Long-Term Follow-Up on Morbidity Among Women With a History of Gestational Diabetes Mellitus: A Systematic Review.

BACKGROUND: Gestational diabetes mellitus (GDM) complicates up to 10% of pregnancies and is a well-known risk factor for type 2 diabetes mellitus (T2DM) and cardiovascular disease. Little is known about possible long-term risks of other diseases.
BACKGROUND: The aim was to review the literature for evidence of associations with morbidity other than T2DM and cardiovascular disease and with long-term mortality.
METHODS: A systematic review based on searches in Medline, Embase, and Cochrane Library until March 31, 2021, using a broad range of keywords. We extracted study characteristics and results on associations between GDM and disease occurrence at least 10 years postpartum, excluding studies on women with diabetes prior to pregnancy or only diabetes prior to outcome. The results are reported according to the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA). Newcastle-Ottawa Scale was used to assess risk of bias.
RESULTS: We screened 3084 titles, 81 articles were assessed full-text, and 15 included in the review. The strongest evidence for an association was for kidney diseases, particularly in Black women. We found indication of an association with liver disease, possibly restricted to women with T2DM postpartum. The association between GDM and breast cancer had been studied extensively, but in most cases based on self-reported diagnosis and with conflicting results. Only sparse and inconsistent results were found for other cancers. No study on thyroid diseases was found, and no study reported on short-term or long-term mortality in women with a history of GDM.
CONCLUSION: Given the frequency of GDM, there is a need for better evidence on possible long-term health consequences, in particular, studies based on comprehensive records of diagnosis of GDM and long-term health outcomes.

Gestational diabetes mellitus (GDM) is defined as abnormal glucose intolerance diagnosed during pregnancy. GDM complicates approximately 5% of pregnancies in Denmark (1), in line with the 2% to 9% of pregnancies reported in other parts of Europe, in Australia, and in North America (2). Several recent studies suggest that the prevalence of GDM is increasing worldwide, making it the most common medical complication of pregnancy (3-5). In most women, the glucose tolerance returns to normal within a few days after giving birth. Women with a history of GDM are at significantly increased risk of developing diabetes mellitus (DM) later in life, particularly type 2 DM (T2DM) (6). It is well-known that T2DM is associated with increased risk of morbidity and mortality, for example, from cardiovascular events such as myocardial infarction, and from heart and renal failure (7-9). It is therefore not surprising that women with a history of GDM are also at significantly increased risk of cardiovascular diseases and heart failure (10, 11) following increased occurrence of hypertension and hyperlipidemia even during the fertile years (12).

However, little is known about the risk of other types of morbidity, such as cancer in women with a history of GDM. Studies based on a median 8-year follow-up of women with GDM have shown an increased risk of thyroid cancer but a reduced risk of premenopausal breast cancer as compared with women without GDM (13). In a study with a follow-up period of 1 to 9 years, development of nonalcoholic fatty liver disease (NAFLD) was also significantly associated with a history of GDM (14). Most studies do not have a follow-up period longer than 10 years, and therefore the possible long-term consequences of GDM apart from T2DM and cardiovascular disease are poorly described (15, 16).

Therefore, the aim of the present systematic review was to investigate whether women with a history of GDM as compared with other women have 1) an increased long-term risk of chronic diseases independently of and apart from T2DM and cardiovascular diseases; and 2) an increased long-term risk of premature mortality. The review was restricted to studies with a mean follow-up period of at least 10 years after a pregnancy complicated by GDM.

Methods

Search Strategy

We conducted a systematic search in the databases Medline (from 1950), Embase (from 1949), and the Cochrane Library (from 1993). Search strategies were developed and conducted in collaboration with a professional librarian, using a combination of medical subject headings (MeSH) terms and free text search terms in titles, keywords, key headings, and abstract fields. We suspected that our findings would be limited, and therefore we conducted a broad search strategy. For GDM, the MeSH terms used were “Diabetes” and “Gestational”, and other terms were (gestation*adj2 diabet*).ab,kf,ti., gdm.ab,kf,ti., (pregnan*adj2 induce*adj2 diabet*).ab,kf,ti. For the full list of search terms used for outcomes, see Supplementary table 1 (17). In addition to searches, the reference lists of relevant studies were checked manually for missing studies, which we imported for screening. Searches were conducted until March 31, 2021.

Inclusion and Exclusion Criteria

We included peer-reviewed original articles of cohort studies, case-control studies, and clinical trials, in which women diagnosed with and/or had self-reported GDM in a previous pregnancy, were investigated for subsequent long-term morbidity or mortality. We chose to include outcomes related to organs other than the heart, well-known to be affected by GDM; hence, included outcomes were liver disease (NAFLD, fibrosis and cirrhosis of the liver, liver failure, and liver transplant), renal disease (chronic renal failure and renal insufficiency), cancer (breast, endometrial, ovarian, cervix uteri, kidney, bladder, thyroid, and urological cancer), thyroid disease (hypothyroidism, hyperthyroidism, and thyrotoxicosis), and deaths from these causes. We excluded studies on women with diabetes (without any differentiation of type) prior to pregnancy or only diabetes prior to outcomes, studies investigating only outcomes in offspring, as well as studies with short-term follow-up (a reported mean or median of less than 10 years postpartum). Review articles, book chapters, protocols of ongoing studies, conference abstracts, and letters were also excluded. Only articles written in English, Danish, Swedish, or Norwegian were included, which are the languages the reviewers comprehend. No limitation regarding publication year was used.

Study Selection

Title and abstract screening and subsequently full-text screening were performed manually by 2 independent reviewers (L.R.F.M. and S.G.-K.) with disagreements resolved by consensus. Findings were reported following the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) reporting guidelines (Fig. 1).

