The association between type 1 and 2 diabetes mellitus and the risk of leukemia: a systematic review and meta-analysis of 18 cohort studies

Abstract

Diabetes mellitus (DM) is widely considered to be associated with the risk of diverse cancers; however, the association between DM and the risk of leukemia is still controversial. Thus, a detailed meta-analysis of cohort studies was conducted to elucidate this association. Eligible studies were screened through the electronic searches in PubMed, Web of Science, and Embase from their inception to August 11, 2020. Summary relative risks (RRs) and 95% confidence intervals (CIs) were computed through the random-effects model. Eighteen articles involving 10,516 leukemia cases among a total of 4,094,235 diabetic patients were included in this meta-analysis. Overall, twenty-five RRs were synthesized for type 2 diabetes mellitus (T2DM) and yielded a summary RR of 1.33 (95%CI, 1.21–1.47; p < 0.001). For type 1 diabetes mellitus (T1DM), 7 RRs were combined, however, the pooled RR was insignificant (RR, 1.08; 95%CI, 0.87–1.34; p = 0.48). Interestingly, the summary RR for East Asia (RR, 1.83, 95%CI, 1.63–2.06) was significantly higher than that for Europe (RR, 1.11, 95%CI, 1.06–1.15), Western Asia (RR, 1.40, 95%CI, 1.25–1.54), North America (RR, 1.14, 95%CI, 1.08–1.20), and Australia (RR, 1.47, 95%CI, 1.25–1.71). Moreover, we found that patients with a shorter T2DM duration (1–5 years) had a higher risk of leukemia compared to those with a longer duration (5.1–10 years). Overall, this meta-analysis suggests there is a moderately increased risk of leukemia among T2DM patients, but not in T1DM patients. Further investigation is warranted.

LEUKEMIA is a cluster of life-threatening haemopoietic cancers [1], characterized by abnormal proliferation and development of white blood cells and their precursors [2]. Leukemia was reported to be the 15th and 11th most common cause of cancer occurrence and death worldwide, respectively, with an estimated 437,033 new leukemia cases and 309,006 leukemia deaths in 2018 [3].

The global upsurge in the incidence of diabetes mellitus (DM) has become one of the largest challenges to public health in the 21st century [4]. As documented in the International Diabetes Federation (IDF) Diabetes Atlas 2017, 425 million individuals aged between 20 to 79 years suffered from diabetes in 2017, which increased by 10 million compared with 2015 [5]. Moreover, diabetes has been reported as a predisposing factor for the onset of diverse cancers, including the liver, bladder, colorectal, breast, ovarian, endometrial and non-Hodgkin’s lymphoma [6-9].

The risk of leukemia among diabetic patients has been explored by some epidemiological studies, but the results were inconsistent. A meta-analysis of 11 observational studies in 2012 revealed a 22% increased risk of leukemia in patients suffering from type 2 diabetes mellitus (T2DM) as compared to the euglycemic population [10]. However, the association between T1DM and leukemia risk remains ambiguous. A number of new cohort studies investigated the association between T2DM and leukemia [11-21]. Therefore, we conducted an updated meta-analysis of cohort studies to provide a more accurate estimate of the association between diabetes and the risk of development of leukemia.

Methods

Search strategy and selection criteria

To identify eligible studies investigating the association between diabetes and risk of leukemia, two researchers (YPF and WYB) independently conducted the literature search in PubMed, Web of Science, and Embase from their inception to August 11, 2020. The following search key terms were used in combination: “diabetes mellitus”, “DM”, “leukemia”, “leukaemia”, “cohort”, “follow-up”, “prospective” and “longitudinal” (Supplementary Table 1). The reference lists of previously published systematic reviews were also examined to find additional published articles.

We included studies that met the following selection criteria: 1) the study was cohort design; 2) the association between diabetes mellitus (type 1 diabetes mellitus (T1DM) or T2DM) and leukemia risk was reported; 3) the study reported the effect sizes (ESs) (standard incidence ratio (SIR), relative risk (RR), or hazard ratio (HR)) and their corresponding 95% confidence intervals (CIs). If there were multiple published papers from the same population or cohort, priority was given to the study with the longest follow-up time. Studies in which gestational diabetes was the exposure of interest were excluded. The researchers scanned the titles and abstracts to exclude the irrelevant articles, and the eligibility of the remaining studies were determined by reading the full text.

