Over the past decades, the prevalence of inflammatory bowel disease (IBD) has surged globally. It is commonly believed that IBD occurs mainly in Western countries, with more than 2 million people in North America and more than 3.2 million people in Europe.¹ However, since the beginning of the 21st century, the incidence of IBD has increased substantially in Asia. From 1990 to 2019, the age-standardized prevalence rate of IBD has risen dramatically, from 29.81 to 39.37 (per 100,000 people).² A study in the United States showed that the annual cost of IBD nursing exceeds 10 billion dollars.³ IBD not only reduces the quality of life of patients but also predisposes them to a variety of malignancies, especially colorectal cancer. Several studies have shown a potential correlation between IBD and hematologic malignancy.^4,5

Hematologic malignancies are a group of diseases of the hematopoietic hierarchy, including leukemia, lymphoma, and multiple myeloma. They have the highest morbidity and mortality rates. Repeated virus infections and chronic exposure to organic solvents or heavy metals are all risk factors for hematologic malignancies. Several studies have reported that the use of immunosuppressive therapy may increase the risk of malignancies in patients with IBD.⁶

Previous meta-analysis suggested that hematologic malignancies were more likely to occur in patients with IBD, but the result showed borderline significance.⁷ For specific types of hematologic malignancies, the results of research vary and deserve further analysis and exploration.⁸ In addition, with the development of research, the therapeutic agents of IBD have changed, and more and more biologics have been adopted. At the same time, cohort studies on the correlation between IBD and hematologic malignancies are constantly being updated. Especially in Asia, cohort studies are increasing. Due to genetic and environmental factors, the specific profile of malignancies in Asian patients with IBD may vary from that of Western countries. Based on the above, it is necessary to update the meta-analysis and make a more detailed subgroup analysis in terms of IBD type, hematologic malignancy type, pharmacotherapy, age, and geographical factors. In this study, a meta-analysis was conducted to investigate the correlation between IBD and hematologic malignancies in studies conducted after 2000.

1. Literature search strategy

This study was conducted in line with the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) and MOOSE (Meta-analysis of Observational Studies in Epidemiology) reporting guidelines. Only cohort studies on the relationship between patients with IBD and hematologic malignancies were selected for this study and were based on publicly procurable data, all of which were extracted from previously ethically approved studies. The protocol for this study had been registered in the International Prospective Register of Systematic Reviews, with the registration number CRD42024462331.

Two investigators independently conducted a systematic and comprehensive literature retrieval in PubMed, Web of Science, Embase, and Cochrane Library databases to identify all published articles. The search terms were used as follows: (hematological cancers OR hematologic malignancies OR hematologic malignant tumors OR leukemia OR multiple myeloma OR lymphoma OR lymphoproliferative OR myeloproliferative) AND (ulcerative colitis OR inflammatory bowel disease OR Crohn's disease OR IBD OR UC OR CD). The retrieval period was from January 2000 to June 2023. In addition, a manual search of references to relevant literature was conducted to avoid the omission of any potential articles.

2. Inclusion and exclusion criteria

According to the PICOS principle, the inclusion criteria of the articles were as follows: (1) population: people without hematologic malignancies or related high-risk factors; (2) exposure: patients diagnosed with IBD; (3) control: people who did not suffer from IBD; (4) outcome: the incidence of hematologic malignancies; or (5) study design: the cohort studies. The exclusion criteria were as follows: (1) full texts were not available; (2) articles were published before 2000; (3) articles were not prepared in English; (4) the patients had developed hematologic malignancies before the diagnosis of IBD was made; or (5) when encountering different articles with the same source of the cohort data, more comprehensive or newer updated studies were included.

3. Outcome measures

The main outcome indicator was the incidence of hematologic malignancies in patients with IBD during the follow-up. Malignant hematologic diseases include lymphoma, leukemia, multiple myeloma, and so on.

4. Data extraction

Two investigators independently extracted data from all eligible studies, and a third adjudicator would review and confirm the ultimate results. The following information was collected: (1) research characteristics (authors, publication year, duration of cohorts, and countries); (2) patient characteristics (number of patients, age, gender, type of IBD, ulcerative colitis (UC) or Crohn's disease (CD) phenotypes, pharmacotherapy, and regions); or (3) the specific tumor types of hematologic malignancies and their corresponding overall incidence rates.