Figure 1.

Flow diagram of the literature search and selection process.

Data extraction

Characteristics of each study were retrieved manually and organized in Table 1. Results were retrieved manually and organized in Table 2. For each study, 2 reviewers (L.R.F.M. and E.L.), independently reported the number of studied women with a history of GDM and number of women without. For each group, we reported the number of women with the studied outcome in the follow-up period. We reported the association between GDM and the outcome studied, using the association measure reported in the original study—odds ratio (OR), rate ratio (RR), or hazard ratio (HR)—together with the reported 95% CI. Where available in the original papers, we included both crude results, adjusted results, and results stratified by ethnicity and/or a DM diagnosis subsequent to the GDM diagnosis.

Table 1.

Descriptive data on included studies

First author, year, (ref)	Country	Study design	Ethnicity	Exposure: +/- number in cohort	Outcomes: +/- number in case-control	Duration of follow-up, years	GDM assessment method
Kidney disease
Bomback 2010 (19)	USA	Cross-sectional, screening data for women at high risk of chronic kidney disease	White, African American, other	GDM+: 571 GDM-: 25045	Microalbuminuria; Macroalbuminuria; Chronic kidney disease (CKD)	Cross-sectional screening study	Self-reported at mean age 51-53 years
Beharier 2015 (20)	Israel	Cohort, hist	Nonselected population	GDM+: 9542 GDM-: 88426	Renal disease	Mean: 11.2	Administrative health data
Dehmer 2018 (21)	USA	Cohort, pros, CARDIA	White, Black	GDM+: 101 GDM-: 719	Chronic kidney disease (urine albumin and creatinine)	Mean: 20.8	Self-reported
Rawal 2018 (22)	Denmark	Cohort, pros. Only participants with examined specimens	European	GDM+: 607 GDM-: 619 (GDM- random selected from cohort)	Chronic kidney disease (estimated glomerular filtration rate [eGFR] and urinary albumin to creatinine ratio [UACR])	Mean: 13 (9-16)	Interviews, administrative health data: ICD-10 DO24.4, DO24.9
Breast cancer
Troisi 1998 (23)	USA	Case-control	White, African American, other	GDM	Breast cancer (medical records): BC+: 1239 BC-: 1166	Not relevant	Self-reported
Perrin 2008 (24)	Israel	Cohort, hist	European, Israeli, other West Asian, North African	GDM+: 410 GDM-: 37516	Breast cancer (Israel Cancer Register ICD-10: C50)	Median: 34	Labor ward logs (GDM based on screening for glucosuria at antenatal visits and subsequently OGTT)
Rollison 2008 (25)	USA	Case-control	Non-Hispanic White, Hispanic, Native American	GDM	Breast cancer (Cancer register: ICD-10: C50) BC+: 2324 BC-: 2523	Reported at mean age 57.0	Self-reported
Brasky 2013 (26)	USA	Case-control	Predominantly Caucasian	GDM	Breast cancer (Medical records, interview): BC+: 960 BC-: 1852	Reported at mean age 58.2	Self-reported
Powe 2017 (27)	USA	Cohort, pros	White, African American, Asians	GDM+: 5188 GDM-: 81784	Breast cancer (questionnaire, medical records)	>22	Self-reported
Fuchs 2017 (28)	Israel	Cohort, hist	Nonselected population	GDM+: 9893 GDM-: 94822	Ovary, uterine, breast, cervix cancer. Medical records	Mean: 12	Perinatal database
Pace 2020 (29)	Canada	Cohort, hist	European ancestry	GDM+: 34294 GDM-: 34294	Breast cancer, thyroid cancer, and other female genital organ cancer	Mean: 13.1	Administrative health data: ICD-codes
Bertrand 2020 (30)	USA	Cohort, pros	African American	GDM+: 2059 GDM-: 39708	Breast cancer (self-report, death certificates, cancer register records, medical records)	Years since last birth: ≥10	Self-reported
Liver disease
Ajmera 2016 (32)	USA	Cohort, pros, CARDIA	Caucasian, African American	GDM+: 124 GDM-: 991	NAFLD (CT scan)	25	Self-reported
Retnakaran 2019 (33)	Canada	Cohort, hist	Chinese, South Asian, other	GDM+: 17932 GDM-: 680146	Serious liver disease (hospitalized for cirrhosis, liver failure, liver transplant)	Median: 14	Administrative health data
Other female genital organ and urological cancer
Fuchs 2017 (28)	Israel	Cohort, hist	Nonselected population	GDM+: 9893 GDM-: 94822	Ovary, uterine, breast, cervix cancer. Medical records	Mean: 12	Perinatal database
Wartko 2017 (35)	USA	Case-control	Non-Hispanic White, non-Hispanic Black, American Indian/ Alaska Native, Asian, Native Hawaiian/Pacific Islander, Hispanic	History of GDM	Endometrial cancer (ICD-9-CM 182.0) EC+: 340 EC-: 5743	Median: 14	ICD-9-CM 648.8 in hospital discharge records
Pace 2020 (29)	Canada	Cohort, hist	European ancestry	GDM+: 34294 GDM-: 34294	Breast, thyroid, other female genital organ, and urological cancer	Mean: 13.1	Administrative health data: ICD-codes
Thyroid cancer
Pace 2020 (29)	Canada	Cohort, hist	European ancestry	GDM+: 34294 GDM-: 34294	Breast, thyroid, other female genital organ, and urological cancer	Mean 13.1	Administrative health data: ICD-codes

Abbreviations: BC, breast cancer; EC, endometrial cancer; GDM, gestational diabetes mellitus; Hist, historical; ICD, International Classification of Diseases; NAFLD, nonalcoholic fatty liver disease; OGTT, oral glucose tolerance test; Pros, prospective; USA, United States of America.