Data extraction and quality assessment

The baseline characteristics of each included study was extracted independently by two reviewers (YPF and WYB): the first author’s last name, year of publication, country, duration of follow-up, participant characteristics (number of cases, sex, age range or mean age at entry), source of control group (population-based or hospital-based), number of leukemia cases, identification of DM and leukemia, RRs with their 95%CIs, and adjusted confounders in the analysis. When studies reported several multivariate adjusted effect estimates, we extracted the RRs with the greatest control for confounders. Two reviewers (YPF and WYB) checked each other’s data after the extraction process and disagreements were addressed by discussion with the third investigator (ZJZ).

Two researchers (YPF and WYB) independently assessed the quality of the included studies based on the Newcastle-Ottawa scale (NOS) [22]. There were 3 parameters to assess cohort studies: selection (4 points), comparability (2 points) and outcome assessment (3 points). Studies scoring 7–9 points were considered to be of high quality, studies scoring 4–6 points were defined as moderate quality, and 0–3 as low quality.

Statistical analysis

Summary relative risks (RRs) and corresponding 95% CIs were calculated to explore the association between DM and leukemia risk through random effects model [23]. We synthesized RRs according to different subtype (T1DM and T2DM) of DM respectively. Since T2DM accounted for the vast majority of diabetes worldwide (about 90%), studies that did not distinguish diabetes subtypes and only reported RRs for the pooled diabetes in adults were assigned to the T2DM group. We used the Cochran Q and I² statistics to evaluate statistical heterogeneity among included studies. A p value of <0.10 or an I² statistic of ≥50% is considered to represent substantial heterogeneity among studies.

Subgroup and sensitivity analyses were conducted for all included studies and studies which focused on T2DM. Subgroup and meta-regression analyses were stratified by effect estimate (RR&HR, and SIR), gender (males and females), study location (Europe, Western Asia, East Asia, North America, and Australia), control population (DM-free as control and population as control), the number of cases (≤100, >100), follow-up time (<10 years, and ≥10 years), T2DM measured by robust methods (hospital diagnosis, medication use, and database records), NOS score (<7 and ≥7), and adjustments of potential confounding factors, such as alcohol use, smoking, BMI/obesity, race/ethnicity and socioeconomic status (Yes vs. No). For sensitivity analyses, we excluded one study at a time and recalculated the summary RRs to assess whether the single study had a significant effect on the overall estimate. However, we did not conduct these analyses for studies investigating T1DM because of the limited number of studies.

We used funnel plots to assess publication bias. Furthermore, Egger’s linear regression tests were conducted and a p value of <0.10 indicated potential publication bias. All statistical analyses were performed using Stata, version 14.0 (Stata Corp, College Station, Texas).

Results

Online searching retrieved a total of 1,466 relevant articles originally, whereas 15 additional studies were retrieved through reference searches. The meta-analysis finally incorporated a total of 18 eligible articles [11-21, 24-30], one of which was obtained by manually searching the reference list of relevant literature [26]. The identification process of relevant studies was illustrated in Fig. 1.

Fig. 1

PRISMA flow diagram for identification of relevant articles for the meta-analysis

Characteristics in included studies

The baseline characteristics of the eligible studies are displayed in Supplementary Table 2. We included 22 cohort studies involving 10,516 leukemia cases among a total of 4,094,235 diabetic patients (the study of Carstensen et al. [16] included five cohorts). The publication year of these articles ranged from 1982 to 2019. Of these 18 articles, 16 focused on both types of DM or specifically investigated T2DM [11-15, 17-21, 24-27, 29, 30], while T1DM was exclusively explored in only 5 of these articles [14, 16, 25, 27, 28]. The studies were conducted in the following countries: UK (n = 4) [16, 19, 27, 30], the United States (n = 3) [12, 24, 29], China (n = 3) [11, 13, 20], Sweden (n = 3) [15, 16, 28], Finland (n = 2) [16, 21], Australia (n = 2) [14, 16], Denmark (n = 2) [16, 25], South Korea (n = 1) [26], Italy (n = 1) [18], Israel (n = 1) [17].

Among these 18 studies, 10 studies [11, 13-15, 18, 20, 21, 25, 27, 28] applied SIR to identify the association between DM and risk of leukemia, while the remaining 8 studies [12, 16, 17, 19, 24, 26, 29, 30] adopted incidence rate (RR or HR). Fifteen studies were retrospective cohort studies based on registry-data [11, 13-21, 25, 27-30] and 3 were prospective cohort studies [12, 24, 26]. Besides, out of 18 studies, 13 were population-based [11-14, 16-18, 21, 24-28] and 5 studies [15, 19, 20, 29, 30] were based on hospital. Furthermore, hospital diagnostic records and cancer registry data were utilized in all these studies to ascertain leukemia cases, which was reliable and robust compared to the self-report methods. Nine studies [12, 13, 15-17, 19, 28-30] were categorized as high quality, whereas the other 9 studies [11, 14, 18, 20, 21, 24-27] as moderate quality according to the Newcastle-Ottawa quality tool (Supplementary Table 3).

