-analysis, we searched MEDLINE, Web of Science, and EMBASE utilizing comprehensive search terms involving: (1) “prediabetes” OR “prediabetic” OR “pre-diabetes” OR “pre-diabetic” OR “borderline diabetes” OR “prediabetic state” OR “impaired glucose tolerance” OR “impaired fasting glucose” OR “IGT” OR “IFG” OR “HbA1c” OR “fasting glucose”; and (2) “colorectal” OR “colon” OR “rectal” OR “colorectum” OR “rectum” AND “cancer” OR “tumor” OR “neoplasms” OR “malignancy” OR “carcinoma”. The search was restricted to studies involving human subjects. We only included studies that had been published as complete articles in English in peer-reviewed journals. Additionally, we manually examined the references of relevant original and review articles for potentially pertinent studies. The literature published from the establishment of the databases to March 12, 2024 was reviewed.

Inclusion and exclusion criteria

The inclusion criteria for the potential studies were as follows: (1) Cohort studies published as full-length articles, including prospective and retrospective cohorts; (2) The inclusion of a general population without CRC at baseline; (3) Prediabetes was evaluated at baseline as exposure and was diagnosed according to the criteria in the original studies; and (4) The incidence of CRC was compared between participants with prediabetes and those with normoglycemia during follow-up.

The exclusion criteria were as follows: (1) Cross-sectional studies or case-control studies; (2) The inclusion of patients diagnosed with specific diseases rather than a general population; (3) No evaluation of prediabetes at baseline; (4) Failure to report the incidence of CRC during follow-up; and (5) Preclinical studies, reviews, and editorials. If studies with overlapping populations were retrieved, the study with the largest sample size was included for the meta-analysis.

Study quality evaluation and data extraction

The literature search, study identification, study quality assessment, and data collection were performed independently by two authors. In cases of disagreement, the corresponding author was consulted to reach a resolution. To evaluate the quality of the included studies, we utilized the Newcastle-Ottawa scale (NOS)[32], which assesses three aspects: Selection of the population, control of confounders, and outcome measurement and analysis. The NOS score ranges from 1 to 9, with 9 indicating superior quality. We extracted various data from each study for subsequent analyses including study information (author, year, country, and design), participant characteristics (sample size, age, and sex), baseline prediabetes diagnosis (along with the definition and the number of participants diagnosed), mean follow-up duration, and CRC diagnosis (type reported and validation methods). Furthermore, variables adjusted when reporting the association between prediabetes and CRC were also included in the extraction process.

Statistical analysis

The relationship between prediabetes and the CRC incidence was assessed using risk ratios (RRs) and the corresponding 95% confidence intervals (CIs) between individuals with prediabetes and those with normoglycemia. RRs and standard errors were computed according to 95%CIs or P values, followed by logarithmic transformation for variance stabilization. Heterogeneity among studies was evaluated using the Cochrane Q test and I² statistic[33], where I² > 50% indicated significant statistical heterogeneity. The findings were combined utilizing a random-effects model that accounted for the influence of heterogeneity[31]. Sensitivity analyses involving the exclusion of one study at a time were conducted to assess the robustness of the results. Predefined subgroup analyses were also conducted to examine the influence of study characteristics on the outcome, with median values of continuous variables used as cutoffs for defining subgroups. Publication bias in the meta-analysis was assessed through the construction of funnel plots along with visual inspection for plot symmetry[34]. Additionally, Egger’s regression test was performed[34]. Statistical analysis utilized RevMan (Version 5.1, Cochrane Collaboration, Oxford, United Kingdom) and Stata software (version 12.0, Stata Corporation, College Station, TX, United States).

RESULTS

Study inclusion

The process of study inclusion is presented in Figure 1. In brief, 977 potentially relevant records were obtained after comprehensive searches of the three databases, and 208 duplicate records were excluded. Subsequently, screening of the titles and abstracts of the remaining records led to the exclusion of 739 studies, mostly because they were not related to the aim of the meta-analysis. Accordingly, the full texts of the 30 remaining records were independently read by two authors, and 19 studies were further removed for the reasons listed in Figure 1. Finally, 11 cohort studies were considered suitable for the subsequent quantitative analyses[18-28].