Table 2.

Association between history of GDM and long-term morbidity outcomes

First author, year, (ref)	Exposure number		Outcome number		Measure	Unadjusted (age-adjusted when specified) (95% CI)	Adjusted (95% CI)	Adjustment variables
	GDM +	GDM -	GDM +	GDM -
Kidney disease
Bomback 2010 (19)	571	25045	[57] [4] [55] [67]	[1928] [150] [1603] [3506]	OR	Microalbuminuria: 1.34 (1.02-1.77) Macroalbuminuria: 1.13 (0.42–3.07) CKD stages 1–2: 1.54 (1.16–2.05) CKD stages 3–5: 0.84 (0.65–1.09) Not reported Not reported	Microalbuminuria: 1.36 (1.03–1.80) Macroalbuminuria: 1.13 (0.41–3.09)aCKD stages 1–2: 1.54 (1.16–2.05)aCKD stages 3–5: 0.94 (0.71–1.25)aCKD stages 1-2:African American women: 2.32 (1.50–3.60)bWhite women: 1.12 (0.68–1.84)b	a) Age, race, BMI, current smoking, alcohol use, hypertension, dyslipidemia, eGFR, and family history of kidney disease b) Age, BMI, current smoking, alcohol use, hypertension, dyslipidemia, eGFR, and family history of kidney disease
Beharier 2015 (20)	9542	88426	23	91	OR HR	2.34 (1.4-3.7) Not reported	Not reported 1.7 (1.05-2.6)	Parity and maternal age
Dehmer 2018 (21)	101 42 59	719 289 430	17 13 4	88 45 43	HR	1.46 (0.87-2.45) Not reported Not reported	1.33 (0.78-2.26)cBlack women: 1.96 (1.04-2.67)dWhite women: 0.65 (0.23-1.83)d	c) Age, race, BMI, smoking, family history of diabetes, fasting blood glucose, baseline eGFR, education, high-density lipoprotein cholesterol, systolic blood pressure and physical activity score d) Age, race, BMI, smoking, family history of diabetes, fasting blood glucose, baseline eGFR, education, high-density lipoprotein cholesterol, systolic blood pressure and physical activity score + interaction term for race and GDM
Rawal 2018 (22)	[601] [181] [420] [597] [179] [418]	[613] [604] [604] [607] [598] [598]	38 22 16 26 14 12	24 23 23 14 13 13	RR	Glomerular hyperfiltration: 1.6 (1.0-2.7) GDM+/DM+: 3.2 (1.8-5.7) GDM+/DM-: 1.0 (0.5-1.9) UACR: 1.9 (1.0-3.6) GDM+/DM+: 3.6 (1.7-7.6) GDM+/DM-: 1.3 (0.6-2.9)	Glomerular hyperfiltration: 1.1 (0.6-2.1) GDM+/DM+: 3.2 (1.4-7.0) GDM+/DM-: 0.8 (0.4-1.6) UACR: 1.4 (0.6-2.9) GDM+/DM+: 2.3 (1.1-5.9) GDM+/DM-:1.0 (0.4-2.3)	Age at index pregnancy, smoking during pregnancy (yes vs no), education (high school or less vs more than high school education), family history of diabetes (yes vs no), prepregnancy BMI, and hypertension before pregnancy
Breast cancer
Troisi 1998 (23) Case-control	Cases [1235]	Controls [1163]	GDM in cases 67	GDM in controls 65	RR	0.99 (0.70-1.4) Last pregnancy ≥5 years: 1.2 (0.74-1.9)	1.1 (0.73-1.5) Last pregnancy ≥5 years: 1.3 (0.77-2.1)	Age, site, race, and a combination variable representing parity and age at first birth, BMI, age at menarche, mammography, and alcohol intake
Perrin 2008 (24)	410 [410] [396]	[37516] [37516] [34551]	29 7 22	1597 637 960	RR	Age-adjusted: 1.6 (1.1-2.3) Not reported Not reported	1.5 (1.0-2.1) Age at diagnosis <50: 1.0 (0.5-2.1) Age at diagnosis ≥50: 1.7 (1.1-2.5)	Age and birth order at first observed birth, social class, ethnic origin, education, and immigration status
Rollison 2008 (25) Case-control	Cases 2324 Not reported Not reported	Controls 2523 Not reported Not reported	GDM in cases 75 47 25	GDM in controls 106 86 18	OR	Age-adjusted: 0.70 (0.51-0.94) Age at GDM onset 15-34: 0.54 (0.37-0.77) Age at GDM onset ≥35: 1.35 (0.73-2.48)	0.71 (0.52-0.98) Age at GDM onset 15-34: 0.56 (0.38-0.82) Age at GDM onset ≥35: 1.34 (0.72-2.52)	Age (5-year categories), study site (Arizona, New Mexico, Colorado, Utah), menopausal status (pre-, peri, or postmenopausal), BMI at study entry (<25, 25-29, or ≥30), BMI at age 15 years (<20, 20-24, or ≥25), number of full-term pregnancies (0, 1-2, 3-4, or ≥5), age at first pregnancy (<30 years vs ≥30 years), age at menarche (<13, 13, or >13 years), lifetime physical activity, family history of breast cancer and breastfeeding history (ever vs never)