DM and leukemia risk

As depicted in Fig. 2, twenty-five risk estimates were synthesized for T2DM, and the summary RR was 1.33 (95%CI, 1.21–1.47; p < 0.001) with high heterogeneity (Q = 239.97; p < 0.001; I² = 90.0%). However, the summary RR for T1DM was 1.08 (95%CI, 0.87–1.34; p = 0.48) and moderate heterogeneity was detected (Q = 10.89; p = 0.09; I² = 44.9%).

Fig. 2

Summary RR of diabetes mellitus on leukemia risk. Data are presented as RR for each study (boxes), 95%CIs (horizontal lines) and summary as RR with 95%CI (diamond). RR, relative risk; CI, confidence interval; F, females; M, males; LL, lymphoid leukemia; ML, myeloid leukemia; AML, acute myeloid leukemia; CML, chronic myeloid leukemia

Results of subgroup analysis were represented in Table 1. When grouped by gender, an enhanced risk of leukemia for both males (the pooled RR, 1.35; 95%CI, 1.15–1.59; I² = 93.3%) and females (the pooled RR, 1.41; 95%CI, 1.07–1.87; I² = 94.8%) were observed among the participants suffering from DM. Interestingly, the summary RR of studies conducted in East Asia (RR, 1.83; 95%CI, 1.63–2.06; I² = 60.9%) was significantly higher than that for Europe (RR, 1.11; 95%CI, 1.06–1.15; I² = 0.0%), Western Asia (RR, 1.40; 95%CI, 1.25–1.54; I² = 0.0%), North America (RR, 1.14; 95%CI, 1.08–1.20; I² = 0.0%) and Australia (RR, 1.47, 95%CI, 1.25–1.71; I² = 66.2%). According to the univariate meta-analysis, the primary source of variability between studies was the study location (I-squared_res = 2.16%; Adjusted R Square = 97.80%). Moreover, the studies that adjusted for BMI/obesity (RR, 1.13, 95%CI, 1.07–1.19; I² = 0.0%) were also found to have lower risk of leukemia with T2DM compared with those that did not (RR, 1.42, 95%CI, 1.26–1.60; I² = 91.1%).

Table 1 The results of subgroup analyses for association between T2DM and leukemia

Subgroup		Number of studies	RR (95%CI)	Test(s) of heterogeneity			p values for meta-regression
				Q	p	I2
Effect size	SIR	9	1.47 (1.25–1.71)	207.88	<0.001	94.7%	0.089
	RR&HR	7	1.22 (1.12–1.33)	19.96	0.068	39.9%
Gender	Male	10	1.35 (1.15–1.59)	148.61	<0.001	93.3%	0.821
	Female	8	1.41 (1.07–1.87)	154.22	<0.001	94.8%
Study location	East Asia	4	1.83 (1.63–2.06)	12.77	0.026	60.9%	0.001
	Europe	7	1.11 (1.06–1.15)	3.31	0.973	0.0%	0.002
	North America	3	1.14 (1.08–1.20)	1.37	0.712	0.0%	0.008
	Australia	1	1.47 (1.25–1.71)	2.96	0.085	66.2%	0.631
	Western Asia	1	1.40 (1.25–1.54)	0.02	0.892	0.0%	Reference
Control population	DM-free as control	6	1.29 (1.10–1.50)	110.57	<0.001	90.1%	0.572
	Population as control	10	1.36 (1.20–1.55)	163.40	<0.001	93.9%
Number of leukemia	≤100	8	1.26 (1.08–1.46)	19.53	0.021	53.9%	0.523
	>100	8	1.37 (1.20–1.56)	206.64	<0.001	94.70%
Follow-up time	<10	4	1.59 (1.07–2.37)	33.2	<0.001	84.9%	0.337
	≥10	12	1.30 (1.18–1.44)	188.7	<0.001	90.5%
T2DM measured by robust methods	Yes	9	1.44 (1.25–1.67)	204.28	<0.001	94.1%	0.094
	No	7	1.19 (1.10–1.29)	15.07	0.179	27.0%
NOS scrore	<7	9	1.47 (1.23–1.75)	99.47	0	89.9%	0.196
	≥7	7	1.26 (1.10–1.43)	132.97	0	90.2%
Adjusted for alcohol use	Yes	4	1.15 (1.03–1.29)	7.97	0.158	37.2%	0.110
	No	12	1.41 (1.24–1.59)	211.13	<0.001	91.5%
Adjusted for somking	Yes	3	1.16 (0.97–1.39)	7.86	0.097	49.1%	0.194
	No	13	1.38 (1.24–1.55)	230.24	<0.001	91.7%
Adjusted for BMI/obesity	Yes	3	1.13 (1.07–1.19)	1.59	0.811	0.0%	0.035
	No	13	1.42 (1.26–1.60)	213.35	0	91.1%
Adjusted for race/ethnicity	Yes	2	1.29 (1.10–1.50)	10.85	0.004	81.6%	0.825
	No	14	1.34 (1.19–1.51)	220.69	<0.001	90.5%
Adjusted for socioeconomic status	Yes	2	1.29 (1.12–1.49)	8.87	0.012	77.5%	0.847
	No	14	1.34 (1.19–1.51)	228.6	<0.001	90.8%