5. Quality assessment

Two investigators evaluated the quality independently. When there was a disagreement, a third investigator could be introduced to resolve the contradiction. We used the Newcastle-Ottawa Scale (NOS) to evaluate the potential risk of bias.⁹ NOS evaluates the risk of bias by evaluating the quality of the cohort study in terms of selection, comparability, and outcomes. In the process of evaluation, each aspect is assigned a score out of a total score of 9. The score of each aspect can be modified according to the specific situation of the study, making the NOS more suitable for evaluating specific studies. Articles with a score of 0–6 are defined as low quality, and articles with a score of 7–9 are regarded as high quality.

6. Statistical analysis

Statistical analyses were carried out using Stata 12.0 (Stata Statistical Software for data science, College Station, TX, USA). Standardized incidence ratio (SIR) with 95% confidence intervals (CI) was calculated to evaluate the incidence of hematologic malignancies in IBD patients. Heterogeneity across studies was estimated by the chi-square test and quantified by I² statistics,¹⁰ with I²≤50% representing low heterogeneity, 50%2≤75% representing moderate heterogeneity, and I²>75% representing high heterogeneity. Considering the potential and inevitable heterogeneity among included articles, such as ethnicity, IBD type, and hematologic malignancy type, all analyses in this study were performed using a random-effects model to improve the credibility of the results.¹¹ Additionally, subgroup analyses were conducted to explore the sources of heterogeneity, and sensitivity analysis was used to test the stability of the results. The publication bias was evaluated by Egger test.¹² p<0.05 was considered t statistically significant in bilateral tests.

1. Studies selection

In this study, a total of 8,619 studies were retrieved from four databases. After eliminating duplicate records, 5,514 articles remained. For articles obtained without other sources, 5,427 unrelated records were eliminated through titles and abstracts, leaving 87 studies. By manually scrutinizing the full texts of 87 studies, 67 studies were excluded for the following reasons: 18 were duplicated database sources, 29 had nonsignificant results, nine were without available data, 11 were non-cohort studies, and 20 studies were finally included.^8,13-31 The specific results are shown in Fig. 1.

Figure 1. Flow diagram of study selection.

2. Characteristics of the included studies

From 2000 to 2023, there were 20 cohort studies to observe the incidence of hematologic malignancies in patients with IBD and to compare it with the non-IBD population. These 20 studies involved 756,377 subjects from nine national cohorts and 11 regional cohorts covering five continents, including five in Asia (China, South Korea, and Japan), three in North America (USA), and 11 in Europe (Hungary, Netherlands, Denmark, Finland, Switzerland, Italy, Swiss, Norway, France, Britain, and Sweden). The earliest cohort started in 1940 and the latest cohort ended in 2019. The mean follow-up period of all studies ranged from 3 years to 19.9 years. Of these 20 studies, four reported the proportion of thiopurine used in IBD patients and four reported the steroid used in patients with IBD. Diseases can be classified into different phenotypes according to the different positions of the diseases in the Montreal classification. Seven studies reported the proportion of patients with UC patients in the four phenotypes (proctitis, left-sided colitis, extensive colitis, and unknown), and eight studies reported four phenotypes in CD patients (terminal ileum, colonic, ileocolonic, and upper gastrointestinal). The details are shown in Table 1 and Supplementary Table 1.

Table 1 . Characteristics of All the Studies Included in the Meta-Analysis.