Brasky 2013 (26) Case-control	Cases 960 267 693	Controls 1852 512 1340	GDM in cases 28 13 15	GDM in Controls 66 36 30	OR	Age- adjusted: 0.81 (0.52-1.27) Not reported Not reported	0.79 (0.48-1.30) Premenopausal 0.60 (0.29-1.27) Postmenopausal 1.03 (0.52-2.05)	Age, education, history of benign breast disease, family history of breast cancer, age at first pregnancy, number of pregnancies, menopausal status, and age at menopause (among postmenopausal women)
Powe 2017 (27)	5188	81784	100 5 95	2227 2224 2224	HR	Age-adjusted: 0.68 (0.55-0.84) GDM+/T2DM+: 0.23 (0.09-0.61) GDM+/T2DM-: 0.72 (0.58-0.90)	0.68 (0.55-0.84) GDM+/T2DM+: 0.26 (0.10-0.68) GDM+/T2DM-: 0.72 (0.58-0.89)	Age, BMI at age 18 (continuous), weight gain since age 18 (continuous), height (continuous), total physical activity (MET-hours/week, quintiles), alcohol intake (none, 1-14 grams/day, ≥15 grams/day), age at menarche (≤10 years old, 11-12, 13-14, ≥15), birth index (continuous), total breastfeeding (none, <6 months, ≥6 months), menopausal status (premenopausal, postmenopausal, unknown), hormone therapy use (never, ever use of estrogen + progesterone, past: estrogen only or other, current: estrogen only or other), family history of breast cancer in mother or sister (yes/no), personal history of benign breast disease (yes/no), White race/ethnicity (yes/no), and mammography within the past 2 years (<40 years old, ≥40 and no mammography, ≥40 and mammography for screening, ≥40 and mammography for abnormality/symptoms)
Fuchs 2017 (28)	9893	94822	[91]	[436]	OR	2.0 (1.60-2.51)	Not reported
Pace 2020 (29)	34294	34294	346	392	HR	Not reported	0.93 (0.80-1.09)	Gestational hypertension, preterm delivery, infant size, parity, prior comorbidity, maternal deprivation index and ethnicity
Bertrand 2020 (30)	2059	39708	70	1609	HR	1.00 (0.78-1.27) Not reported	0.98 (0.77-1.25) 10 + years since last birth 0.92 (0.69-1.22)	Age, questionnaire cycle, BMI at 18, recent BMI, parity, menarche, age at first birth, oral contraceptive duration, and family history of breast cancer
Liver disease
Ajmera 2018 (32)	124 61 62	991 75 909	17 [12] [5]	58 [15] [41]	OR	2.56 (1.44-4.55) Not reported Not reported	2.29 (1.23-4.27) GDM+/DM+: 1.18 (0.45-3.10) GDM+/DM-: 1.93 (0.72-5.14)	Baseline HOMA-IR, triglycerides, and history of GDM
Retnakaran 2019 (33)	17932	680146	15/105	11/105	HR	1.40 (1.01-1.94) GDM+/T2DM+: 1.88 (1.23-2.87) GDM+/T2DM-: 1.26 (0.76-2.09)	Not reported GDM+/T2DM+: 1.56 (1.02-2.39) GDM+/T2DM-: 1.15 (0.69-1.91)	Age, income, region of residence, hypertension, dyslipidemia, and ethnicity
Endometrial cancer
Fuchs 2017 (28)	9893	94822	[11]	[47]	OR	2.1 (1.01-4.05)	Not reported
Wartko 2017 (35) Case-control	Cases 340 [354]	Controls 5743	GDM in Cases [32]	GDM in Controls [322]	OR	Nonimputed, adjusted: 1.7 (1.14-2.55)e	Imputed, adjusted 1.30 (0.85-1.98)f	e) Race/ethnicity, year of delivery, maternal age at delivery f) Race/ethnicity, year of delivery, maternal age at delivery and body mass index
Pace 2020 (29)	34294	34294	7	3	HR	Not reported	0.31 (0.07-1.46)	Gestational hypertension, preterm delivery, infant size, parity, prior comorbidity, maternal deprivation index and ethnicity
Ovarian cancer
Fuchs 2017 (28)	9893	94822	[10]	[46]	OR	2.0 (1.03-4.04)	Not reported
Pace 2020 (29)	34294	34294	24	32	HR	Not reported	1.02 (0.66-1.58)	Gestational hypertension, preterm delivery, infant size, parity, prior comorbidity, maternal deprivation index and ethnicity
Cervical cancer
Fuchs 2017 (28)	9893	94822	[23]	[199]	OR	1.1 (0.70-1.65)	Not reported
Pace 2020 (29)	34294	34294	44	38	HR	Not reported	1.21 (0.76-1.92)	Gestational hypertension, preterm delivery, infant size, parity, prior comorbidity, maternal deprivation index and ethnicity
Urological cancer
Pace 2020 (29)	34294	34294	25	21	HR	Not reported	0.60 (0.24-1.48)	Gestational hypertension, preterm delivery, infant size, parity, prior comorbidity, maternal deprivation index and ethnicity
Thyroid cancer
Pace 2020 (29)	34294	34294	125	96	HR	Not reported	1.39 (1.03-1.89)	Gestational hypertension, preterm delivery, infant size, parity, prior comorbidity, maternal deprivation index and ethnicity