In addition, when excluded one study at a time and recalculate the pooled RRs, the results did not appreciably change and the estimates in each case were well within the confidence limits of the overall estimate (Supplementary Fig. 1).

T2DM duration and leukemia risk

The effect of T2DM duration on the risk of leukemia was elucidated in only 5 of the 18 included studies. As illustrated in Table 2, the highest risk of leukemia occurrence was discovered in patients within one year of T2DM diagnosis (RR, 3.48; 95%CI, 1.75–6.94; p < 0.001). The positive association was also obtained in the 1–5 years group (RR, 1.17; 95%CI, 1.04–1.33; p = 0.012) and 5.1–10 years group (RR, 1.10; 95%CI, 1.01–1.20; p = 0.029). However, when T2DM was diagnosed ≥10 years, the pooled RR reduced to 1.03 (95%CI, 0.90–1.19; p = 0.644).

Table 2 Summary RRs for the association between T2DM and leukemia according to diabetes duration

Diabetes duration, years	References	Number of RR	RR (95%CI)	p
<1	[15, 17]	3	3.48 (1.75–6.94)	<0.001
1–5	[17, 30]	2	1.17 (1.04–1.33)	0.012
5.1–10	[17, 30]	2	1.10 (1.01–1.20)	0.029
≥10	[17, 30]	2	1.03 (0.90–1.19)	0.644
>1	[14, 15, 19, 28, 30]	7	1.40 (1.16–1.70)	0.001
>2	[14, 15, 17, 30]	9	1.18 (1.07–1.30)	0.001

Publication bias

The funnel plot for meta-analysis could not rule out potential publication bias (Supplementary Fig. 2). The Egger’s test of all included risk estimates showed no evidence of publication bias (p value for T1DM = 0.635; p value for T2DM = 0.315).

Discussion

Evaluating the 18 articles, including 22 cohort studies, a 33% elevated risk of leukemia was found to be associated with T2DM. However, this significantly positive association was not witnessed in T1DM. According to the subgroup analyses, both male and female T2DM patients developed an elevated risk of leukemia as compared to the general population or with those without DM. Furthermore, the risk of leukemia appeared to be inversely associated with the duration of T2DM, which may be related to the treatment of diabetes. When T2DM was diagnosed over 10 years, there was no significant difference in the risk of leukemia in diabetics compared to the general population. Although some variations were marked in the pooled RRs of sensitivity analyses, all of them had point estimates greater than 1 and most of the corresponding CIs overlapped with that of the summary RR in overall meta-analysis. These findings presumed a robust positive association.

In the subgroup analysis of T2DM, compared with the western countries, East Asia portrayed a more substantial association between T2DM and leukemia. Differential genetic background and environmental factors, such as the country-specific dietary patterns, may justify this outcome. Higher levels of trans-fat and saturated fat prevailing the western dietary pattern [31] may promote cancer development [32, 33]. A multicenter case-control study in China reported that elevated risk of leukemia was correlated with dietary intake of animal fat and dietary habits with frequent intakes of fat, deep-fried, and smoked foods (p for trend <0.05) [34]. Moreover, Solans et al. estimated the association between overall diet and chronic lymphocytic leukemia (CLL), and found that high adherence to a western dietary pattern (i.e., high intake of high-fat dairy products, processed meat, refined grains, sweets, caloric drinks, and convenience food) was associated with CLL (OR, 1.19; 95%CI, 1.03–1.37) [35]. Notably, dietary pattern is a stable eating habit formed over a long period and hyperglycemia always exists before T2DM diagnosis. Therefore, in the case that individuals with hyperglycemia caused by the long-term high-fat diet were assigned to the control population, T2DM can only display a relatively weak impact on leukemia development.