Author (year)	Country	Cohort year	No. of patient	CD/UC	UC phenotype, %	CD phenotype, %
Ashworth et al. (2012)13	USA	1979–2008	1,374	791/535 Unspecified 48	NA	NA
Lakatos et al. (2013)14	Hungarian	1977–2008	1,420	506/914	E1: 25.4 E2: 50.2 E3: 24.4	L1: 32.8 L2: 35.9 L3: 30.6 L4: 0.7
Yano et al. (2013)15	Japan	1985–2010	770	770	NA	L1: 29.6 L2: 14.7 L3: 55.7
Yadav et al. (2015)16	USA	1940–2011	839	377/462	NA	NA
van den Heuvel et al. (2016)17	Netherlands	1991–2013	2,801	1,157/1,644	E1: 34 E2: 48 E3: 18	L1: 43 L2: 32 L3: 23 L4: 11
Wang et al. (2016)18	China	1998–2013	3,348	685/2,663	NA	NA
So et al. (2017)19	China	1990–2016	2,621	1,018/1,603	E1: 35 E2: 30.8 E3: 34.2	L1: 24.5 L2: 31.7 L3: 3.8 L4: 8.1
Malham et al. (2019)20	Denmark and Finland	1992–2014	6,689	2,921/3,741	NA	NA
Scharl et al. (2019)21	Swiss	2006–2016	3,119	1,474/1,106	E1: 16.3 E2: 33.1 E3: 50.6	L1: 30.2 L2: 29.6 L3: 37.8 L4: 2.4
Taborelli et al. (2020)8	Italy	1995–2013	3,664	1,306/2,358	NA	NA
Zhang et al. (2022)22	China	1998–2018	1,385	516/869	E1: 32.6 E2: 12.3 E3: 42.6 Unknown: 12.5	L1: 37.2 L2: 29.3 L3: 36.4 L4: 5.8
Olén et al. (2023)23	Sweden and Denmark	1969–2019	164,716	49,237/99,492 Unspecified 15,978	E1: 13.1 E2: 15.9 E3: 50.2 Unknown: 20.8	L1: 34.7 L2: 36.3 L3: or not defined 29 Perianal disease: 12.2
Yu et al. (2023)24	Norway and Sweden	1987–2015 1993–2016	131,492	42,247/78,344 Unspecified 10,901	E1: 15.7 E2: 16.8 E3: 50.1 Unknown: 17.2	L1: 21 L2: 20.4 L3: 43.3 Unknown: 15.1
Lopez et al. (2014)25	France	2004–2007	19,486	11,759/7,044 Unspecified 683	NA	NA
Chiorean et al. (2011)26	USA	1990–2009	3,585	2,277/1,308	NA	NA
Beaugerie et al. (2009)27	France	2004–2007	19,486	11,759/7,044 Unspecified 683	NA	NA
Palli et al. (2000)28	Italy	1978–1992	920	231/689	NA	NA
Lewis et al. (2001)29	UK	1988–1997	16,996	6,605/10,391	NA	NA
Jung et al. (2023)30	Korea	2008–2019	90,370	23,640/66,730	NA	NA
Farrell et al. (2000)31	Republic of Ireland	1990–1999	782	267/515	NA	NA

In the Montreal classification, disease extent (E) for UC is divided into three categories. The E1 describes patients with proctitis, E2 describes patients with left-sided colitis, and E3 describes patients with extensive colitis. In the Montreal classification, disease location (L) for CD is divided into four categories. The L1 describes the disease is confined to terminal ileum, L2 describes the disease is confined to colonic, L3 describes the disease involves ileocolonic; L4 describes the disease involves upper gastrointestinal..

CD, Crohn's disease; UC, ulcerative colitis; NA, not available..

3. Quality assessment

NOS was used to evaluate the quality of 20 cohort studies. The scoring results showed that all the articles used in this meta-study scored ≥6, demonstrating the high quality of the articles (Supplementary Table 2).

4. Analysis of the primary result

Combining the statistical results of the 20 cohort studies, the incidence of hematologic malignancies was higher in patients with IBD compared to the non-IBD population, with an SIR of 3.05 (95% CI, 1.77 to 5.27; p<0.001). However, this result had high heterogeneity with I²=94% (Fig. 2).

Figure 2. Forest plot of the relationship between inflammatory bowel disease and the incidence of hematologic malignancies (p<0.001). SIR, standardized incidence ratio; CI, confidence interval.

5. Subgroup analysis

Depending on the specific type of IBD disease, IBD patients can be divided into UC patients and CD patients. Compared to the non-IBD population, UC patients had a higher incidence of hematologic malignancies (SIR=2.29, p=0.05), and similarly, CD patients had a higher incidence of hematologic malignancies (SIR=3.56, p=0.005) (Fig. 3).

Figure 3. (A) Forest plot of the relationship between ulcerative colitis and the incidence of hematologic malignancies (p=0.05). (B) Forest plot of the relationship between Crohn’s disease and the incidence of hematologic malignancies (p=0.005). SIR, standardized incidence ratio; CI, confidence interval.