Abbreviations: BMI, body mass index; CKD, chronic kidney disease; DM, diabetes mellitus; GDM, gestational diabetes mellitus; HOMA-IR, homeostatic model assessment index; HR, hazard ratio; OR, odds ratio; RR, relative risk; T2DM, type 2 diabetes; UACR, urine albumin to creatinine ratio.

As disease outcomes were defined differently across the studies and/or reported for different subgroups, it was not meaningful to attempt summarizing the results statistically in meta-analyses, including tests of heterogeneity, and hence it was not possible to meaningfully perform any subanalyses.

Assessment of risk of bias

Two reviewers (L.R.F.M. and J.L.) assessed the risk of bias in each included study independently. Discrepancies were resolved through discussion and consensus. The Newcastle-Ottawa Scale was used systematically for the assessment of risk of bias, for cohort studies as well as for case-control studies, as recommended by the Cochrane Non-Randomized Studies Methods Working Group. The Newcastle-Ottawa Scale assessment scores the selection, comparability, and ascertainment of exposure and outcome, and summarizes the quality with a total score between 0 and 9 (18). The outcome of this assessment is reported in Supplementary table 2 (17).

Results

We imported a total of 3084 studies for screening from databases and reference lists (PRISMA flow diagram Fig. 1). After exclusion of 1344 duplicates, 1740 studies remained for manual screening by title and abstract, of which 1659 were found irrelevant, leaving 81 studies that were assessed for full-text eligibility. Sixty-six studies were excluded for the reasons listed in Fig. 1. A final selection of 15 original articles matching our criteria was included in the review. Descriptive data of the studies are summarized in Table 1.

Kidney Disease

Four studies on the long-term association between GDM history and kidney diseases were identified, all cohort studies (19-22).

Bomback et al (2010) (19) examined whether GDM, in the absence of subsequent overt DM, increased the risk of abnormal urinary albumin excretion and impaired glomerular filtration rate among the participants of the National Kidney Foundation’s Kidney Early Evaluation Program (KEEP). This was a screening study of U.S. adults at high risk of chronic kidney disease (CKD). Among women without DM, 571 had a self-reported history of GDM while 25 045 did not. They found an increased risk of microalbuminuria (OR 1.36 [95% CI, 1.03–1.80]) and CKD stages 1-2 (OR 1.54 [95% CI, 1.16–2.05]), while no association was found between a history of GDM and macroalbuminuria (OR 1.13 [95% CI, 0.41–3.09]) or CKD stages 3-5 (OR 0.94 [95% CI, 0.71–1.25]). Stratification by race showed a significantly increased risk of CKD stages 1-2 among African American women (OR 2.32 [95% CI, 1.50–3.60]), but not in White women; (OR 1.12 [95% CI, 0.68–1.84]).

Beharier et al (2015) (20) conducted a population-based cohort study in Israel including 9542 women with a history of GDM and 88 426 women without. Mean follow-up was 11 years, and 23 GDM women developed renal morbidity compared with 91 women without GDM. In unadjusted analyses, GDM was found to be a significant risk factor for renal morbidity (OR 2.34 [95% CI, 1.4-3.7]) based on a significantly increased risk of hypertensive renal disease, while renal failure, chronic renal failure, and end-stage renal disease were not individually associated with increased risk. In analyses adjusted for maternal age and parity, GDM was associated with increased risk of hospitalization for renal disease (HR 1.7 [95% CI, 1.05-2.6]).

Dehmer et al (2018) (21) studied the association between self-reported history of GDM and CKD in the Coronary Artery Risk Development in Young Adults (CARDIA) cohort. They extracted data from 820 women, who were nulliparous at enrollment, delivered at least once during the follow-up period, and had kidney function measured during the 25 years of follow-up. There were 17 cases of CKD in women with GDM, and 88 in women without GDM, with an overall adjusted HR of 1.33 (95% CI, 0.78-2.26), and HR 1.96 (95% CI, 1.04-2.67) in the subgroup of Black women.

Rawal et al (2018) (22) performed a 9- to 16-year follow-up of women with a history of GDM within the Danish National Birth Cohort. They investigated the risk of glomerular hyperfiltration (defined as estimated glomerular filtration rate [eGFR] of 116.4 mL/min/1.73 m2), elevated urine albumin to creatinine ratio (UACR) (defined as ≥ 20 mg/g) and micro- or macroalbuminuria (defined as UACR > 30 mg/g). The adjusted risks were not increased; RR 1.1 (95% CI, 0.6-2.1) for glomerular hyperfiltration and RR 1.4 (95% CI, 0.6-2.9) for UACR. However, both risks were significantly increased among women with a history of GDM and subsequent DM; RR 3.2 (95% CI, 1.4-7.0) and RR 2.3 (95% CI, 1.1-5.9), respectively.

In summary, in studies of Black women from the United States, a history of GDM was associated with an increased risk of later kidney disease, and an increased risk was also reported from an Israeli hospital database. Glomerular hyperfiltration and UACR were increased in Danish women, but only in those with DM subsequent to the GDM diagnosis.

Breast Cancer

Eight studies were detected, investigating long-term risk of breast cancer after a pregnancy complicated with GDM (23-30), 3 case-control and 5 cohort studies.

A U.S. population-based case-control study by Troisi et al (1998) (23) evaluated the risk of breast cancer following various pregnancy outcomes, including self-reported history of GDM. They identified women between 20 and 44 years of age who were diagnosed with in situ or invasive breast cancer, as well as matched controls, and obtained in-person interviews regarding details on their pregnancies. Sixty-seven breast cancer cases reported a history of GDM. No association was found between GDM and breast cancer: unadjusted RR 0.99 (95% CI, 0.70-1.4), adjusted RR 1.1 (95% CI, 0.73-1.5).