The studies that adjusted for BMI/obesity were appeared to have a lower risk of leukemia with T2DM compared with those that did not. The association between BMI and the onset of leukemia is convincing. A meta-analysis including 65 different studies documented that the risk of leukemia, including acute, chronic myeloid lymphoma (AML and CML), and CLL, was aggravated with BMI (RR, 1.09; 95%CI, 1.03–1.15) [36]. However, only three of the included studies adjusted for BMI/obesity and these three studies controlled for many confounders. Therefore, it was difficult to reliably determine whether variation in relative risk was due to differences in obesity adjustment or other factors.

In our meta-analysis, we also investigated the impact of diabetes duration on the leukemia risk. The first year of follow-up estimated 3.48 folds risk of leukemia in T2DM than among the euglycemic population. The incidence of leukemia was also significantly increased in patients suffering from T2DM for 1–5 years and 5.1–10 years. However, the risk of leukemia declined to the level of the general population among patients with T2DM diagnosed for ≥10 years. Detection bias and reverse causality may account for the highest risk for leukemia observed in the first year following the diagnosis of diabetes. Leukemia cases identified due to increased screening or physical examination after diabetes diagnosis may be affected by detection bias. Besides, undiagnosed leukemia could indulge in insulin sensitivity and trigger various biological pathways to suppress insulin secretion [37], which may lead to diabetes. The risk of leukemia appears to be inversely associated with diabetes duration. Certain cancers, such as pancreatic cancer [38], non-Hodgkin’s lymphoma [6], exhibit a similar association with T2DM, though the mechanism has not yet been established. However, it should be noted that there are a limited number of studies investigating the effect of T2DM duration on the incidence of leukemia, and thus we can not rule out the possibility that this result was just accidental. The treatment of diabetes may be responsible for the inverse association between T2DM duration and leukemia incidence, if real. The hypoglycemic drugs reduce the fasting serum glucose of diabetic patients and maintain it at a normal level. This, in turn, forbids the development of hyperglycemia-induced leukemia.

We found no significant association between T1DM and leukemia risk. Acute lymphocytic leukemia (ALL) and T1DM is reported to share an environmental etiology and common epidemiological features [39], including similar age distributions and correlated international incidence, which suggested a significant positive correlation between them (r = 0.53). However, in another study, Feltbower et al. scrutinized the association between the two diseases within small areas in a region in the north of the United Kingdom and only reported a correlation coefficient of 0.33. Furthermore, after controlling the deprivation index, the correlation coefficient dropped to a negligible level (r = 0.06) [40]. Based on population-based registries, Richiardi et al. also analyzed the incidence of the two diseases and opined that their incidence was not correlated in the same birth cohorts (r = –0.34; p = 0.41) [41]. These findings support our results that T1DM was not associated with increased incidence of leukemia.

The pathophysiological mechanism behind the diabetes-induced increased leukemia risk remains unknown. Based on the following facts, we speculated that hyperglycemia might play a crucial role in the relationship between T2DM and leukemia risk. First, elevated fasting serum glucose levels have been identified as an independent risk factor for certain cancers, and the risk enhances with rising glucose levels [42]. Second, adequate research has substantiated the fact that hyperglycemia can trigger DNA damage and then cause DNA mutation, which may be responsible for the development of leukemia [43]. Third, some studies have also documented that the response to glucose process is associated with CML [44, 45].

Several strengths of the present study are detailed as follows. First, the available evidence was gathered and summarized from the cohort studies that dealt with the impact of DM on leukemia risk. The large sample size of our study was powerful enough to estimate the association between DM and the risk of leukemia. Second, the recall bias in case-control studies could be avoided as our meta-analysis concentrated only cohort studies. Third, this is the first meta-analysis that investigated the association between T1DM and leukemia risk. Finally, although with limited data, we also explored the effect of T2DM’s duration upon leukemia risk.

