The association between gestational diabetes mellitus and cancer in women: A systematic review and meta-analysis of observational studies

Journal Pre-proof The association between gestational diabetes mellitus and cancer in women: A systematic review and meta-analysis of observational studies Yongbo Wang Pengfei Yan Tao Fu Jingping Yuan Guifang Yang Yu Liu Zhi-Jiang Zhang PhD

PII:

S1262-3636(20)30030-6

DOI:

https://doi.org/doi:10.1016/j.diabet.2020.02.003

Reference:

DIABET 101149

To appear in:

Diabetes & Metabolism

Received Date:

11 August 2019

Revised Date:

30 January 2020

Accepted Date:

9 February 2020

Please cite this article as: Wang Y, Yan P, Tao F, Yuan J, Yang G, Liu Y, Zhang Z-Jiang, The association between gestational diabetes mellitus and cancer in women: A systematic review and meta-analysis of observational studies, Diabetes and amp; Metabolism (2020), doi: https://doi.org/10.1016/j.diabet.2020.02.003

This is a PDF file of an article that has undergone enhancements after acceptance, such as the addition of a cover page and metadata, and formatting for readability, but it is not yet the definitive version of record. This version will undergo additional copyediting, typesetting and review before it is published in its final form, but we are providing this version to give early visibility of the article. Please note that, during the production process, errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain. © 2020 Published by Elsevier.

The association between gestational diabetes mellitus and cancer in women: A systematic review and meta-analysis of observational studies

Running title: Gestational diabetes mellitus and cancer risk Yongbo Wang1, Pengfei Yan1, Tao Fu2, Jingping Yuan2, Guifang Yang3, Yu Liu4, Zhi-Jiang Zhang1* 1

Department of Preventive Medicine, School of Health Sciences, Wuhan University, Wuhan

of

430071, China Renmin Hospital, Wuhan University, Wuhan 430071, China

3

Zhongnan Hospital, Wuhan University, Wuhan 430071, China

4

Department of Statistics, College of Management, Wuhan Institute of Technology, Wuhan

ro

2

Zhi-Jiang Zhang, PhD

re

* Correspondence to:

-p

430071, China

rn al P

Department of Preventive Medicine, School of Health Sciences, Wuhan University, No. 185 Donghu Road, Wuhan 430071, China Tel: + 86 6875 6848

Fax: + 86 6875 6848

Email: [email protected]

Jo u

Received 11 August 2020; Accepted 9 February 2020

Abstract

Aim. – Both type 1 and type 2 diabetes are associated with greater risk of a variety of cancers. However, the association between gestational diabetes mellitus (GDM) and risk of cancer has so far not been well addressed. This study aimed to summarize the epidemiological evidence of the association between GDM and subsequent risk of cancer. Methods. – PubMed and Embase databases were searched for relevant studies, and a random-effects model was used to calculate the summary relative risks (RRs) along with the corresponding 95% confidence intervals (CIs). Results. – A total of 17 observational studies were selected, comprising 7 case–control and 10 1

Page 1 of 28

cohort studies. Pooled effect estimates retrieved from these 17 studies showed that GDM was associated with an increased risk of breast cancer in Asia (pooled RR: 1.31, 95% CI: 1.01– 1.70), but not in other regions, and also with thyroid cancer (RR: 1.28, 95% CI: 1.16–1.42), stomach cancer (RR: 1.43, 95% CI: 1.02–2.00) and liver cancer (RR: 1.27, 95% CI: 1.03– 1.55). However, GDM was not associated with any increased risk of colon (RR: 1.41, 95% CI: 0.90–2.21), colorectal (RR: 1.16, 95% CI: 0.95–1.41), ovarian (RR: 1.14, 95% CI: 0.90–1.44), cervical (RR: 1.02, 95% CI: 0.81–1.29), pancreatic (RR: 3.49, 95% CI: 0.80–15.23), brain and nervous system (RR: 1.26, 95% CI: 0.80–1.97), blood (leukaemia, RR: 0.77, 95% CI: 0.45–1.30), endometrial (RR: 0.77, 95% CI: 0.20–2.98), skin (RR: 1.13, 95% CI: 0.81–1.59)

of

or urological (RR: 0.98, 95% CI: 0.73–1.31) cancers. Conclusion. – GDM is associated with a greater risk of cancer in women, including breast,

ro

thyroid, stomach and liver cancers. However, further investigation is nonetheless warranted.

-p

Keywords: Cancer; Epidemiological; Risk; Meta-analysis

re

Introduction

Gestational diabetes mellitus (GDM), considered the first diagnosis of glucose intolerance

rn al P

that arises during pregnancy, promotes the risk of maternal and infant pregnancy complications [1]. The main risk factors for GDM include polycystic ovary syndrome, maternal obesity or overweight, family history of type 2 diabetes mellitus (T2DM), prediabetes, previous history of fetal death and increased maternal age [2]. In addition, GDM is characterized by hyperglycaemia, insulin resistance and hyperinsulinaemia, all of which have been implicated in uncontrolled cell growth and cancer [1]. During pregnancy, women

Jo u

with GDM have higher levels of glucose and C-reactive protein, lower levels of sex hormone-binding globulin and a greater possibility of hyperinsulinaemia compared with pregnant women without GDM [3, 4]. Women are also considered to be at greatest risk of developing T2DM within 5 years of having had a pregnancy with GDM, with the incidence of T2DM plateauing after around 10 years [5, 6]. T2DM is associated with a higher risk of various cancers [7–10], and may be particularly detrimental in women [11, 12]. As GDM has characteristics similar to T2DM and is a predictor for subsequent significant T2DM, it is reasonable to infer that GDM may also be associated with greater cancer risk in women. Indeed, a number of studies [13–29] have been conducted to address whether GDM increases the risk of cancer. However, it remains unclear as to whether or not GDM is 2

Page 2 of 28

associated with an increased cancer risk, considering the mixed results in studies so far. Given that the global prevalence of GDM has been on the rise [30, 31], it is now of public-health significance to examine the impact of GDM on cancer incidence. In the present work, our aim was to conduct a meta-analysis to examine the association between GDM and cancer risk in women.

Materials and Methods Search strategy The present meta-analysis was performed according to the Preferred Reporting Items for

of

Systematic Reviews and Meta-Analyses (PRISMA) guidelines [32]. PubMed and Embase databases were searched up to 30 January 2020 for epidemiological studies of GDM and

ro

cancer risk, using the following search terms: (gestational diabetes OR gestational diabetes mellitus OR GDM) AND (cancer OR oncology OR tumour OR tumours OR neoplasm * OR

-p

carcinoma OR malignant *) AND (case–control OR retrospective OR cohort OR cohorts OR prospective OR longitudinal OR follow-up). The reference lists of previous reviews of the

re

subject and of studies included in the analysis were also searched for any further relevant

Selection criteria

rn al P

studies.