An Israeli cohort study, based on hospital records, conducted by Perrin et al (2008) (24), included 410 women with a history of GDM and 37 516 women without. Presence of GDM was determined from notes on “pre-diabetes” in the labor ward logs. At follow-up in the Israeli Cancer Register over a median of 34 years, 29 cases of breast cancer were identified in the GDM cohort, and 1597 cases in the other women, with an adjusted RR 1.5 (95% CI, 1.0-2.1). Stratified by age at diagnosis, the adjusted RR was 1.0 (95% CI, 0.5-2.1) for women diagnosed younger than age 50, and 1.7 (95% CI, 1.1-2.5) for those diagnosed at or older than age 50.

Rollison et al (2008) (25) conducted a population-based case-control study in non-Hispanic White/Hispanic/Native American women in the United States who were diagnosed with invasive breast cancer. Among 2324 breast cancer patients, 75 reported a history of GDM, and among 2523 controls, the number was 106, giving a statistically significantly decreased adjusted OR of 0.71 (95% CI, 0.52-0.98). The OR for women with onset of GDM before the age of 35 was 0.56 (95% CI, 0.38-0.82), and for those with onset at age 35 or older 1.34 (95% CI, 0.72-2.52).

In another population-based case-control study from the United States, Brasky et al (2013) (26) conducted an investigation in women aged 35 to 79 years with a primary, histologically confirmed breast cancer and matched controls. Of the 960 breast cancer cases, 28 reported a history of GDM, and of the 1852 controls, there were 66 who did. There was no association between a history of GDM and breast cancer: OR 0.79 (95% CI, 0.48-1.30), neither in premenopausal women (OR 0.60 [95% CI, 0.29-1.27]), nor in postmenopausal women (OR 1.03 [95% CI, 0.52-2.05]).

In the Nurses’ Health Study II, Powe et al (2017) (27) studied 5188 parous women with a self-reported history of GDM and 81 784 without any diabetes followed over 22 years. The self-reported GDM data had previously been validated against medical records in a subgroup of study participants, with 94% of cases being confirmed (31). One hundred women with GDM and 2277 women without developed breast cancer. There was a significantly decreased risk of invasive breast cancer in GDM women as compared with other women: HR 0.68 (95% CI, 0.55-0.84). In those who subsequently developed T2DM, the HR was 0.72 (95% CI, 0.58-0.89), and in those without, the HR was 0.26 (95% CI, 0.10-0.68).

Fuchs et al (2017) (28) undertook a hospital-based cohort study in Israel comparing the incidence of long-term female malignancies (breast, ovary, uterine, and cervix cancer) in women with and without a recorded diagnosis of GDM. The cohort was followed for 26 years. In 9893 women with GDM, 91 developed breast cancer, and in the 94 822 women without GDM, 436 developed breast cancer. There was a statistically significantly increased risk of breast cancer after a pregnancy complicated by GDM: OR 2.0 (95% CI, 1.60-2.51).

Pace et al (2020) (29) undertook a Canadian cohort study of 34 294 women with a singleton birth and at least 2 diagnostic codes for GDM and a matched reference cohort of women without GDM. Cancer diagnoses (breast, reproductive organs, urological, thyroid etc.) were retrieved from discharge records after a mean follow-up period of 13.1 years. Among women with a history of GDM, 346 developed breast cancer, and in those without GDM, 392 developed breast cancer. No association between a history of GDM and breast cancer was found (HR 0.93 [95% CI, 0.80-1.09]).

Bertrand et al (2020) (30) studied the association between GDM and breast cancer in the U.S. Black Women’s Health Study (BWHS). They included 41 767 parous women, of whom 2059 reported a history of GDM. Women were followed for a maximum of 32 years. Among women with GDM, 70 had developed breast cancer. There was no association between a history of GDM and risk of breast cancer: HR 0.98 (95% CI, 0.77-1.25). Restricted to women with at least 10 years since their last birth, the result was HR 0.92 (95% CI, 0.69-1.22).

In summary, among studies exploring the association between a history of GDM and risk of breast cancer, no association or a decreased risk were found in 3 cohort and 3 case-control studies from North America, while elevated risks were found in 2 Israeli studies based on hospital records.

Liver Disease

Two studies on the association between prior GDM and long-term liver disease met our inclusion criteria (32, 33), Table 2.

Ajmera et al (2016) (32) evaluated the impact of a history of GDM on the prevalence of NAFLD in the CARDIA study, which was a multicenter, population-based, prospective cohort study that in 1985-1986 enrolled Caucasian (50%) and African American (50%) adults aged 18 to 30 years (34). Participants underwent an initial examination on anthropometric and metabolic profiling and follow-up examinations at 2, 5, 7, 10, 15, 20, and 25 years (2010-2011) with a retention rate of 72% in the surviving cohort. Of the 2787 women in the CARDIA cohort, Ajmera et al included 1115 women with ≥ 1 births, free of diabetes, and who underwent computed tomography (CT) quantification of hepatic steatosis at the 25-year follow-up. One hundred and twenty-four women had a self-reported history of GDM and 17 of those were diagnosed with NAFLD at follow-up. Compared with women with no history of GDM, and after adjustments for age, parity, and metabolic risk factors, the long-term risk of NAFLD after a history of GDM was significantly increased: OR 2.29 (95% CI, 1.23-4.27). When further stratified by subsequent development of DM before diagnosis of NAFLD; the OR for having NAFLD was 1.18 (95% CI, 0.45-3.10) for those with DM, and 1.93 (95% CI, 0.72-5.14) for those without (32).