Hematologic malignancies can be divided into non-Hodgkin's lymphoma (NHL), Hodgkin's lymphoma (HL), and leukemia. The incidence of different types of hematologic malignancies was higher in IBD patients compared to controls, with the SIR of 1.70 (95% CI, 1.14 to 2.54; p=0.01) for NHL, 3.47 (95% CI, 1.56 to 7.73; p=0.002) for HL and 3.69 (95% CI, 2.14 to 6.37; p<0.001) for leukemia (Fig. 4).

Figure 4. (A) Forest plot of the relationship between inflammatory bowel disease (IBD) and the incidence of non-Hodgkin’s lymphoma (p=0.01). (B) Forest plot of the relationship between IBD and the incidence of Hodgkin's lymphoma (p=0.002). (C) Forest plot of the relationship between IBD and the incidence of leukemia (p<0.001). SIR, standardized incidence ratio; CI, confidence interval.

In terms of the use of therapeutic drugs, the three groups, including immunomodulators, anti-tumor necrosis factor (TNF) and thiopurines, encompassed the outcomes of IBD patients who had utilized the respective medications, irrespective of whether the administration was as monotherapy or as part of a combination therapy regimen. The incidence of hematologic malignancies was higher in IBD patients treated with immunomodulators compared to the control group (SIR=4.46, p<0.001) (Fig. 5). According to the specific drug type, the SIR for IBD patients who used anti-TNF drugs was 5.10 (95% CI, 3.52 to 7.39; p<0.001), while the SIR for IBD patients who used thiopurine was 3.80 (95% CI, 2.46 to 5.87; p<0.001) (Table 2).

Figure 5. Forest plot of the relationship between inflammatory bowel disease patients using immunosuppressants and the incidence of hematologic malignancies (p<0.001). SIR, standardized incidence ratio; CI, confidence interval.

Table 2 . Subgroup Analysis of the Relationship between Inflammatory Bowel Disease and Hematologic Malignancies.

Types of subgroup	No. of studies	SIR	95% CI	p-value	Heterogeneity (I²), %
Age
<40 yr	6	2.49	1.67–3.69	<0.001	29.7
≥40 yr	3	2.44	1.15–5.19	0.02	84.7
Drug
Anti-TNF	3	5.10	3.52–7.39	<0.001	0
Thiopurine	6	3.80	2.46–5.87	<0.001	0
Region
Asia	5	3.60	2.37–5.45	<0.001	0
Europe	12	2.78	1.29–6.01	0.009	96.5
America	3	2.50	1.42–4.39	0.001	21.6

SIR, standardized incidence ratio; CI, confidence interval; TNF, tumor necrosis factor..

Compared with the non-IBD population, the incidence of hematologic malignancies was higher in patients with IBD, both younger than 40 years old (SIR=2.49, p<0.001) and older than 40 years old (SIR=2.44, p=0.02). Geographically, the incidence of hematologic malignancies was higher in IBD patients than in the non-IBD population, either in Asia (SIR=3.60, p<0.001), Europe (SIR=2.78, p=0.009) or the United States (SIR=2.50, p=0.001) (Table 2).

6. Publication bias and sensitivity analysis

The Egger test in Stata12.0 was applied to visually measure publication bias, and the result was p=0.768, suggesting no obvious publication bias (Supplementary Fig. 1). By excluding individual studies one by one, the sensitivity of the meta-analysis of the correlation between IBD and hematologic malignancy was analyzed. The results suggested that the combined statistical results were stable (Supplementary Fig. 2).

Combined with the statistical results of 20 cohort studies, the incidence of hematologic malignancies is higher in patients with IBD compared to the non-IBD population, a phenomenon observed in both UC and CD patients, with a higher incidence in CD. Among hematologic malignancies, IBD patients were at higher risk of developing NHL, HL, and leukemia. The incidence of hematologic malignancies in IBD patients treated with immunomodulators. The results were consistent across geographical environments. To summarize, there is a potential correlation between IBD and hematologic malignancies.