Two reviewers (Y.W. and P.Y.) independently assessed the eligibility of each article, and any disagreements regarding study inclusion were resolved by consultation with a third reviewer (Z.-J.Z.). Studies that met the following criteria were included: (i) case–control or cohort design, published as full-text manuscripts; (ii) main focus of interest was presence of

Jo u

preexisting GDM and main outcome of interest was incidence of cancer; (iii) sample size with cancer was available; and (iv) effect sizes (ES) [including relative risk (RR), odds ratio (OR), hazard ratio (HR), standardized incidence rate (SIR) and incidence risk ratio (IRR)]

with corresponding 95% confidence intervals (CIs) were reported, but only those ES with 95% CIs from models reflecting the maximum extent of adjustment were included. When data from the same population were reported in more than one study, the publication containing the larger sample size with more complete and sufficient information was selected.

Data extraction and quality assessment Two reviewers (Y.W. and P.Y.) independently extracted all data using a standardized form, 3

Page 3 of 28

with any disagreements regarding study inclusion resolved by consultation with a third reviewer (Z.-J.Z.). These data included the first author, publication year, case source, control source, study location, years of follow-up, maternal age at pregnancy, number of GDM patients and cancer patients, and criteria used to evaluate GDM and cancer. Also collected were data on adjusted ES and 95% CIs, and the potential confounding factors adjusted for. According to the Newcastle–Ottawa Scale (NOS) guidelines for assessing the quality of non-randomized studies in meta-analyses [33], nine domains were established for quality assessment: these were related to selection (four items), comparability (two items) and outcome (three items). A study awarded with ≥ 5 stars was defined as a high-quality study,

of

with 0–4 stars considered low quality.

ro

Statistical analysis

A random-effects model [34] was used to calculate summary ES and 95% CIs for the

-p

association between GDM and cancer risk. As cancer is a relatively rare disease, differences between the above-mentioned risk estimates were ignored, and all measures were interpreted

re

as RR for simplicity. Also, I2 statistics were used to assess statistical heterogeneity across studies, using the following cut-off points: low (I2 < 25%); moderate (I2 25–50%); high (I2 >

rn al P

50–75%); and severe (I2 > 75%) [35]. In addition, to assess whether the results could have been markedly affected by a single study, sensitivity analyses were conducted in which one study at a time was removed and the rest analyzed. Any potential publication bias was assessed by Egger’s linear regression test and Begg’s adjusted rank correlation test [36, 37]. A P value < 0.10 was considered to indicate statistically significant publication bias. All statistical analyses were performed using Stata version 14.0 software (StataCorp LLC,

Jo u

College Station, TX, USA).

Results

Literature search

Our literature search yielded 1327 articles. However, most of the studies did not report the risk of GDM on risk of cancer or lacked original data and were therefore excluded (Fig. 1). Other articles were excluded for being commentaries, case reports or reviews. Of the initial abstracts reviewed, 42 studies were deemed eligible for inclusion and were reviewed in detail. Among these, 17 articles [13–29] were finally identified for inclusion. Descriptive data of the studies included in our meta-analysis are presented in Table I and Table II. 4

Page 4 of 28

Description of included studies The 17 articles included in our meta-analysis were rather heterogeneous: seven were case– control studies [15, 20, 21, 23, 24, 27, 28] and 10 were cohort studies [13, 14, 16–19, 22, 25, 26, 29]. Most (n = 12) of the selected studies focused on a single cancer site, although five studies [13, 14, 18, 22, 29] evaluated the relationship between GDM and cancer risk in a group of cancer patients with multiple types of cancer. The included studies were carried out in the following regions: the US (n = 8); Europe (n = 1); Asia (n = 6); and Canada (n = 2). Sample sizes ranged from 181 to 185,315, and the mean/median maternal age of the subjects at the time of pregnancy ranged from 24.0 to 33.1 years.

of

Methods used for determining GDM status were mixed across these studies: seven studies [16, 17, 21, 23, 24, 27, 28] were self-reported; and 10 studies [13–15, 18–20, 22, 25, 26, 29]

ro

used medical records of oral glucose tolerance tests (OGTTs) or glucose challenge tests (GCTs). To ascertain outcomes, six studies used cancer registries [19, 20, 22, 24–26], nine

-p

used medical records [13–16, 18, 21, 23, 27–29] and one was self-reported [17]. Seven studies [14–17, 22, 24, 28] were adjusted for body mass index (BMI), three [16, 20, 28] were

re

adjusted for alcohol intake, three [14, 20, 22] were adjusted for smoking, and only one [17] was adjusted for physical activity. In five of the cohort studies [13, 14, 17, 19, 22], the

rn al P

mean/median follow-up durations were < 10 years whereas, in the other five cohort studies [16, 18, 25, 26, 29], the mean/median follow-up time was ≥ 10 years. Table S1 (see supplementary materials associated with this article online) shows the risk of bias for the cohort and case–control studies. Quality scores ranged from 6 to 9, with an average score of 7.4 for cohort studies and 6.4 for case–control studies.

Jo u

Quantitative findings Breast cancer

As shown in Fig. 2, 15 studies [13, 14, 16–25, 27–29] investigated breast cancer risk. The results revealed that GDM was not associated with risk of breast cancer incidence in either case–control (OR: 0.88, 95% CI: 0.69–1.12, I2: 19.3%) or cohort (RR: 1.08, 95% CI: 0.89– 1.32, I2: 88.6%) studies. Sensitivity analyses, conducted by leaving out one study at a time, showed good consistency for case–control studies, with results ranging from 0.84 (95% CI: 0.68–1.05) to 0.95 (95% CI: 0.74–1.23). As for cohort studies, the results, which were not substantially altered upon removing any one study, were within the fairly narrow range of 1.00 (95% CI: 0.85–1.17) to 1.14 (95% CI: 0.95–1.39). 5

Page 5 of 28

Given the high heterogeneity of the cohort studies, subgroup analyses were also performed. However, summary associations did not vary, whether by GDM ascertainment methods, breast cancer ascertainment methods, sample size or follow-up in the cohort studies (Table III). In a subgroup analysis by region, for the five studies [13, 14, 18, 22, 25] carried out in Asia, the summary RR was 1.31 (95% CI: 1.01–1.70; P = 0.043) and, for the four studies [16, 17, 19, 29] carried out in North America, the summary RR was 0.88 (95% CI: 0.75–1.04; P = 0.121). A subsequent analysis stratified by source control could find no significant association between GDM and risk of breast cancer incidence in population-based studies [14, 16–19, 22, 25, 29].

of

Egger’s tests (P = 0.995 for case–control studies, P = 0.698 for cohort studies) and Begg’s tests (P = 0.927 for case–control studies, P = 0.466 for cohort studies) found no publication

ro

biases.