The risk of developing serious liver disease defined as liver cirrhosis, liver failure, and liver transplantation after a history of GDM was investigated by Retnakaran et al in 2019 (33) in a Canadian cohort with 698 078 participants. They hypothesized that the risk would be increased, due to the well-described increased risk of liver disease in individuals diagnosed with T2DM. They found 17 932 women with GDM, of whom 15 cases per 100 000 person years developed serious liver disease during a median follow-up duration of 14 years. The long-term association between GDM and serious liver disease was slightly increased: HR 1.40 (95% CI, 1.01-1.94), but only in the women with a history of GDM who subsequently developed T2DM: HR 1.56 (95% CI, 1.02-2.39) vs HR 1.15 (95% CI, 0.69-1.91) in those without T2DM.

Both studies indicated a positive association between GDM and long-term risk of liver disease, but the association was attenuated when data were stratified by subsequent development of DM, and no clear pattern was seen.

Female Genital Organ Cancer

Three studies have examined the association between GDM and genital cancer (28, 29, 35).

Fuchs et al (2017) (28) investigated a history of GDM in relation to both breast cancer (described above) and to female genital cancers. The future risks of developing ovarian and endometrial cancer after a pregnancy with GDM were significantly increased: OR 2.0 (95% CI, 1.03-4.04) and OR 2.1 (95% CI, 1.01-4.05), respectively. No association was found with cervical cancer, OR 1.1 (95% CI, 0.70-1.65).

In a hospital-based case-control study from Washington State, Wartko et al (2017) (35) estimated the association between a history of GDM and risk of endometrial cancer. Among 340 women with endometrial cancer, 32 had a history of GDM, and among 5743 matched controls, 322 had a history of GDM: OR 1.70 (95% CI, 1.14-2.55) in observed, nonimputed data, and OR 1.30 (95% CI, 0.85-1.98) when missing data were accounted for by imputation and further adjustment for BMI.

Pace et al (2020) (29) (described above) found no association between overall risk of reproductive organ malignancies (HR 1.08 [95% CI, 0.91-1.29]), endometrial cancer (HR 0.31 [95% CI, 0.07-1.46]), ovarian cancer (HR 1.02 [95% CI, 0.76-1.92]), or cervical cancer (HR 1.21 [95% CI, 0.76-1.92]).

No consistent pattern was found for the association between a history of GDM and later occurrence of female genital cancers, although the overall trend was toward no association in 2 of the 3 studies.

Thyroid Disease Including Thyroid Cancer

In this disease group, only 1 study was found.

Pace et al (2020) (29) (described above) is the only study of thyroid cancer in women with a history of GDM. Among the 34 294 women with GDM, 125 were later diagnosed with thyroid cancer compared with 96 women without GDM, resulting in an increased risk in GDM women: HR 1.39 (95% CI, 1.03-1.89).

No study investigating the long-term relationship between history of GDM and subsequent development of thyroid disease (hypothyroidism, hyperthyroidism, thyrotoxicosis) was found.

Other Outcomes

Pace et al (2020) (29) found no association between a history of GDM and later risk of urological cancer (HR 0.60 [95% CI, 0.24-1.08]). Pace et al (2020) also reported on a number of other cancer sites and found no HR statistically significantly different from 1. No study on long-term mortality was detected.

Discussion

Main Finding

In the present systematic review, we mapped potential long-term health consequences of GDM other than T2DM and cardiovascular disease. Given that GDM is among the most common medical complication of pregnancies, surprisingly few studies were identified.

We found some indication for an increased risk of liver disease following a history of GDM. It was uncertain, however, whether this risk was restricted to those with a subsequent diagnosis of DM. Three studies indicated that women, especially Black women, with a history of GDM had an increased risk of subsequent kidney disease. There was also some indication of increased risk of glomerular hyperfiltration and UACR but restricted to women with subsequent DM. The possible risk of breast cancer following a history of GDM showed somewhat inconsistent results. In 3 cohort and 3 case-control studies from North America no association was found, but an increased risk was indicated in 2 Israeli cohort studies based on hospital records. Only sparse and somewhat inconsistent results were found for the association between a history of GDM and other cancers. No study on the association between a history of GDM and thyroid diseases was found, except 1 on thyroid cancer, and no study had reported on mortality in women with GDM.

Kidney Diseases

The strongest indication for long-term health consequences of GDM came from studies of kidney diseases. This is consistent with the fact that inflammatory markers shown to predict both cardiovascular events and chronic kidney disease are elevated in women with a history of GDM (36), supporting the hypothesis that women with a history of GDM can develop subclinical inflammation and persistent generalized vascular dysfunction (20). Additionally, there have been indications of DM playing an important role in the development of future renal damage (22). This was not surprising, as it is well established that T2DM is a risk factor for chronic kidney disease (37, 38).

The excess risk of kidney disease following a history of GDM largely came from studies in Black women (19, 21). Previous studies strongly indicate that African American women with GDM developed T2DM and hypertension more often than White women with GDM (39, 40). This could very well explain the higher incidence of cardiovascular disease in African American women with a history of GDM.

In 3 of the studies indicating an association with kidney disease, the GDM diagnoses were self-reported (19-21). However, in the U.S. CARDIA cohort, self-reported GDM had been validated by review of prenatal glucose tolerance tests for a subsample of 165 women with 200 births, showing a sensitivity of 100%, and a specificity of 92% (41). The Israeli cohort study was based on hospital records from 1988-2013 (20). Data from this study were unfortunately not reported stratified for development of subsequent T2DM.

Breast cancer was the most thoroughly studied possible long-term consequence of GDM, but the results of the 8 available studies were not entirely consistent.