Due to the immune imbalance and chronic inflammation in the intestines in IBD patients, there can be changes in their immune components, which in turn affect the immune response of tumors. For example, human leukocyte antigen G (HLA-G) is a potent immunosuppressive molecule that binds to immunosuppressive receptors. Expression of HLA-G induced by microenvironment cues, including cytokines, may participate in anti-inflammatory immune responses and malignant transformation. The expression of HLA-G during chronic inflammation leads to the growth of transformed cells beyond the recognition range by host immune surveillance.³² The findings show that the risk of malignant tumor development is increased in the case of intestinal inflammatory diseases, such as UC and CD. At the same time, the report shows that the increase of HLA-G is common in hematologic malignancies. In a study by Martín-Lorenzo et al.,³³ HLA-G was also found to increase significantly in B-cell chronic lymphocytic leukemia. HLA-G is preferentially detected in tumor tissues but rarely in adjacent normal tissues, suggesting a specific correlation with tumor growth and progression. Aberrant expression of HLA-G in cancer has been associated with the establishment of anti-tumor immune responses and tumor escape. Studies have shown that in HL patients, 54% of HLA-G protein is up-regulated, suggesting that increased expression of HLA-G proteins by tumor cells may be an immune escape mechanism.³³

Patients with IBD are chronically exposed to an inflammatory environment in the body that may lead to abnormal gene mutations, which in turn increase the risk of hematologic malignancies. Clonal hematopoiesis (CH) is a precancerous lesion, a somatic mutation in hematopoietic stem cells and progenitor cells that increases the risk of hematologic malignancies.³⁴ There is growing evidence of a correlation between inflammation and CH.³⁵ Its presence in the context of IBD has been well established. Several studies have detected multiple mutations in several genes in patients with IBD who exhibit one or more mutations. Chronic infections and inflammatory responses caused by intestinal dysfunction may promote the selection of CH-related DNMT3A gene mutation through interferon-γ signal transduction induced by these diseases.³⁵ The inflammatory environment may selectively induce the development of CH-specific mutations, which will underlie the biological basis of hematologic malignancies.³⁶

The imbalance of the intestinal flora and the disruption of the mucosal barriers may play an important role in the development of hematologic malignancies in patients with IBD.³⁷ Some host bacteria are directly involved in promoting oxidative stress, leading to DNA damage and consequently cancer.^38,39 Studies have shown that the microbiota-intestinal-lymphoma axis is one of the pathways of lymphoma.⁴⁰ The most common site of primary extranodal lymphoma is the gastrointestinal tract and a large number of lymphoma patients have documented intestinal dysbiosis.⁴¹ It has been proven that Helicobacter pylori is a risk factor in more than 90% of cases of gastric mucosa-associated lymphoid tissue lymphoma,⁴² and the same pathological pattern exists in mice with chronic H. pylori infection, especially Felis H. pylori.⁴³ In addition, studies have shown that bacteria such as Campylobacter jejuni, Chlamydia psittaci, and Borrelia berenii are also involved in the development of human lymphoma.³³ It has been found that a series of bacteria causing intestinal infections, such as Campylobacter,⁴⁴ will increase the risk of IBD, and bacterial infections recur throughout the course of the disease in IBD patients. Therefore, bacteria may play an important role in the process of hematologic malignancies in patients with IBD.

Through the above potential mechanisms, IBD patients may be at increased risk of developing hematologic malignancies, including leukemia and lymphoma. In this study, patients with IBD had the highest incidence of leukemia compared to NHL and HL. One possible reason for this is that patients with IBD have genes that are prone to mutate abnormally in response to factors such as inflammation, and the resulting CH is more likely to evolve into myeloid malignant tumors. The clonal hematopoietic population eventually progresses to myelodysplastic syndrome or acute myeloid leukemia, whereas the malignant tumor of the lymphatic system is less likely to develop into leukemia.⁴⁵ The higher risk of developing HL in patients with IBD compared to NHL may be related to the role of inflammatory factors in HL. For example, the up-regulation of the HLA-G protein mentioned above is more pronounced in patients with HL. In addition, the choice of therapeutic agents for IBD may also affect the types of hematologic malignancies to some extent. The thiopurine mentioned below makes lymphoma relatively common.

