Published online Oct 15, 2026. doi: 10.4239/wjd.124799
Revised: July 17, 2026
Accepted: August 28, 2026
Published online: October 15, 2026
Processing time: 110 Days and 11.8 Hours
Glucagon-like peptide-1 receptor agonists (GLP-1RAs) and related incretin-based dual agonists are increasingly used by women of reproductive age for type 2 diabetes, obesity, polycystic ovary syndrome, and infertility-related metabolic dysfunction. This expanding use has created a clinical dilemma about how to counsel women during periconception or inadvertent early-pregnancy exposure.
To map and critically synthesize the available human evidence on maternal, pregnancy, fetal, neonatal, and offspring outcomes following periconceptional or early-pregnancy GLP-1RA exposure.
We conducted a scoping review guided by the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews, using PubMed and Google Scholar from January 1, 2000 to June 10, 2026. Evidence was organized by study design and inferential strength, including case reports; case series; pharmacovigilance analyses; regulatory clinical-trial pregnancy datasets; comparative cohort, registry, and controlled studies; and systematic reviews with meta-analyses. Systematic reviews without meta-analysis were retained only for contextualization and reference cross-checking. A Mea
Forty-seven reports were included, comprising 41 core evidence reports and 6 contextual systematic reviews without meta-analysis. Core evidence included 17 case reports, 3 case series, 1 pharmacovigilance study, 1 regulatory clinical-trial pregnancy dataset, 16 cohort studies, and 3 systematic reviews with meta-analyses. Across comparative cohorts and meta-analyses, inadvertent preconceptional or early-pregnancy GLP-1RA exposure was not associated with a consistent increase in overall major congenital malformations. However, signals for spontaneous abortion, elective termination, preeclampsia, preterm birth, altered gestational weight gain, fetal growth abnormalities, and renal/genitourinary anomalies were inconsistently reported. Interpretation was limited by confounding from type 2 diabetes, obesity, polycystic ovary syndrome, infertility, hypertension, nephropathy, comparator choice, exposure-window heterogeneity, and prescription-based exposure misclassification. Evidence was strongest for cautious reassurance after inadvertent exposure and weakest for planned or continued pregnancy use.
Current human evidence does not indicate a reproducible teratogenic signal after inadvertent preconceptional or early-pregnancy GLP-1RA exposure. Nevertheless, available data remain observational, heterogeneous, and confounded. GLP-1RAs should be discontinued when pregnancy is recognized, and women should be transitioned to pregnancy-compatible metabolic care. Planned or continued GLP-1RA use during pregnancy is not supported.
Core Tip: This scoping review maps the full human evidence spectrum on glucagon-like peptide-1 receptor agonist exposure before conception and during early pregnancy, from case reports to meta-analyses. The most important finding is cautiously reassuring: no consistent overall signal of major congenital malformations has emerged after inadvertent early exposure. However, pregnancy loss, termination, hypertensive disorders, preterm birth, fetal growth, and renal/genitourinary signals remain uncertain and highly confounded. Clinicians should reassure women after accidental exposure, stop glucagon-like peptide-1 receptor agonists once pregnancy is recognized, and avoid planned or continued use during pregnancy until stronger prospective evidence is available.