The U.S. case-control studies were conducted in 1990-1992 (23), 1996-2001 (26), and 1999-2004 (25). Data on GDM were self-reported, and given that the data were collected from breast cancer patients and controls from 1990 to 2004, these self-reported data are likely to date back to pregnancies in the 1960s. Some uncertainty must be considered as to whether GDM in pregnant women was investigated systematically at that time; and if this was the case, whether the result was conveyed to the woman; and to what extent recall bias could affect self-reported data at the time of a breast cancer diagnosis many years later. One cohort study from the United States was also based only on self-reported data from 1995 onward (30), while the self-reported data from 1989-2001 in the other U.S. cohort study had been validated against medical records (27). In the Canadian cohort study, GDM diagnoses were retrieved from health administrative records from 1999-2007 (29). In all of these 6 studies, no association was found between a history of GDM and breast cancer.

The 2 Israeli cohort studies were both based on hospital records (24, 28). Perrin et al (2008) (24) used data from the Jerusalem Perinatal Study comprising births from 1964 to 1976. Data on obstetric information were copied from the labor ward log at the time of birth, and the rubric “pre-diabetes” was supposed to correspond, approximately, to GDM. The pregnant women were screened for glucosuria at each antenatal visit and if found positive, they were referred for an oral glucose tolerance test. Like in the Israeli study on kidney diseases (20), Fuchs et al (2017) (28) used hospital records from another Israeli area from 1988 to 2013. Data on GDM came from the perinatal database, but further information on how GDM was defined was not provided.

The differences in time periods for diagnosis of GDM, and the differences in ascertainment of GDM diagnosis could have played a role in the inconsistencies in results across studies. The ICD8 coding system did not have a separate code for GDM. Yet, the awareness of the possible negative consequences for development of diabetes during pregnancy has been present for decades, and years before the oldest study included in this review. Furthermore, the excess risk found in the Jerusalem study was restricted to women diagnosed with breast cancer after the age of 50 years, and this study might be the only one with a sufficiently long follow-up period to cover breast cancer cases diagnosed in postmenopausal age.

Liver Diseases

Based on 2 studies, there was some evidence for a long-term increased risk of liver diseases after a history of GDM. This was consistent with earlier results by Forbes et al (2011) (14), which showed GDM to be associated with a significantly increased short-term risk of NAFLD (OR 2.77 [95% CI, 1.43–5.37]).

However, the elevated long-term risk of liver cirrhosis, failure or need of transplant after a diagnosis of GDM was significantly increased only among women who developed T2DM after GDM, reinforcing the association between DM and liver disease. Obesity and weight gain are risk factors for GDM, T2DM, and liver disease (33). Further investigations are needed in which the possible association between a history of GDM and the long-term risk of liver disease is studied with adjustment for the presence of DM and/or obesity at follow-up.

Liver biopsy is the gold standard diagnostic ascertainment method for NAFLD (32). In the Canadian cohort study, diagnoses of liver diseases were extracted from hospital records (33). However, in the CARDIA study, diagnoses were based on CT scans (32), which might imply a risk of misclassification with non-alcoholic fatty liver disease.

Female Genital Cancer and Urological Cancer

An excess risk of endometrial cancer was indicated in the cohort study based on hospital records from 1988-2013 from an Israeli hospital (28). An excess risk in a U.S.-based case-control study based on hospital records disappeared when the risk assessment was based on imputed data and further adjusted for body mass index (35). Another study also failed to reach significance on increased risk of endometrial cancer, but this could be ascribed missing data (29). The attenuation of the association between GDM and endometrial cancer when controlled for body mass index is expected, as obesity is known to be the strongest nongenetic risk factor for endometrial cancer (35).

Other Morbidity and Mortality

Based on only 1 study, there might be a small long-term risk for thyroid cancer but not other thyroid diseases (29). Since thyroid disease was recently found to be associated with GDM in short-term follow-up studies (5 years postpartum) (42), further investigation of the long-term consequences is needed.

Strengths and Limitations

The strength of this study is the broad scope of diagnoses explored, adding knowledge to the overall long-term morbidity, besides overt diabetes and heart disease, related to a history of GDM.

The limitation of the study is that the included studies differ on important issues such as design, outcome measures, and study period. Some studies are based on the diagnosis of GDM in hospital records, others on self-reports with the inherent risk of recall bias. Some studies included cohorts from private hospitals, others based on national registers. Some studies adjusted for intermediate DM and BMI, others did not. The time span, which the studies covered is broad with different diagnosis systems and differences in how to diagnose. Hence, we refrained from performing a meta-analysis of the reported results.

Further, we restricted our search to morbidity related to specific organs and not morbidity in general. Therefore, long-term morbidity related to, for example, the pancreas and intestine is not covered.

We are aware of the risk of publication bias. However, the reported studies are all based on the hypothesis that GDM might cause long-term health problems. For each study, much work, demanding extra testing and/or extraction of data from existing databases was needed. On this basis, we find it unlikely that researchers have completed this work without reporting the results. By far, the majority of reported results are not statistically significant, indicating that researchers have also reported results not supporting their hypothesis.

Conclusion

The evidence on long-term health consequences after a history of GDM is sparse for the conditions explored in this review. The most consistent pattern for an association between GDM and long-term disease was found for kidney diseases, especially in Black women, but in 3 of the 4 studies, the diagnosis of GDM was self-reported, and the studies used different outcome measures for kidney disease. Studies based on self-reported GDM diagnosis, some published as far back as the 1960s, dominated the evidence base for the association between GDM and breast cancer; and some diagnoses might therefore have been misclassified. Overall, the findings in this systematic review did not provide firm evidence for associations between GDM and long-term excess risks of the studied diseases.

Given the complex associations between GDM, development of different kinds of diseases after birth, and long-term health outcomes, future studies based on comprehensive hospital records or well-defined cohorts are strongly recommended.