Immunosuppressants and biological agents are increasingly being used to treat IBD, making it necessary to consider the risks of therapeutic drugs. The present study found that patients who used thiopurine or anti-TNF-α drugs have an increased risk of hematologic malignancies. Some studies believed that TNF-α is a cytokine involved in systemic inflammation and immune system regulation, and that inhibition of TNF-α favors the process of tumor formation.⁴⁶ Two experimental studies support the view that anti-TNF promotes tumor formation, either alone or in combination with thiopurine, and contributes to the development of lymphoma.^47,48 One study found that anti-TNF could inhibit the activity of natural killer cells and harm anti-lymphoma activity in vitro.⁴⁷ Another study showed that the lymphoma-like transformation driven by the Epstein-Barr virus was amplified by anti-TNF and 6-mercaptopurine in the lymphoma model in vitro.⁴⁸

This study found that the incidence of hematologic malignancies in CD patients was higher than that in UC patients. The reason may be that UC lesions are mainly located in the large intestine, and retrograde development may also involve the entire colon and the terminal ileum. However, CD lesions may involve both the terminal ileum and the adjacent right colon at the same time, or only the small intestine or colon, as well as the oral cavity, esophagus, stomach, and duodenum. Lesions that are segmental or jumping, with extensive inflammatory involvement, tend to be more severe, which also causes heavier dosages and duration of treatment in patients with CD than in patients with UC. In addition, CD patients are exposed to higher levels of harmful diagnostic radiation from repeated radiography, which may also be related to their increased risk of developing hematologic malignancies.⁴⁹

This study found that, compared to other regions, Asian patients with IBD have a higher risk of developing hematologic malignancies. It is generally recognized that environmental pollution is often more severe in developing countries due to industrial and economic factors, potentially increasing risks in Asian regions. Studies have shown that hematologic malignancies are positively correlated with factors such as ultraviolet exposure and pesticide use.^50,51 Moreover, anti-inflammatory dietary patterns are more common in Europe than in Asia, which may play a role in the hematologic malignancy risks among Asian IBD patients. Mediterranean diet is considered a beneficial diet for patients with IBD due to its modulating effect on inflammation. The research shows that the high score of the Mediterranean diet is negatively correlated with the risk and progression of IBD.⁵² Therefore, the areas dominated by the Mediterranean diet may inhibit the development of inflammation during the process of IBD because of its anti-inflammatory effect, thereby reducing the occurrence of hematologic malignancies.

This study is based on cohort studies with a high degree of confidence. In contrast to a previous study, which had merely briefly touched upon the incidence of different hematologic malignancy types in patients with IBD subtypes,⁷ our study delves into a comprehensive analysis of the incidence of different hematologic malignancy types in IBD patients, as well as the impact of other factors such as pharmacotherapy, age and region on the risk of hematologic malignancies. This study endeavored to offer fresh perspectives through a more thorough analysis, encompassing more nuanced subgroup analyses, and has yielded more pronounced results. The limitation is that this study does not have enough data to study the influence of the dose effect of drugs on the occurrence of hematologic malignancies in IBD patients and the impact of monotherapy and combination therapy on the incidence of hematologic malignancies. At the same time, due to the lack of data, it was not possible to evaluate the influence of follow-up time on the occurrence of hematologic malignancies in IBD patients. In addition, the high heterogeneity (I²=94% in Fig. 2) can lead to decreased credibility of the statistical results.

In conclusion, this meta-analysis provides the latest information on the incidence of hematologic malignancies in patients with IBD. The incidence of hematologic malignancies is higher in patients with IBD than in the non-IBD population, including NHL, HL, and leukemia. This relationship may be influenced by factors such as the type of IBD, pharmacotherapy, and other variables. Additionally, more large multicenter cohort studies are needed to confirm the phenomenon.

No potential conflict of interest relevant to this article was reported.

Study concept and design: L.T. Data acquisition: X.Z. Data analysis and interpretation: Q.Z. Drafting of the manuscript: Q.Z. Critical revision of the manuscript for important intellectual content: D.W. Statistical analysis: Z.X. Obtained funding: L.T. Administrative, technical, or material support; study supervision: J.R. Approval of final manuscript: all authors.

Supplementary materials can be accessed at https://doi.org/10.5009/gnl240119.

The datasets supporting this article’s conclusions are included within the article and its additional files.