Figure 1 The flowchart depicts the process of database search and study inclusion. CRC: Colorectal cancer.

Overview of study characteristics

Table 1 presents the summarized characteristics of the included studies. Overall, nine prospective cohorts[18-23,25,26,28] and two retrospective cohorts[24,27] were included in the meta-analysis. These studies were reported from 2005 to 2024, and they were performed in Korea, Austria, the United States, the United Kingdom, China, and Japan. Overall, 4996352 participants from community-derived general populations were included. The mean ages of the participants ranged from 37.5-59.3 years. The diagnosis of prediabetes was based on IFG in six studies[18,19,21,22,24,27], IGT in one study[25], and mildly elevated HbA1c levels (5.7%-6.5%) in four studies[20,23,26,28]. Accordingly, 383917 (7.7%) of the included subjects had prediabetes at baseline. The mean follow-up durations in the studies ranged 3.1-35.0 years. The incidence of CRC during follow-up was reported in all but one of the included studies, which reported the incidence of colon cancer[21]. The CRC incidence was mostly validated using local or national cancer registries or the International Classification of Diseases codes of the healthcare databases, whereas it was validated by pathological reports in one study[21]. Multivariate analysis was performed in all of the included studies to evaluate the association between prediabetes and the risk of CRC, and factors such as age, sex, smoking, alcohol drinking, and body mass index (BMI) or waist circumference (WC) were adjusted to varying extents. The NOS score of the included studies ranged from 6-9, suggesting overall moderate-to-good study quality (Table 2).

Table 1 Characteristics of the included cohort studies.

Ref.	Country	Design	Population	Sample size	Mean age (years)	Male sex (%)	Definition of PreD	No. of participants with PreD	Mean follow-up duration (years)	Type of cancer	Methods for validation of patients with CRC	Variables adjusted
Jee et al[18], 2005	Korea	PC	General population	1298385	46.9	63.9	IFG	80598	10	Overall CRC	National cancer registry	Age and sex
Rapp et al[19], 2006	Austria	PC	General population	140813	43	45.2	IFG	6828	8.4	Overall CRC	Local cancer registry	Age, sex, smoking, occupation, and BMI
Joshu et al[20], 2012	United States	PC	General population	12792	56.9	45.3	HbA1c (5.7%-6.5%)	2861	13.2	Overall CRC	State cancer registry	Age, sex, ethnicity, study site, education, smoking, BMI, and WC
Parekh et al[21], 2013	United States	PC	General population	4615	37.5	50	IFG	700	35	Colon cancer	Pathological reports	Age, sex, alcohol, smoking, and BMI
Shin et al[22], 2014	Korea	PC	General population	175677	41.8	58.9	IFG	29461	4.7	Overall CRC	National cancer registry	Age, sex, BMI, smoking, alcohol consumption, and regular exercise
Peila et al[23], 2020	United Kingdom	PC	General population	476517	57.8	46	HbA1c (5.7%-6.5%)	64167	7.1	Colon and rectal cancer	ICD codes	Age, sex, ethnicity, education, smoking, physical activity, and HRT
Ke et al[25], 2021	China	PC	General population	9224	52.3	34.3	IGT	1454	7.5	Overall CRC	Local cancer registry	Age and sex
Itoh et al[24], 2021	Japan	RC	General population	1441311	46	56.6	IFG	74836	3.1	Overall CRC	ICD codes	Age, sex, obesity, high WC, hypertension, dyslipidemia, cigarette smoking, alcohol consumption, and physical inactivity
Campbell et al[26], 2022	United States	PC	General population	5049	NR	51.5	HbA1c (5.7%-6.5%)	916	9.5	Overall CRC	Local cancer registry	Age, sex, BMI, smoking, physical activity, alcohol, and HRT
Ahn et al[27], 2024	Korea	RC	General population	1430054	49.8	71.4	IFG	120906	6.4	Overall CRC	National cancer registry	Age, sex, BMI, smoking status, alcohol consumption, physical activity, hypertension, and dyslipidemia
Wang et al[28], 2024	China	PC	General population	1915	59.3	49.9	HbA1c (5.7%-6.5%)	1190	12.9	Overall CRC	Local cancer registry	Age, sex, WC, smoking, alcohol drinking, household annual income, education, physical activity, and intake of vegetables and red meat