-p

Thyroid cancer

Four cohort studies [13, 14, 19, 29] reported a significant positive association, whereas one

re

cohort study [22] reported no significant association between GDM and thyroid cancer risk. The summary RR was 1.28 (95% CI: 1.16–1.42; P = 0.001; Fig. 3A), and there was no

Gastric cancer

rn al P

evidence of heterogeneity (I2: 0.0%).

Three cohort studies [13, 14, 22] found a positive, albeit non-significant, association between GDM and stomach cancer risk. The summary RR was 1.43 (95% CI: 1.02–2.00; P = 0.039;

Jo u

Fig. 3B), with no evidence of heterogeneity (I2: 0.0%).

Liver cancer

One cohort study [13] reported a positive but non-significant association between GDM and risk of liver cancer, whereas another cohort study [14] found no association. The summary RR was 1.27 (95% CI: 1.03–1.55; P = 0.022; Fig. 3C), and there was no evidence of heterogeneity (I2: 0.0%).

Colon cancer Two cohort studies [14, 22] showed no association between GDM and colon cancer risk—the summary RR was 1.41 (95% CI: 0.90–2.21; P < 0.136; Fig. 3D)—and there was no evidence 6

Page 6 of 28

of heterogeneity (I2: 0.0%). One cohort study [22] found no association between GDM and rectal cancer risk (RR: 1.25, 95% CI: 0.28–5.58), and two cohort studies [13, 29] could find no association between GDM and colorectal cancer risk (summary RR: 1.16, 95% CI: 0.95– 1.41; P = 0.142; Fig. 3E). There was also no evidence of heterogeneity (I2: 0.0%).

Ovarian cancer One cohort study [18] found a positive association, and two cohort studies [14, 29] found a positive but non-significant association, between GDM and ovarian cancer risk, whereas one cohort study [13] reported no association. The summary RR was 1.14 (95% CI: 0.90–1.44; P

of

= 0.281; Fig. 3F), and there was evidence of moderate heterogeneity (I2: 31.8%).

ro

Cervical cancer

Two cohort studies [18, 29] found a positive, albeit non-significant, association between

-p

GDM and cervical cancer risk, and one cohort study [13] reported no association. The

Pancreatic cancer

rn al P

of heterogeneity (I2: 0.0%).

re

summary RR was 1.02 (95% CI: 0.81–1.29; P = 0.843; Fig. 3G), and there was no evidence

Two cohort studies [22, 26] found a significant positive association between GDM and pancreatic cancer risk, whereas another cohort study [13] did not. The summary RR was 3.49 (95% CI: 0.80–15.23; P = 0.097; Fig. 3H), with evidence of severe heterogeneity (I2: 87.6%).

Brain and nervous system tumours

Jo u

Two cohort studies [13, 22] showed no association between GDM and brain and nervous system cancer risk. The summary RR was 1.26 (95% CI: 0.80–1.97; P = 0.313; Fig. 3I), and there was no evidence of heterogeneity (I2: 0.0%).

Blood cancer Three cohort studies [14, 22, 29] found a positive but non-significant association between GDM and haematological neoplasm risk: the summary RR was 1.48 (95% CI: 1.04–2.09; P = 0.027; Fig. 3J), and there was no evidence of heterogeneity (I2: 0.0%). However, two cohort studies [13, 29] showed no association between GDM and leukaemia risk: the summary RR was 0.77 (95% CI: 0.45–1.30; P = 0.324; Fig. 3K); and there was no evidence of 7

Page 7 of 28

heterogeneity (I2: 0.0%).

Endometrial cancer One cohort study [15] found a positive but non-significant association between GDM and endometrial cancer risk, while another cohort study [29] reported no such association. The summary RR was 0.77 (95% CI: 0.20–2.98; P = 0.705; Fig. 3L). There was also evidence of high heterogeneity (I2: 68.5%).

Skin cancer

of

Two cohort studies [13, 22] found a positive, albeit non-significant, association between GDM and skin cancer risk, whereas one cohort study [29] reported no such association. The

ro

summary RR was 1.13 (95% CI: 0.81–1.59; P = 0.468; Fig. 3M). There was also evidence of

-p

moderate heterogeneity (I2: 36.4%).

Urological cancer

re

One cohort study [13] found a positive but non-significant association between GDM and urological cancer risk, whereas two other cohort studies [22, 29] reported no such association.

rn al P

The summary RR was 0.98 (95% CI: 0.73–1.31; P = 0.885; Fig. 3N) with no evidence of heterogeneity (I2: 0.0%).

Other types of cancer

Although there were insufficient studies identified of other cancer types to warrant meta-analyses, their findings are nevertheless presented in Table IV. One cohort study [13]

Jo u

found significant positive associations between GDM and risk of kidney and lung and bronchial cancers. Another cohort study [18] showed a significant positive association between GDM and uterine cancer risk.

Discussion The aim of the present study was to determine whether women with GDM are at higher risk of cancer incidence. In fact, according to 17 observational studies, women with GDM are indeed susceptible to thyroid, stomach and liver cancers. However, GDM was not associated with any greater risk of other cancers, including those of the colon, colorectum, ovary, uterine cervix, pancreas, brain and nervous system, and blood (leukaemia), although a major 8

Page 8 of 28

limitation is the sparse data for certain types of cancer, such as of the kidney, lung and bronchus, uterus and lymphatic system. In fact, given that cancer develops more often with increasing age (especially in the ovary), women with GDM might have been more likely to be diagnosed with malignancy if the follow-up periods had been longer. Therefore, it could be that the follow-up duration from the time of pregnancy (mean/median maternal age of women at the time of pregnancy ranged from 24.0 to 33.1 years) was insufficient. Nevertheless, our present results indicate that GDM is indeed correlated with an increased risk of cancer. In the subgroup analyses for breast cancer, while the summary ES for studies conducted in

of

North America were not statistically significant, a positive association was found in cohort studies carried out in Asia. However, it is unclear whether this positive association plays a

ro

role in the greater susceptibility to insulin resistance observed in Asian populations. Disorders associated with insulin resistance include the metabolic syndrome and obesity [38], and the

-p

latter, especially central obesity [39, 40], has been associated with a higher risk of breast cancer. Given the rapidly increasing incidence of central obesity and GDM in the Asian

re

population, further studies of the GDM–breast cancer association appear to be indicated [41, 42].

rn al P

In the present meta-analysis, it was found that GDM correlated with a 28% increased risk of thyroid cancer incidence. Intriguingly, a previous meta-analysis [43] found that women with T2DM have a significantly greater risk of thyroid cancer compared with those without T2DM. In addition, higher levels of fasting blood glucose are associated with higher risk of thyroid cancer [44], while preclinical studies have shown that hyperinsulinaemia has a direct carcinogenic effect on thyroid cancer cells [44]. Moreover, women with diabetes during

Jo u

pregnancy are at higher risk of developing thyroid cancer, which may be partly related to more monitoring by doctors after delivery. This observation is itself consistent with the fact that diabetes patients are more likely to be diagnosed with early-stage thyroid cancer due to more healthcare exposure.