- Citation: Singh R, Singh A, Singh AK. Human pregnancy outcomes after early glucagon-like peptide-1 receptor agonist exposure: A scoping review. World J Diabetes 2026; 17(10): 124799
- URL: https://www.wjgnet.com/1948-9358/full/v17/i10/124799.htm
- DOI: https://dx.doi.org/10.4239/wjd.124799
Glucagon-like peptide-1 receptor agonists (GLP-1RAs) have become central therapies for type 2 diabetes (T2D) and obesity, and their clinical use is rapidly expanding into broader cardiometabolic and reproductive-metabolic contexts. Their benefits include glycemic improvement, weight reduction, appetite regulation, reduced insulin resistance, and potential improvement in metabolic complications associated with obesity and polycystic ovary syndrome (PCOS). As use increases among women of reproductive age, inadvertent exposure before conception or during early pregnancy has become increasingly likely. The question of pregnancy safety is clinically important and methodologically complex. Women receiving GLP-1RAs frequently have obesity, T2D, PCOS, infertility, hypertension, dyslipidemia, renal disease, advanced maternal age, previous pregnancy loss, or other metabolic risk factors. These conditions independently increase the risk of miscarriage, congenital anomalies, gestational diabetes mellitus (GDM), hypertensive disorders of pregnancy (HDP), preterm birth, stillbirth, fetal growth abnormalities, neonatal hypoglycemia, neonatal intensive care unit (NICU) admission, and cesarean delivery (CD). Therefore, adverse pregnancy outcomes after GLP-1RA exposure may reflect underlying maternal risk rather than a direct drug effect. Exposure timing further complicates interpretation. “Preconceptional exposure,” “periconceptional exposure,” “first-trimester exposure,” and “continued pregnancy exposure” repre
Current clinical recommendations advise avoiding GLP-1RAs during pregnancy. These recommendations are based on concerns about animal reproductive toxicity, the lack of definitive human safety data, and the precautionary principle, rather than on clear evidence of human teratogenicity. However, the number of human reports has increased substantially. Evidence now includes isolated case reports, case series, pharmacovigilance analyses, regulatory clinical-trial pregnancy data, teratology information service cohorts, electronic health record (EHR) studies, population registries, and systematic reviews and meta-analyses. These types of evidence do not carry equal inferential weight. Case reports and case series provide clinical detail but no denominator. Pharmacovigilance can identify disproportionate reporting signals but cannot estimate incidence or causality. Regulatory clinical-trial pregnancy data provide structured but post hoc information from accidental pregnancies occurring in trials where pregnancy was an exclusion criterion. Comparative cohorts and registries provide the strongest primary evidence for risk estimation but are vulnerable to confounding by indication, disease severity, exposure misclassification, and comparator choice. Systematic reviews and meta-analyses provide a pooled synthesis but inherit limitations due to a small number of events, the absence of a balanced comparator arm, and overlap among the primary studies.
The purpose of this scoping review was to map the available human evidence on GLP-1RA exposure before conception and during early pregnancy, organize the evidence by study design and clinical indication, evaluate methodological quality using appropriate appraisal tools, and provide a clinically meaningful interpretation to inform counseling of women after inadvertent exposure or before planned conception.
This review was designed as a scoping review because the evidence base is broad, evolving, and methodologically heterogeneous. The review includes multiple types of human evidence: Case reports, case series, pharmacovigilance analyses, regulatory clinical-trial pregnancy datasets, retrospective and prospective cohort studies, registry studies, controlled preconception/fertility studies reporting pregnancy outcomes, and systematic reviews and meta-analyses. The review was conducted using a predefined methodology and reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses extension for Scoping Reviews (PRISMA-ScR)[1]. The completed checklist is provided as Supplementary material. The protocol was not prospectively registered. Because the work was designed as a scoping review intended to map the breadth and structure of human evidence rather than to generate a pooled effect estimate, prospective registration was not completed. We acknowledge that prospective registration on platforms such as the Open Science Framework would have further strengthened transparency and reduced the possibility of selective reporting. To mitigate this limitation, we report the eligibility criteria, database-specific search strategy, data-charting framework, PRISMA-ScR flow, and critical-appraisal approach in full. A scoping approach was preferred over a de novo meta-analysis because the available literature varies substantially in exposure window, indication, comparator group, outcome definition, data source, and risk of bias.
The review addressed the following question: What human evidence is available regarding maternal, pregnancy, fetal, neonatal, and offspring outcomes after GLP-1RA exposure before conception, around conception, or during pregnancy? The review question was structured using the Population-Concept-Context framework rather than the Population-Intervention-Comparator-Outcome approach because the objective was to map the breadth and nature of available human evidence, not to estimate the effect of a single intervention against a uniform comparator. The included literature revealed heterogeneous designs, including case reports, case series, pharmacovigilance analyses, regulatory clinical-trial pregnancy data, comparative cohort studies, and systematic reviews with or without meta-analyses. Many reports had no comparator group, while exposure windows and outcomes varied substantially. Therefore, the Population-Concept-Context framework was more appropriate for defining the scope of evidence: The population comprised women with pregnancy outcomes after GLP-1RA exposure; the concept comprised preconceptional, periconceptional, or early-pregnancy GLP-1RA exposure and associated maternal-fetal outcomes; and the context comprised human clinical evidence across diabetes, obesity, PCOS/infertility, pharmacovigilance, regulatory, registry, and cohort settings.