PreD: Prediabetes; CRC: Colorectal cancer; PC: Prospective cohort; RC: Retrospective cohort; IFG: Impaired fasting glucose; IGT: Impaired glucose tolerance; HbA1c: Hemoglobin A1c; BMI: Body mass index; WC: Waist circumference; HRT: Hormone replacement therapy; ICD: International Classification of Diseases.

Table 2 Study quality evaluation via the Newcastle-Ottawa scale.

Ref.	Representativeness of the exposed cohort	Selection of the non-exposed cohort	Ascertainment of exposure	Outcome not present at baseline	Control for age and sex	Control for other confounding factors	Assessment of outcome	Sufficient long follow-up duration	Adequacy of the follow-up of cohorts	Total
Jee et al[18], 2005	1	1	1	1	1	0	0	1	1	7
Rapp et al[19], 2006	1	1	1	1	1	1	0	1	1	8
Joshu et al[20], 2012	1	1	1	1	1	1	0	1	1	8
Parekh et al[21], 2013	1	1	1	1	1	1	1	1	1	9
Shin et al[22], 2014	1	1	1	1	1	1	0	0	1	7
Peila et al[23], 2020	1	1	1	1	1	1	0	1	1	8
Ke et al[25], 2021	1	1	1	1	1	0	0	1	1	7
Itoh et al[24], 2021	0	1	1	1	1	1	0	0	1	6
Campbell et al[26], 2022	1	1	1	1	1	1	0	1	1	8
Ahn et al[27], 2024	0	1	1	1	1	1	0	1	1	7
Wang et al[28], 2024	1	1	1	1	1	1	0	1	1	8

Results of the meta-analysis

Because six studies reported the outcomes separately in men and women[18-20,22,24,28] and one study reported the outcomes of colon cancer and rectum cancer separately[23], these datasets were independently included, resulting in 18 datasets from 11 studies. The pooled results revealed that compared to people with normoglycemia, those with prediabetes at baseline had a higher incidence of CRC (RR = 1.18, 95%CI = 1.11-1.25, P < 0.001, Figure 2A) with mild statistical heterogeneity (I² = 27%).

Figure 2 Forest plots of the subgroup analyses of the association between prediabetes and the risk of colorectal cancer. A: Overall subgroup analysis; B: Subgroup analysis according to study design; C: Subgroup analysis according to the study country; D: Subgroup analysis according to mean age of the participants; E: Subgroup analysis according to the sex of the participants; F: Subgroup analysis according to the definition of prediabetes; G: Subgroup analysis according to the mean follow-up duration; H: Subgroup analysis according to adjustment for body mass index or waist circumference. CI: Confidence interval.

Results of sensitivity analysis

Further analysis assessing the impact of excluding individual datasets consistently demonstrated similar results (RR = 1.16-1.19, all P < 0.05). In particular, when the single study reporting colon cancer opposed to CRC was excluded, comparable findings were observed (RR = 1.16, 95%CI = 1.11-1.21, P < 0.001), and a significant decrease in statistical heterogeneity was noted (I² = 0%).