In the present study, it was found that women with GDM have a 43% higher risk of stomach cancer incidence. Common risk factors for gastric cancer include tobacco use (smoking), radiation, Helicobacter pylori infection, family history, obesity, and dietary factors such as high intakes of salty and smoked food, and low consumption of vegetables and fruits [45]. Among these factors, smoking, dietary habits and obesity also characterize GDM [46]. In addition, GDM is correlated with a greater risk of developing T2DM [47]. Moreover, it has 9

Page 9 of 28

been shown that patients who develop T2DM have a higher risk of developing stomach cancer [48, 49]. Our present study also found that women with GDM have a 27% increased risk of developing liver cancer. The prevalences of both GDM and liver cancer have been on the rise worldwide [30, 50], although whether this correlation is causal has yet to be determined. In fact, GDM and liver cancer have some risk factors in common, such as smoking, alcohol use, dietary habits and obesity [51, 52]. However, it remains uncertain whether manipulation of these risk factors would modulate the association. In addition, a positive but non-significant association was found between GDM and pancreatic cancer risk. This higher risk of liver and pancreatic

of

cancer in GDM patients may be related to the fact that these organs are particularly exposed to insulin, especially if tumour cells remain sensitive to insulin [53].

ro

On the other hand, our study found no significant association between GDM and the risk of cancers of the colon, colorectum, ovary, uterine cervix, brain and nervous system, and blood

-p

(leukaemia). However, those studies included fewer cases of cancer, which may have markedly reduced their statistical power. When exploring the relationship between GDM and

re

the relatively rarer cancers, the small number of cases may have led to misunderstandings of the correlations between them. Therefore, such results should be interpreted with caution.

durations.

rn al P

Future studies could be improved by recruiting larger cohorts and/or longer follow-up

Nevertheless, in general, clarification of the association between GDM and subsequent cancer risk may be difficult. There is evidence that GDM can affect cancer risk via certain direct or indirect pathophysiological mechanisms, such as hyperinsulinaemia secondary to insulin resistance, hyperglycaemia and chronic inflammation. Common risk factors for cancer

Jo u

and GDM, such as BMI, smoking habits, alcohol use and low physical activity, may be further explanations [54]. In addition, hyperglycaemia can trigger the production of reactive oxygen species, which can induce carcinogenesis by destroying cellular DNA [55, 56]. In recent years, the biological role of systemic inflammation in the pathogenesis of GDM has been a topic of increasing interest. Independently of obesity, enhanced circulating levels of C-reactive and interleukin-6 proteins have been observed in GDM [57, 58], indicating that GDM is a low-inflammation state. Inflammation, which is also a hallmark of cancer, has generally been recognized to mediate all aspects of cancer, from cell transformation to metastasis. Thus, systemic and chronic inflammation may be the underlying mechanism linking GDM 10

Page 10 of 28

and cancer development. The risk of diabetes, and T2DM in particular, is especially high in women with a history of GDM, and diabetes has been consistently reported to increase the risk of cancer. Indeed, the higher risk of diabetes after GDM may be one of the reasons for the greater risk of cancer in such a population. The results of our present study extend those of the recently published (2014) meta-analysis by Tong et al. [59] of the association between GDM and subsequent risk of cancer. Nine papers were included in that study [59], whereas 17 papers were included in ours. In addition, the range of effects revealed in our present analysis is also considerably greater than that of a previously published meta-analysis [60] of the association between GDM and breast cancer

of

risk, as our analysis looked at the relationship between GDM and breast cancer risk by controlling for various confounding factors.

ro

Nevertheless, our present study has some limitations. First, most of the included studies lacked any information on multiple GDM pregnancies and histology, parity, and stage and

-p

subtype of cancer and, therefore, it was not possible to consider those variables. However, the Fuchs et al. study [18] found no association between the number of times that GDM was

re

diagnosed and the RR for malignancy. Second, the shared common risk factors for GDM and cancer, such as obesity, alcohol intake and smoking, were not considered in our meta-analysis,

rn al P

as only a few studies adjusted for those potential confounders; it is also not clear whether the reported high cancer risk of GDM was due to the high prevalence of these factors. Third, given the small number of studies reporting site-specific cancers, it is difficult to report the overall cancer risk subsequent to GDM while also checking all major contributors to the high overall risk of cancer at all sites. Fourth, most studies gave no information on potential confounders, such as physical activity, dietary habits, genetic factors and environmental

Jo u

exposures. Fifth, in some studies, the follow-up period may not have been long enough to allow cancer detection. Indeed, the lengthy time lag and low prevalence of cancer make it difficult to conduct prospective studies. Sixth, providers of antenatal, gynaecological and family-planning services do not always use standardized criteria, nor do they necessarily record data as completely and accurately as possible. Furthermore, differences in regions, races or ethnicities and levels of medical service utilization can also influence results. Finally, there may be different degrees of use of cancer screening services between patients with and without GDM.

11

Page 11 of 28

Conclusion The present meta-analysis has indicated that women who develop GDM have an increased risk of cancer, including those of the stomach, liver and thyroid. Thus, the prevention of GDM may play an important role in inhibiting the development of certain types of cancer in future. Also, as the incidence of GDM rises, more studies are needed to further establish the association between GDM and long-term risk for cancer. Authors’ contributions Y.W. and Z.-J.Z. reviewed the literature, screened medical records, assessed the quality of

of

studies, extracted data, performed the statistical analysis and wrote the manuscript. P.Y. screened medical records, assessed the quality of studies, extracted data, performed the

ro

statistical analysis and reviewed/edited the manuscript. T.F., J.Y., G.Y. and Y.L. contributed to

-p

the discussion and reviewed/edited the manuscript. All authors approved the final manuscript. Funding sources

re

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

Ethics approval

rn al P

(grant number 2042017kf0193) to Dr Zhi-Jiang Zhang.

Formal consent is not required for this type of study (retrospective).

Research involving human participants and/or animals

Jo u

This article included no studies involving humans or animals performed by any of the authors.

Conflicts of interest

The authors report no conflicts of interest in this work.

12

Page 12 of 28

References [1]

Metzger BE, Gabbe SG, Persson B, Buchanan TA, Catalano PA, Damm P, et al. International association of diabetes and pregnancy study groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy. Diabetes Care 2010;33:676-82.

[2]

Chen P, Wang S, Ji J, Ge A, Chen C, Zhu Y, et al. Risk Factors and Management of Gestational Diabetes. Cell Biochem Biophys 2015;71:689-94.

[3]

Wolf M, Sauk J, Shah A, Vossen Smirnakis K, Jimenez-Kimble R, Ecker JL, et al. Inflammation and glucose intolerance: a prospective study of gestational diabetes mellitus. Diabetes Care 2004;27:21-7.