Limitations of our study should also be acknowledged. First, the association between T2DM and leukemia risk may be modified by confounders. We observed that the pooled RRs for studies that adjusted for alcohol use, smoking, and BMI/obesity was somewhat weaker than those that did not. Additionally, most of the included studies used registry-based data, and details on some important potential confounders, such as a family history of hematological malignancies, obesity status, career, and exposure to benzene and radioactive materials, were not available. Therefore, the role of these confounders in the association between T2DM and leukemia risk remains imprecise. Second, recent research has highlighted the inverse association of some antidiabetic drugs, such as metformin, with the risk of certain cancers [46-49]. Unfortunately, the role of these drugs in the relationship between DM and leukemia risk has rarely been elucidated in the included studies. Finally, different leukemia subtypes have varied epidemiological patterns and risk factors. Thus, in-depth research is essential to determine the differences in the impact of T2DM on the risk of different leukemia subtypes.

Conclusion

The results of this meta-analysis indicate that T2DM is associated with an elevated risk of leukemia in both men and women. Future studies should differentiate between the different leukemia subtypes and explore their risk in persons with T2DM. The role of potential confounders in the association between T2DM and leukemia risk, such as BMI, family history of leukemia and drug use, also requires more well-designed studies to clarify.

Funding Sources

This work was financially supported by the National Natural Science Foundation of China (grant number 81641123) and the Fundamental Research Funds for the Central Universities (grant number 2042017kf0193).

Ethical Approval

For this type of study (i.e. retrospective), formal consent is not required.

Research Involving Human Participants and/or Animals

This article does not contain any studies with human or animal subjects performed by any of the authors.

Acknowledgments

The authors would like to express special gratitude to all the personnel who supported or helped with this study.