Results of subgroup analyses

Further subgroup analyses revealed similar results in prospective (RR = 1.19) and retrospective cohorts (RR = 1.18, P for subgroup difference = 0.93, Figure 2B), in studies from Asian (RR = 1.17) and Western countries (RR = 1.21, P for subgroup difference = 0.62, Figure 2C), in participants aged < 50 years (RR = 1.24) and ≥ 50 years (RR = 1.11, P for subgroup difference = 0.09, Figure 2D), and in men (RR = 1.23) and women (RR = 1.15, P for subgroup difference = 0.37, Figure 2E). Subgroup analysis also suggested that the association between prediabetes and an increased risk of CRC was mainly influenced by studies of prediabetes defined by IFG (RR = 1.24) and mildly elevated HbA1c levels (RR = 1.18), but not by studies that defined prediabetes by IGT (RR = 1.06), although the between-subgroup difference was not significant (P for subgroup difference = 0.18, Figure 2F). Finally, similar results were obtained in studies with follow-up durations of < 8 years (RR = 1.17) and ≥ 8 years (RR = 1.23, P for subgroup difference = 0.49, Figure 2G) and in studies with (RR = 1.24) and without (RR = 1.12) adjustment for BMI or WC (P for subgroup difference = 0.09, Figure 2H).

Publication bias evaluation

Upon visual inspection, the funnel plots for the meta-analysis of the association between prediabetes and the risk of CRC appeared symmetrical, indicating a low likelihood of publication bias (Figure 3). Additionally, Egger’s regression test results (P = 0.42) also support this conclusion by suggesting a low risk of publication bias.

Figure 3 Funnel plots of the association between prediabetes and the risk of colorectal cancer. RR: Risk ratio.

DISCUSSION

This meta-analysis presented current evidence regarding the connection between prediabetes and the risk of CRC. Through the consolidation of data from 11 cohort studies involving an extensive cohort of more than 4.9 million individuals, we noted a significant 18% increase in the occurrence of CRC among individuals with prediabetes compared to those with normoglycemia. These results further suggested a possible influence of high blood glucose on the development of CRC, even prior to a diabetes diagnosis. To the best of our knowledge, two previous meta-analyses assessed the connection between prediabetes and CRC[16,35]. The first analysis from 2014 indicated that prediabetes increases the overall risk of cancer across various sites[16]. Furthermore, a subgroup analysis of three studies focusing on stomach cancer and CRC outcomes also found a similar association. However, it should be noted that both cancer incidence and mortality studies were included in this earlier meta-analysis, which could have influenced the results[16]. Additionally, the limited datasets for CRC outcomes might have impacted the reliability of the findings[16]. Although the other meta-analysis also suggested a possible relationship between prediabetes and CRC, only studies published before 2022 were included[35]. In addition, their meta-analysis included studies with non-standard definitions of prediabetes (e.g., IGT: 8.9-12.1 mmol/L), which might have confounded the results[35]. Our meta-analysis offers several methodological strengths compared with the previous studies. First, we conducted an extensive literature search across three commonly used databases to gather current literature relevant to our aim. Second, all included studies were cohort studies investigating the longitudinal relationship between prediabetes and CRC incidence. Thirdly, we conducted several sensitivity and subgroup analyses, thereby supporting the robustness of our findings. Finally, statistical heterogeneity was eliminated by removing the only study that reported colon cancer rather than CRC, which explained the source of heterogeneity.

The molecular mechanisms linking prediabetes to CRC involve chronic inflammation, insulin resistance, and oxidative stress. Hyperglycemia and hyperinsulinemia in prediabetic states activate the insulin-like growth factor/insulin axis, promoting cell proliferation and inhibiting apoptosis, thereby fostering carcinogenesis[36]. Chronic low-grade inflammation, mediated by cytokines such as interleukin-6 (IL-6) and tumor necrosis factor α, disrupts immune surveillance and supports tumor progression, whereas oxidative stress induces DNA damage and mutations[37,38]. Additionally, hyperglycemia can trigger epigenetic changes, such as methylation of tumor suppressor genes (e.g., adenomatous polyposis coli) and mismatch repair genes (e.g., mutL homolog 1)[39]. Alterations in gut microbiota further exacerbate CRC risk by increasing intestinal permeability and activating inflammatory pathways via toll-like receptor 4 signaling[40]. These overlapping mechanisms underscore the shared biological pathways between prediabetes and CRC.