[4]

Thadhani R, Wolf M, Hsu-Blatman K, Sandler L, Nathan D, Ecker JL. First-trimester sex

of

hormone binding globulin and subsequent gestational diabetes mellitus. Am J Obstet Gynecol 2003;189:171-6.

Kim C, Newton KM, Knopp RH. Gestational diabetes and the incidence of type 2 diabetes - A systematic review. Diabetes care 2002;25:1862-8.

[6]

ro

[5]

Chodick G, Elchalal U, Sella T, Heymann AD, Porath A, Kokia E, et al. The risk of overt

-p

diabetes mellitus among women with gestational diabetes: a population-based study. Diabetic Med 2010;27:779-85.

Wang Y, Liu X, Yan P, Bi Y, Liu Y, Zhang ZJ. Association between type 1 and type 2

re

[7]

diabetes and risk of non-Hodgkin's lymphoma: A meta-analysis of cohort studies. [8]

rn al P

Diabetes Metab 2019; pii: S1262-3636(19)30066-7.

Wei M, Liu Y, Wang C, Yu C, Li D, Zhou W, et al. The need for differentiating diabetes-specific mortality from total mortality when comparing metformin with insulin regarding cancer survival. Acta Diabetol 2017;54:219-20.

[9]

Zhang ZJ. Re: "Type 2 diabetes and the risk of colorectal adenomas: Black Women's Health Study". Am J Epidemiol 2014;179:1276-7.

[10]

Zhang ZJ. The association between type 2 diabetes and head and neck squamous cell

[11]

Jo u

carcinoma. Cancer Epidemiol 2014;38(2):209. Zhang ZJ, Zhao G, Song Y. Attention is needed for women in the control of diabetic complications in China. Acta Diabetol 2014;51:1081-2.

[12]

Zhang ZJ, Zhao G, Yu C, Bi Y, Zhang Q, Song Y. Diabetic women suffer more years of life lost than diabetic men. Int J Endocrinol 2014;2014:208369.

[13]

Peng YS, Lin JR, Cheng BH, Ho C, Lin YH, Shen CH, et al. Incidence and relative risk for developing cancers in women with gestational diabetes mellitus: a nationwide cohort study in Taiwan. BMJ open 2019;9:e024583.

[14]

Han KT, Cho GJ, Kim EH. Evaluation of the Association between Gestational Diabetes Mellitus at First Pregnancy and Cancer within 10 Years Postpartum Using National Health Insurance Data in South Korea. Int J Environ Res Public Health 2018;15(12).

[15]

Wartko PD, Beck TL, Reed SD, Mueller BA, Hawes SE. Association of endometrial 13

Page 13 of 28

hyperplasia and cancer with a history of gestational diabetes. Cancer Cause Control 2017;28:819-28. [16]

Powe CE, Tobias DK, Michels KB, Chen WY, Eliassen AH, Manson JE, et al. History of Gestational Diabetes Mellitus and Risk of Incident Invasive Breast Cancer among Parous Women in the Nurses' Health Study II Prospective Cohort. Cancer Epidem Biomar 2017;26(3):321-7.

[17]

Park YM, O'Brien KM, Zhao S, Weinberg CR, Baird DD, Sandler DP. Gestational diabetes mellitus may be associated with increased risk of breast cancer. British J Cancer 2017;116:960-3.

[18]

Fuchs O, Sheiner E, Meirovitz M, Davidson E, Sergienko R, Kessous R. The association between a history of gestational diabetes mellitus and future risk for female malignancies. Arch

[19]

of

Gynecol Obstetr 2017;295:731-6.

Bejaimal SA, Wu CF, Lowe J, Feig DS, Shah BR, Lipscombe LL. Short-term risk of cancer

ro

among women with previous gestational diabetes: a population-based study. Diabetic Med 2016;33:39-46. [20]

Ardalan A, Bungum T. Gestational Age and the Risk of Maternal Breast Cancer: A

[21]

-p

Population-Based Case-Control Study. Breast J 2016;22:657-61.

Brasky TM, Li YL, Jaworowicz DJ, Potischman N, Ambrosone CB, Hutson AD, et al. 2013;24:1675-85.

Sella T, Chodick G, Barchana M, Heymann AD, Porath A, Kokia E, et al. Gestational diabetes

rn al P

[22]

re

Pregnancy-related characteristics and breast cancer risk. Cancer Cause Control

and risk of incident primary cancer: a large historical cohort study in Israel. Cancer Causes Control : CCC 2011;22(11):1513-20. [23]

Sanderson M, Peltz G, Perez A, Johnson M, Vernon SW, Fernandez ME, et al. Diabetes, physical activity and breast cancer among Hispanic women. Cancer Epidemiol 2010;34:556-61.

[24]

Rollison DE, Giuliano AR, Sellers TA, Laronga C, Sweeney C, Risendal B, et al.

Jo u

Population-based case-control study of diabetes and breast cancer risk in Hispanic and non-Hispanic White women living in US southwestern states. AM J Epidemiol 2008;167:447-56.

[25]

Perrin MC, Terry MB, Kleinhaus K, Deutsch L, Yanetz R, Tiram E, et al. Gestational diabetes and the risk of breast cancer among women in the Jerusalem Perinatal Study. Breast Cancer Res Treat 2008;108:129-35.

[26]

Perrin MC, Terry MB, Kleinhaus K, Deutsch L, Yanetz R, Tiram E, et al. Gestational diabetes as a risk factor for pancreatic cancer: a prospective cohort study. BMC Med 2007;5:25.

[27]

Lawlor DA, Smith GD, Ebrahim S. Hyperinsulinaemia and increased risk of breast cancer: findings from the British Women's Heart and Health Study. Cancer Causes Control 2004;15:267-75.

[28]

Troisi R, Weiss HA, Hoover RN, Potischman N, Swanson CA, Brogan DR, et al. Pregnancy 14

Page 14 of 28

characteristics and maternal risk of breast cancer. Epidemiology 1998;9:641-7. [29]

Pace R, Rahme E, Dasgupta K. Gestational diabetes mellitus and risk of incident primary cancer: A population-based retrospective cohort study. J Diabetes 2020;12:87-90.

[30]

Ferrara A. Increasing prevalence of gestational diabetes mellitus: a public health perspective. Diabetes care 2007;30 Suppl 2:S141-6.

[31]

Cho GJ, Park JH, Lee H, Yoo S, Shin SA, Oh MJ. Prepregnancy Factors as Determinants of the Development of Diabetes Mellitus After First Pregnancy. J Clin Endocr Metab 2016;101:2923-30.

[32]

Moher D, Liberati A, Tetzlaff J, Altman DG, Group P. Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement. J Clin Epidemiol 2009;62:1006-12.

[33]

Stang A. Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of

of

nonrandomized studies in meta-analyses. Eur J Epidemiol 2010;25:603-5.