References

• 1   Juliusson  G,  Hough  R (2016) Leukemia. Prog Tumor Res 43: 87–100.
• 2   Kampen  KR (2012) The discovery and early understanding of leukemia. Leuk Res 36: 6–13.
• 3   Bray  F,  Ferlay  J,  Soerjomataram  I,  Siegel  RL,  Torre  LA, et al. (2018) Global cancer statistics 2018: GLOBOCAN estimates of incidence and mortality worldwide for 36 cancers in 185 countries. CA Cancer J Clin 68: 394–424.
• 4   Zimmet  P,  Alberti  KG,  Magliano  DJ,  Bennett  PH (2016) Diabetes mellitus statistics on prevalence and mortality: facts and fallacies. Nat Rev Endocrinol 12: 616–622.
• 5  International Diabetes Federation (IDF) (2017) IDF Diabetes Atlas Eighth 12 Edition, https://diabetesatlas.org/resources/2017-atlas.html [Accessed 2020.09.02]
• 6   Wang  Y,  Liu  X,  Yan  P,  Bi  Y,  Liu  Y, et al. (2020) Association between type 1 and type 2 diabetes and risk of non-Hodgkin’s lymphoma: a meta-analysis of cohort studies. Diabetes Metab 46: 8–19.
• 7   Zhang  ZJ (2014) Re: “Type 2 diabetes and the risk of colorectal adenomas: Black Women’s Health Study”. Am J Epidemiol 179: 1276–1277.
• 8   Tsilidis  KK,  Kasimis  JC,  Lopez  DS,  Ntzani  EE,  Ioannidis  JP (2015) Type 2 diabetes and cancer: umbrella review of meta-analyses of observational studies. BMJ 350: g7607.
• 9   Xu  Y,  Huo  R,  Chen  X,  Yu  X (2017) Diabetes mellitus and the risk of bladder cancer: a PRISMA-compliant meta-analysis of cohort studies. Medicine (Baltimore) 96: e8588.
• 10   Castillo  JJ,  Mull  N,  Reagan  JL,  Nemr  S,  Mitri  J (2012) Increased incidence of non-Hodgkin lymphoma, leukemia, and myeloma in patients with diabetes mellitus type 2: a meta-analysis of observational studies. Blood 119: 4845–4850.
• 11   Zhang  PH,  Chen  ZW,  Lv  D,  Xu  YY,  Gu  WL, et al. (2012) Increased risk of cancer in patients with type 2 diabetes mellitus: a retrospective cohort study in China. BMC Public Health 12: 567.
• 12   Lai  GY,  Park  Y,  Hartge  P,  Hollenbeck  AR,  Freedman  ND (2013) The association between self-reported diabetes and cancer incidence in the NIH-AARP Diet and Health Study. J Clin Endocrinol Metab 98: E497–E502.
• 13   Lin  CC,  Chiang  JH,  Li  CI,  Liu  CS,  Lin  WY, et al. (2014) Cancer risks among patients with type 2 diabetes: a 10-year follow-up study of a nationwide population-based cohort in Taiwan. BMC Cancer 14: 381.
• 14   Harding  JL,  Shaw  JE,  Peeters  A,  Cartensen  B,  Magliano  DJ (2015) Cancer risk among people with type 1 and type 2 diabetes: disentangling true associations, detection bias, and reverse causation. Diabetes Care 38: 264–270.
• 15   Liu  X,  Hemminki  K,  Forsti  A,  Sundquist  K,  Sundquist  J, et al. (2015) Cancer risk in patients with type 2 diabetes mellitus and their relatives. Int J Cancer 137: 903–910.
• 16   Carstensen  B,  Read  SH,  Friis  S,  Sund  R,  Keskimaki  I, et al. (2016) Cancer incidence in persons with type 1 diabetes: a five-country study of 9,000 cancers in type 1 diabetic individuals. Diabetologia 59: 980–988.
• 17   Dankner  R,  Boffetta  P,  Balicer  RD,  Boker  LK,  Sadeh  M, et al. (2016) Time-dependent risk of cancer after a diabetes diagnosis in a cohort of 2.3 million adults. Am J Epidemiol 183: 1098–1106.
• 18   Gini  A,  Bidoli  E,  Zanier  L,  Clagnan  E,  Zanette  G, et al. (2016) Cancer among patients with type 2 diabetes mellitus: a population-based cohort study in northeastern Italy. Cancer Epidemiol 41: 80–87.
• 19   Williams  R,  van Staa  TP,  Gallagher  AM,  Hammad  T,  Leufkens  HGM, et al. (2018) Cancer recording in patients with and without type 2 diabetes in the Clinical Practice Research Datalink primary care data and linked hospital admission data: a cohort study. BMJ Open 8: e020827.
• 20   Qi  J,  He  P,  Yao  H,  Song  R,  Ma  C, et al. (2019) Cancer risk among patients with type 2 diabetes: a real-world study in Shanghai, China. J Diabetes 11: 878–883.
• 21   Saarela  K,  Tuomilehto  J,  Sund  R,  Keskimaki  I,  Hartikainen  S, et al. (2019) Cancer incidence among Finnish people with type 2 diabetes during 1989–2014. Eur J Epidemiol 34: 259–265.
• 22   Stang  A (2010) Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in meta-analyses. Eur J Epidemiol 25: 603–605.
• 23   DerSimonian  R,  Laird  N (1986) Meta-analysis in clinical trials. Control Clin Trials 7: 177–188.
• 24   Ragozzino  M,  Melton LJ  3rd,  Chu  CP,  Palumbo  PJ (1982) Subsequent cancer risk in the incidence cohort of Rochester, Minnesota, residents with diabetes mellitus. J Chronic Dis 35: 13–19.
• 25   Wideroff  L,  Gridley  G,  Mellemkjaer  L,  Chow  WH,  Linet  M, et al. (1997) Cancer incidence in a population-based cohort of patients hospitalized with diabetes mellitus in Denmark. J Natl Cancer Inst 89: 1360–1365.
• 26   Jee  SH,  Ohrr  H,  Sull  JW,  Yun  JE,  Ji  M, et al. (2005) Fasting serum glucose level and cancer risk in Korean men and women. JAMA 293: 194–202.