In addition, the potential association between prediabetes and the risk of CRC might also be partially explained by immune dysregulation involving the IL-17 pathway[41]. In prediabetic states, chronic low-grade inflammation, driven by elevated IL-17 Levels, can disrupt immune homeostasis and promote a pro-tumorigenic environment in the colon[42]. This inflammatory response, exacerbated by hyperglycemia-induced oxidative stress, creates conditions conducive to colorectal carcinogenesis[43]. Targeting the IL-17 pathway, as seen with IL-17 inhibitors used in inflammatory diseases, could offer a preventive approach by reducing CRC risk in prediabetic individuals[44]. Future research on IL-17 inhibitors in this population could provide valuable insights into their potential role in CRC prevention.

Interestingly, our subgroup analyses revealed differential effects of prediabetes subtypes on CRC risk, with elevated fasting glucose and HbA1c levels displaying a stronger association than IGT. This highlights the heterogeneity within the prediabetic population and suggests that certain metabolic abnormalities might confer a higher risk of CRC. There is evidence suggesting the potential heterogeneity of prediabetes according to different diagnostic criteria. For instance, it has been proposed that the use of HbA1c for defining prediabetes offers greater precision and provides moderate enhancements in risk assessment for clinical complications such as cardiovascular disease, peripheral arterial disease, chronic kidney disease, and overall mortality[45]. Furthermore, a recent study revealed that the likelihood of progressing from prediabetes to diabetes significantly varies based on the cohort and definitions used for prediabetes, with the risk being highest when prediabetes is defined by IFG[46]. However, only one of the studies included in our meta-analysis assessed prediabetes using IGT. Caution should be exercised when interpreting the results of this subgroup analysis. Additionally, subgroup analysis indicated a similar connection between prediabetes and CRC risk in studies adjusting for BMI or WC. As obesity is a known risk factor for CRC[47], these findings suggest that any link between prediabetes and an increased CRC risk is unlikely to be attributable to obesity-mediated pathways.

The identification of prediabetes as a potential risk factor for CRC also raises important considerations regarding screening and surveillance protocols for CRC in clinical practice. Current guidelines recommend CRC screening primarily based on age and family history, with limited emphasis given to metabolic factors such as prediabetes. Our findings suggest that individuals with prediabetes could benefit from earlier or more frequent CRC screening to detect CRC and intervene in its development at an earlier stage. Integrating metabolic parameters such as fasting glucose and HbA1c levels into existing risk stratification algorithms might enhance the precision and efficacy of CRC screening strategies. Furthermore, our research emphasized the importance of additional investigations to clarify the fundamental connections between hyperglycemia and the development of CRC. Uncovering the molecular pathways associated with the carcinogenic impact of prediabetes may reveal new targets for preventing CRC. It is also important to conduct future studies to confirm our results and evaluate the effects of interventions addressing prediabetes on long-term incidences for CRC.

Our meta-analysis contributes to the growing body of evidence identifying prediabetes as a significant risk factor for CRC. However, it is essential to interpret these findings in the context of several limitations inherent to our study design and the complexities underlying the association between prediabetes and CRC risk. First, our meta-analysis relied on data extracted from observational cohort studies, which are susceptible to inherent biases and confounding factors. Despite efforts to adjust for potential confounders in the included studies, residual confounding remains a concern. Factors such as dietary habits, physical activity levels, and other lifestyle factors, which are closely linked to both prediabetes and CRC risk, might not have been adequately controlled in the original studies. Additionally, variations in definitions of prediabetes, outcome assessment across included studies, and variables included in the regression models might have introduced heterogeneity and limited the generalizability of our findings. Furthermore, the observational nature of the included studies precludes establishing a causal relationship between prediabetes and CRC risk. Finally, we were unable to determine the association between prediabetes with different pathological types of CRC because this data was merely reported among the included studies. Studies are needed for further investigations in the future.

CONCLUSION

In conclusion, our comprehensive analysis supports the potential association between prediabetes and the risk of CRC. These results highlight the significance of identifying and addressing hyperglycemia as early as possible to possibly prevent CRC and stress the necessity for customized prevention and screening approaches in individuals with prediabetes, although further interventional studies are needed for validation. Additional investigation is necessary to understand the underlying mechanisms of the association between prediabetes and the risk of CRC.