DerSimonian R, Laird N. Meta-analysis in clinical trials. Control Clin Trials 1986;7:177-88.

[35]

Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsistency in meta-analyses.

ro

[34]

BMJ 2003;327:557-60. [36]

Egger M, Davey Smith G, Schneider M, Minder C. Bias in meta-analysis detected by a simple, Begg CB, Mazumdar M. Operating characteristics of a rank correlation test for publication bias. Biometrics 1994;50:1088-101.

[38]

Lorincz AM, Sukumar S. Molecular links between obesity and breast cancer. Endocr-Relat

rn al P

Cancer 2006;13:279-92. [39]

re

[37]

-p

graphical test. BMJ 1997;315:629-34.

Connolly BS, Barnett C, Vogt KN, Li T, Stone J, Boyd NF. A meta-analysis of published literature on waist-to-hip ratio and risk of breast cancer. Nutr Cancer 2002;44:127-38.

[40]

Eliassen AH, Colditz GA, Rosner B, Willett WC, Hankinson SE. Adult weight change and risk of postmenopausal breast cancer. Jama-J Am Med Assoc 2006;296:193-201.

[41]

Lee KW, Ching SM, Ramachandran V, Yee A, Hoo FK, Chia YC, et al. Prevalence and risk factors of gestational diabetes mellitus in Asia: a systematic review and meta-analysis. BMC

[42]

Jo u

Pregnancy Childbirth 2018;18:494.

Erber E, Lim U, Maskarinec G, Kolonel LN. Common immune-related risk factors and incident non-Hodgkin lymphoma: The multiethnic cohort. Int J Cancer 2009;125:1440-5.

[43]

Li H, Qian J. Association of diabetes mellitus with thyroid cancer risk: A meta-analysis of cohort studies. Medicine 2017;96:e8230.

[44]

Shih SR, Chiu WY, Chang TC, Tseng CH. Diabetes and thyroid cancer risk: literature review. Exp Diabetes Res 2012;2012:578285.

[45]

Karimi P, Islami F, Anandasabapathy S, Freedman ND, Kamangar F. Gastric cancer: descriptive epidemiology, risk factors, screening, and prevention. Cancer Epidemiol Biomarkers Prev 2014;23:700-13.

[46]

Weinert LS. International Association of Diabetes and Pregnancy Study Groups recommendations on the diagnosis and classification of hyperglycemia in pregnancy: 15

Page 15 of 28

comment to the International Association of Diabetes and Pregnancy Study Groups Consensus Panel. Diabetes Care 2010;33:e97; author reply e8. [47]

Savyon M. The risk of overt diabetes mellitus among women with gestational diabetes: a population-based study. Diabetic Med 2010;27:852.

[48]

Ohkuma T, Peters SAE, Woodward M. Sex differences in the association between diabetes and cancer: a systematic review and meta-analysis of 121 cohorts including 20 million individuals and one million events. Diabetologia 2018;61:2140-54.

[49]

Fang HJ, Shan SB, Zhou YH, Zhong LY. Diabetes mellitus and the risk of gastrointestinal cancer in women compared with men: a meta-analysis of cohort studies. BMC Cancer 2018;18:422.

[50]

Akinyemiju T, Abera S, Ahmed M, Alam N, Alemayohu MA, Allen C, et al. The Burden of

of

Primary Liver Cancer and Underlying Etiologies From 1990 to 2015 at the Global, Regional, and National Level: Results From the Global Burden of Disease Study 2015. JAMA oncology [51]

ro

2017;3:1683-91.

Solomon CG, Willett WC, Carey VJ, RichEdwards J, Hunter DJ, Colditz GA, et al. A prospective study of pregravid determinants of gestational diabetes mellitus. JAMA

[52]

-p

1997;278:1078-83.

Marengo A, Rosso C, Bugianesi E. Liver Cancer: Connections with Obesity, Fatty Liver, and

[53]

re

Cirrhosis. Annual review of medicine 2016;67:103-17.

Giovannucci E, Harlan DM, Archer MC, Bergenstal RM, Gapstur SM, Habel LA, et al. Diabetes

[54]

rn al P

and Cancer A consensus report. Diabetes Care 2010;33:1674-85. Vigneri P, Frasca F, Sciacca L, Pandini G, Vigneri R. Diabetes and cancer. Endocrine-Related Cancer 2009;16:1103-23. [55]

Chandrasekaran K, Swaminathan K, Chatterjee S, Dey A. Apoptosis in HepG2 cells exposed to high glucose. Toxicol In Vitro 2010;24:387-96.

[56]

Zhang Y, Zhou JQ, Wang TL, Cai L. High level glucose increases mutagenesis in human lymphoblastoid cells. Int J Biol Sci 2007;3:375-9. Ategbo JM, Grissa O, Yessoufou A, Hichami A, Dramane KL, Moutairou K, et al. Modulation of

Jo u

[57]

adipokines and cytokines in gestational diabetes and macrosomia. J Clin Endocr Metab 2006;91:4137-43.

[58]

Wolf M, Sandler L, Hsu K, Vossen-Smirnakis K, Ecker JL, Thadhani R. First-trimester C-reactive protein and subsequent gestational diabetes. Diabetes Care 2003;26:819-24.

[59]

Tong GX, Cheng J, Chai J, Geng QQ, Chen PL, Shen XR, et al. Association Between Gestational Diabetes Mellitus and Subsequent Risk of Cancer: a Systematic Review of Epidemiological Studies. Asian Pac J Cancer P 2014;15:4265-9.

[60]

Xie C, Wang W, Li X, Shao N, Li W. Gestational diabetes mellitus and maternal breast cancer risk: a meta-analysis of the literature. J Maternal Fetal Neonatal Med 2019;32:1022-32.

16

Page 16 of 28

Figure legends

Fig. 1. Flow diagram as per Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines for identifying relevant articles for the meta-analysis.

Fig. 2. Association in women between gestational diabetes mellitus (GDM) and breast cancer incidence. Data presented are relative risk (RR) for each study (grey boxes), 95% confidence intervals (CIs; horizontal lines) and summary RR with 95% CI (diamond-shaped boxes).

of

Fig. 3. Meta-analyses of pooled risks of cancer of the thyroid (A), stomach (B), liver (C), colon (D), colorectum (E), ovary (F), uterine cervix (G), pancreas (H), brain (I), blood (J),

ro

leukaemia (K), endometrium (L), skin (M) and urological organs (N) in women with gestational diabetes mellitus (GDM). Data presented are relative risk (RR) for each study

-p

(grey boxes), 95% confidence intervals (CIs; horizontal lines) and summary RR with 95% CI

Jo u

rn al P

re

(diamond-shaped boxes).