• 27   Swerdlow  AJ,  Laing  SP,  Qiao  Z,  Slater  SD,  Burden  AC, et al. (2005) Cancer incidence and mortality in patients with insulin-treated diabetes: a UK cohort study. Br J Cancer 92: 2070–2075.
• 28   Shu  X,  Ji  J,  Li  X,  Sundquist  J,  Sundquist  K, et al. (2010) Cancer risk among patients hospitalized for type 1 diabetes mellitus: a population-based cohort study in Sweden. Diabet Med 27: 791–797.
• 29   Atchison  EA,  Gridley  G,  Carreon  JD,  Leitzmann  MF,  McGlynn  KA (2011) Risk of cancer in a large cohort of U.S. veterans with diabetes. Int J Cancer 128: 635–643.
• 30   Wotton  CJ,  Yeates  DG,  Goldacre  MJ (2011) Cancer in patients admitted to hospital with diabetes mellitus aged 30 years and over: record linkage studies. Diabetologia 54: 527–534.
• 31   Okręglicka  K (2015) Health effects of changes in the structure of dietary macronutrients intake in western societies. Rocz Panstw Zakl Hig 66: 97–105.
• 32   Duan  W,  Shen  X,  Lei  J,  Xu  Q,  Yu  Y, et al. (2014) Hyperglycemia, a neglected factor during cancer progression. Biomed Res Int 2014: 461917.
• 33   Fabiani  R,  Minelli  L,  Bertarelli  G,  Bacci  S (2016) A western dietary pattern increases prostate cancer risk: a systematic review and meta-analysis. Nutrients 8: 626.
• 34   Liu  P,  Holman  CD,  Jin  J,  Zhang  M (2015) Diet and risk of adult leukemia: a multicenter case-control study in China. Cancer Causes Control 26: 1141–1151.
• 35   Solans  M,  Castelló  A,  Benavente  Y,  Marcos-Gragera  R,  Amiano  P, et al. (2018) Adherence to the Western, Prudent, and Mediterranean dietary patterns and chronic lymphocytic leukemia in the MCC-Spain study. Haematologica 103: 1881–1888.
• 36   Abar  L,  Sobiecki  JG,  Cariolou  M,  Nanu  N,  Vieira  AR, et al. (2019) Body size and obesity during adulthood, and risk of lympho-haematopoietic cancers: an update of the WCRF-AICR systematic review of published prospective studies. Ann Oncol 30: 528–541.
• 37   Ye  H,  Adane  B,  Khan  N,  Alexeev  E,  Nusbacher  N, et al. (2018) Subversion of systemic glucose metabolism as a mechanism to support the growth of leukemia cells. Cancer Cell 34: 659–673.e6.
• 38   Zhang  JJ,  Jia  JP,  Shao  Q,  Wang  YK (2019) Diabetes mellitus and risk of pancreatic cancer in China: a meta-analysis based on 26 case-control studies. Prim Care Diabetes 13: 276–282.
• 39   Feltbower  RG,  McKinney  PA,  Greaves  MF,  Parslow  RC,  Bodansky  HJ (2004) International parallels in leukaemia and diabetes epidemiology. Arch Dis Child 89: 54–56.
• 40   Feltbower  RG,  Manda  SO,  Gilthorpe  MS,  Greaves  MF,  Parslow  RC, et al. (2005) Detecting small-area similarities in the epidemiology of childhood acute lymphoblastic leukemia and diabetes mellitus, type 1: a Bayesian approach. Am J Epidemiol 161: 1168–1180.
• 41   Richiardi  L,  Magnani  C,  Bruno  G,  Maule  MM,  Merletti  F, et al. (2005) Re: “Detecting small-area similarities in the epidemiology of childhood acute lymphoblastic leukemia and diabetes mellitus, type 1: a Bayesian approach”. Am J Epidemiol 162: 1132–1133.
• 42   Crawley  DJ,  Holmberg  L,  Melvin  JC,  Loda  M,  Chowdhury  S, et al. (2014) Serum glucose and risk of cancer: a meta-analysis. BMC Cancer 14: 985.
• 43   Lee  SC,  Chan  JC (2015) Evidence for DNA damage as a biological link between diabetes and cancer. Chin Med J (Engl) 128: 1543–1548.
• 44   Fitter  S,  Vandyke  K,  Schultz  CG,  White  D,  Hughes  TP, et al. (2010) Plasma adiponectin levels are markedly elevated in imatinib-treated chronic myeloid leukemia (CML) patients: a mechanism for improved insulin sensitivity in type 2 diabetic CML patients? J Clin Endocrinol Metab 95: 3763–3767.
• 45   Agostino  NM,  Chinchilli  VM,  Lynch  CJ,  Koszyk-Szewczyk  A,  Gingrich  R, et al. (2011) Effect of the tyrosine kinase inhibitors (sunitinib, sorafenib, dasatinib, and imatinib) on blood glucose levels in diabetic and nondiabetic patients in general clinical practice. J Oncol Pharm Pract 17: 197–202.
• 46   Heckman-Stoddard  BM,  DeCensi  A,  Sahasrabuddhe  VV,  Ford  LG (2017) Repurposing metformin for the prevention of cancer and cancer recurrence. Diabetologia 60: 1639–1647.
• 47   Cunha Junior  AD,  Pericole  FV,  Carvalheira  JBC (2018) Metformin and blood cancers. Clinics (Sao Paulo) 73: e412s.
• 48   Wang  Y,  Fu  T,  Liu  Y,  Yang  G,  Yu  C, et al. (2020) The association between metformin and head and neck cancer survival: a systematic review and meta-analysis of cohort studies. Curr Pharm Des 26: 3161–3170.
• 49   Wang  Y,  Liu  X,  Yan  P,  Tang  J,  Chen  T, et al. (2020) Effect of metformin on the risk of prostate cancer in patients with type 2 diabetes by considering different confounding factors: a meta-analysis of observational studies. Eur J Cancer Prev 29: 42–52.