17

Page 17 of 28

Identification

Records excluded (n =1107)

Titles and abstracts screened after duplicates removed (n =220)

of

Screening

1.Records identified through database searching (n =1310) PubMed (470) Embase (840) 2.Records identified through reference list(n=17)

-p re

Studies included in qualitative synthesis (n=17)

Records excluded (n=25), With reasons: Did not report measure of association for cancer (11); Reviews (2) Reported type 1 diabetes and type 2 diabetes (8); Without GDM status well defined (4).

rn al P

Included

Eligibility

Full-text articles assessed for eligibility (n =42)

ro

Records excluded (n =178)

Studies included in quantitative synthesis (n=17)

Jo u

Fig. 1. PRISMA flow diagram for identification of relevant articles for the meta-analysis.

18

Page 18 of 28

of ro -p re rn al P

Jo u

Fig. 2. The association between GDM and breast cancer incidence in women. Data are presented as RR for each study (boxes), 95% CIs (horizontal lines) and summary as RR with 95% CI (diamond). Abbreviation: DM: diabetes mellitus; RR: relative risk; CI: confidence interval.

19

Page 19 of 28

of ro -p re rn al P Jo u Fig. 3. Meta-analysis of the pooled risk of thyroid (A), gastric (B), liver(C), colon (D), colorectal (E), ovary(F), cervical (G), pancreatic (H), brain (I), blood (J), leukemia (K), endometrial (L), skin (M), and urological cancer (N) in women with GDM.

20

Page 20 of 28

Brasky, 2013 [21]

US

Women with a hospital discharge record indicating EH/EC possibly linked to previous delivery hospitalization, or birth record from 1987 to 2013

Randomly selected from remaining deliveries, frequency matched 10:1 by delivery year, maternal age at delivery

Women diagnosed with premenopausal (age < 50 years) breast cancer

Women giving birth to live babies between 1 January 1994 and 31 December 2003, not diagnosed with breast cancer between 1 January 1995 and 31 December 2008

US

US

Women with incident, primary, histologically confirmed breast cancer diagnosed between 1996 and 2001

Controls (n)

Criteria for GDM diagnosis

Criteria for cancer diagnosis

Maternal age at pregnancy, years

Type of cancer

Adjusted variables

RR (95% CI)

5743

ICD-9-CM for GDM (648.8x) in hospital discharge records

ICD-9-CM diagnostic codes

< 25, > 40

Endometrial

Race/ethnicity, year of delivery, maternal age at delivery, BMI

1.30 (0.85– 1.98)

Breast

Age (years), race/ethnicity, education (years), birth weight (lbs), parity, gestational age (weeks), weight gain (lbs), alcohol use, tobacco use, induction of labour, gestational hypertension

1.62 (0.30– 8.68)

Breast

Age, education, history of benign breast disease, family history of breast cancer, age at first pregnancy, number of pregnancies, menopausal status, age at menopause (postmenopausal women)

0.79 (0.48– 1.30)

Pr epr oo

Control source

593

al

Ardalan, 2016 [20]

Case source

Cases (n)

126

503

Recorded birth certificates

Cancer registry

34.4

rn

Wartko, 2017 [15]

Country

Jo u

First author, year [ref]

f

Table I Main characteristics of case–control studies included in the meta-analyses

Frequency-matched to cases by age and race at 2:1 ratio

960

1852

Self-administered questionnaire

Medical records, interview

24.0 (4.8)

21

Page 21 of 28

Lawlor, 2004 [27]

Women with breast cancer

Women without breast cancer

UK

f

Frequency-matched to cases in 5-year age groups

Pr epr oo

US

Non-Hispanic whites, Hispanics or Native Americans living in Colorado, New Mexico, Utah and selected counties of Arizona from 1999 to 2004, diagnosed with first primary histologically confirmed breast cancer

190

106

Self-reported

Self-reported

Surgeons’ and oncologists’ reports

Cancer registry

NA

< 30 (85.8%)

Breast

147

Menopausal status, mammography screening

Age (5-year categories), BMI (kg/m2) at age 15 years Breast

al

75

511

(< 20, 20–24, ≥25), full-term pregnancies (n = 0, 1–2, 3–4, ≥5)

rn

Rollison, 2008 [24]

US

Randomly selected from subjects diagnostically screened at mammography centre where subjects were diagnosed

Jo u

Sanderson, 2010 [23]

Hispanic women aged 30–79 years with histologically confirmed incident primary breast cancer and no history of cancer other than non-melanoma skin cancer

3690

Self-reported

Medical records

NA

Breast

Age

0.26 (0.03– 2.31)

Non-Hispanic white women: 0.68 (0.45– 1.03); Hispanic and/or Native American women: 0.75 (0.46–1.22)

1.59 (0.72– 3.49)

22

Page 22 of 28

US

f Frequency-matched, using random digit dialing, by age and geographical area to derive expected case distributions

Pr epr oo

Troisi, 1998 [28]

Hospital records of eligible patients were abstracted, using rapid ascertainment systems, to document details of clinical and pathological characteristics of diagnosed breast cancer

1235

1163

Self-reported

Medical records

20–44

Breast

Age at menarche, mammography, alcohol intake, BMI, site, race, and a combination variable representing parity and age at first birth

1.10 (0.73–1.5)

Jo u

rn

al

ref: reference; GDM: gestational diabetes mellitus; RR: relative risk; CI: confidence interval; US: United States; EH/EC: endometrial cancer/endometrial hyperplasia; ICD-9-CM: International Classification of Diseases, 9th Revision, Clinical Modification; BMI: body mass index; lbs: pounds; NA: not available; UK: United Kingdom

23

Page 23 of 28

Han, 2018 [14]

Park, 2017 [17]

China

NHIRD Taiwan

South Korea

National Health Insurance data for first pregnancy

US

The Sister Study: 4.2% had at least one pregnancy with GDM, 0.9% had multiple GDM pregnancies

Cases (n)/ sample size

Criteria for GDM diagnosis

Criteria for cancer diagnosis

Pr epr oo

Health administrative data from Quebec (universal health insurance)

Source of controls

Duration of follow-up, years

Maternal age at pregnancy, years

Type of cancer

Adjusted variables Matched by age group, delivery year, health region; adjusted for gestational hypertension, preterm delivery, infant size, parity, previous comorbidity, material deprivation index, ethnicity

Population-based

34,294/68,588

≥ 2 codes

Hospital discharge abstracts using ICD codes for each site-specific malignancy

Hospital-based

1063 + 18,444/ 47,373 + 943,199

ICD-9-CM codes 648, 250

ICD-9-CM codes 140–208

Mean: 6.84 ± 3.05

Mean: 28.97 ± 4.91

Various sites

Age, hypertension, dyslipidaemia, liver disease, infertility, kidney disease

NHSE health interview, examination, ICD-10-CM (codes O24.4, O24.9)

ICD-10-CM codes

Maximum: 10

Mean: 28.25 ± 3.28

Various sites

Age, BMI, smoking status, fasting blood glucose level

Breast

Birth cohort, race or ethnicity, educational attainment, age at first birth, age at menarche, relative weight at age 10, BMI, physical activity (quintiles of MET-h/week) during childhood, teen years

al

Peng, 2019 [13]

Canada

Cohort

rn

Pace, 2019 [29]

Country

Population-based

Jo u

First author, year [ref]

f

Table II Main characteristics of cohort studies included in the meta-analyses

Population-based

6569/102,900

2141/39,198

Self-reported

Self-reports verified by medical records (81%)

Mean: 13.1 ± 5.2

Around half (51.6%, n = 35,404) ≤ 30 years at cohort entry

Mean: 7.4

30–39

Various sites

24

Page 24 of 28

f Nurses in Nurses’ Health Study II cohort

Population-based

Fuchs, 2016 [18]

Israel

Soroka University Medical Center, Be’er Sheva

Population-based

Canada

Health administrative databases, Ontario

Israel

Maccabi Healthcare Services (MHS)

Sella, 2013 [22]

rn

Jo u

Bejaimal, 2016 [19]

Pr epr oo

US

Medical records

> 22 (prospective)

Mean: 33.1 (± 5.5)

9893/1,04,715

Medical records

Medical records

Mean: 12

Mean: 31.8 ± 5.9

2927/149,049

Medical records (75-g or 100-g OGTT)

Ontario Cancer Registry

2034/185,315

100-g OGTT (Carpenter– Coustan criteria)

Israel National Cancer Registry through linkage data

2377/86,972

al

Powe, 2016 [16]

Baseline questionnaire: self-reports of physician’s diagnosis confirmed by medical records (98%)

Population-based

Population-based

Median: 8

Mean: 5.19 ± 3.9

32 (28–35)

30.72 (± 5.22)

Breast

Female malignancies (ovary, uterine, breast, cervix)

Thyroid, breast

Various sites

BMI at age 18 (continuous), weight gain since age 18 (continuous), height (continuous), total physical activity (MET-h/week quintiles), alcohol intake, age at menarche, birth index, menopausal status, hormone therapy use, family history of breast cancer (mother, sister), personal history of benign breast disease, white race/ethnicity, mammography within past 2 years

NA

Income, number of physician visits within 3 years of index date Age, socioeconomic level, smoking, BMI level, gravidity, number of GP visits within 2 years of index date

25

Page 25 of 28

Jerusalem Perinatal Study

Population-based

1626/37,926

Perrin, 2007 [26]

Israel

Israel's National Population Registry

Population-based

54/37,926

Medical records (OGTT)

f

Israel

Pr epr oo

Perrin, 2008 [25]

Labour ward log at time of birth using separate rubrics in printed forms: pregnant women screened for glycosuria at each antenatal visit

Israel National Cancer Registry: ICD-10-CM code C50; 5-digit morphology codes 2, 3

Median: 34

< 25–35+

Breast

Age, birth order at first observed birth, social class, ethnic origin, education level, immigration status

Israel National Cancer Registry

Mean: 38.0

< 25–35+

Pancreas

Age at first observed birth, birth weight

Jo u

rn

al

ref: reference; GDM: gestational diabetes mellitus; ICD-9/10-CM: International Classification of Diseases, 9th/10th Revision, Clinical Modification; NHIRD: National Health Insurance Research Database; NHSE: National Health Screening Examination; US: United States; MET-h: metabolic equivalent of task per hour; BMI: body mass index; NA: not available; OGTT: oral glucose tolerance test

26

Page 26 of 28

Subgroups

Studies (n)

Pr epr oo

f

Table III Summary of associations between gestational diabetes mellitus (GDM) and breast cancer, stratified by cohort study Heterogeneity

2

Subgroup analysis

Q

P

I (%)

RR (95% CI)

P

9

70.41

< 0.001

88.6

1.08 (0.89–1.32)

0.428

4

10.34

0.016

71.0

0.88 (0.75–1.04)

0.121

5

21.94

< 0.001

81.8

1.31 (1.01–1.70)

0.043

> 100,000

5

45.12

< 0.001

91.1

1.17 (0.87–1.56)

0.303

≤ 100,000

4

17.06

< 0.001

82.4

0.98 (0.75–1.28)

0.894

Published codes

7

52.64

< 0.001

88.6

1.16 (0.93–1.45)

0.195

Self-reported

2

9.64

0.002

89.6

0.86 (0.54–1.38)

0.544

6

54.66

< 0.001

90.9

1.12 (0.86–1.45)

0.492

3

7.73

0.021

74.1

1.01 (0.74–1.37)

0.973

8

61.61

< 0.001

88.6

1.07 (0.85–1.34)

0.589

1

0.0

–

–

1.23 (1.09,1.39)

0.001

> 10 years

4

53.54

< 0.001

94.4

1.16 (0.73–1.86)

0.527

≤ 10 years

5

16.85

0.002

76.3

1.04 (0.86–1.24)

0.708

Total Region: North America Asia Sample size:

Medical records

Source control: Population-based Hospital-based Follow-up:

Jo u

Cancer registry system

rn

Breast cancer diagnosis:

al

Diabetes diagnosis:

RR (95% CI): relative risk (95% confidence interval) 27

Page 27 of 28

Type of cancer Lung Bone Mesothelial & soft-tissue Oral & pharyngeal Urinary bladder Kidney Lymphoma Hodgkin's lymphoma Non-Hodgkin's lymphoma

Pr epr oo

f

Table IV Cancer risk associated with gestational diabetes mellitus (GDM) for tumour types with limited numbers of studies Type of study

First author, year [reference]

Relative risk (95% CI)

Cohort

Pace, 2019 [29]

1.20 (0.73–1.99)

Cohort

Pace, 2019 [29]

1.00 (0.06–15.99)

Cohort

Pace, 2019 [29]

10.09 (0.75–135.72)

Cohort

Peng, 2019 [13]

1.11 (0.70–1.76)

Cohort

Peng, 2019 [13]

1.03 (0.53–2.03)

Cohort

Peng, 2019 [13]

2.17 (1.43–3.29)

Cohort

Peng, 2019 [13]

0.77 (0.49–1.23)

Cohort

Pace, 2019 [29]

1.51 (0.30–7.44)

Cohort

Pace, 2019 [29]

1.59 (0.64–3.94)

Cohort

Peng, 2019 [13]

0.58 (0.27–1.24)

Cohort

Peng, 2019 [13]

1.37 (1.04–1.80)

Cohort

Fuchs, 2016 [18]

2.10 (1.10–4.05)

Cohort

Sella, 2013 [22]

1.43 (0.40–5.09)

Cohort

Pace, 2019 [29]

0.39 (0.14–1.10)

Cohort

Sella, 2013 [22]

0.88 (0.11–7.08)

Uterine Respiratory

Jo u

Neurological

rn

Lung & bronchial

Head & neck

al

Bone & connective tissue

28

Page 28 of 28