Studies were eligible if they reported human pregnancy-related outcomes after exposure to a GLP-1RA or related incretin-based therapy, including exenatide, liraglutide, lixisenatide, albiglutide, dulaglutide, semaglutide, or tirzepatide. Only English-language reports were eligible. Eligible exposure windows included preconceptional use, periconceptional use, first-trimester exposure, or exposure during pregnancy. Although the review focused principally on preconceptional and early-pregnancy exposure, reports of continued later-pregnancy exposure were retained to map the full extent of available human evidence. Eligible evidence types included case reports, case series, pharmacovigilance analyses, regulatory clinical-trial pregnancy datasets, retrospective cohort studies, prospective cohort studies, registry studies, controlled preconception or fertility studies reporting pregnancy outcomes, and systematic reviews and meta-analyses of human pregnancy outcomes. Studies were excluded from the primary evidence synthesis if they were animal-only studies, in vitro or mechanistic studies without human pregnancy outcomes, narrative commentaries without extractable human outcome data, editorials, viewpoints, abstracts without sufficient extractable data, or reports in which GLP-1RA exposure could not be linked to pregnancy outcomes. Systematic reviews without meta-analysis were not used as primary evidence because they did not contribute independent patient-level or cohort-level data and overlapped with the primary studies already extracted. They were retained only for background, contextualization, and reference cross-checking.
PubMed and Google Scholar were searched from January 1, 2000 through June 10, 2026, the final date of search completion. Reference lists of included studies and relevant reviews were manually searched to identify additional eligible human reports. Records were eligible if they were available online on or before this date. Studies carrying a 2026 publication year were included only if they were accessible within the final search window. PubMed identified 140 records. Google Scholar retrieved approximately 27700 records. Because Google Scholar ranks results by relevance and does not provide stable database-style indexing, the first 500 relevance-ranked records were screened. This threshold was selected pragmatically because it exceeded the PubMed yield of 140 records by more than 3-fold, while remaining feasible for manual screening. The aim was to improve sensitivity for ahead-of-print articles, citation-linked reports, gray literature, and records not retrieved through PubMed. Thus, 640 records were identified for screening from PubMed and Google Scholar. After manual deduplication of 224 records, 416 records underwent title and abstract screening. Of these, 359 were excluded. Fifty-seven full-text reports were assessed for eligibility, and 10 were excluded. Forty-seven reports were retained in the final evidence map. Of these, 41 were included in the core evidence synthesis, and 6 systematic reviews without meta-analysis were retained only for contextualization and reference cross-checking. Figure 1 summarizes the PRISMA-ScR flow diagram for study selection. The exact PubMed and Google Scholar keyword strings, Boolean operators, search date, and screening depth are provided in Supplementary Table 1.
Titles and abstracts were screened for relevance to GLP-1RA exposure before or during pregnancy. Full texts were then assessed against eligibility criteria. When multiple reports used overlapping datasets, the most complete or most recent report was prioritized for outcome interpretation, while related reports were retained only if they provided unique evidence. Data charting was performed by 1 author and independently verified by the other 2 authors. Disagreements regarding inclusion were resolved by consensus. Data were extracted into a standardized evidence-charting form. Extracted variables included author, year, country, study design, data source, clinical population, indication for GLP-1RA use, GLP-1RA agent, exposed sample size, comparator group where applicable, exposure window, maternal outcomes, fetal/neonatal outcomes, congenital malformation outcomes, pregnancy-loss outcomes, follow-up duration, main findings, strengths, limitations, and interpretation. Evidence was grouped according to both design and clinical indication. Primary evidence categories included case reports/series, pharmacovigilance, regulatory clinical-trial pregnancy data, and cohort, registry, and controlled studies. The latter were further stratified into diabetes-specific cohorts; obesity-, PCOS-, or fertility-focused cohorts; and mixed diabetes/obesity cohorts. Systematic reviews and meta-analyses were considered separately.
Critical appraisal was performed to contextualize the strength of evidence and guide interpretation. Because this was a scoping review, studies were not excluded solely on the basis of quality or risk-of-bias assessment. Case reports and case series were assessed using the respective Joanna Briggs Institute (JBI) critical appraisal checklists, comprising 8 and 10 domains, respectively[2]. Comparative cohort, registry, and controlled observational studies were assessed using the JBI cohort checklist[2], which included 11 domains. The pharmacovigilance study and the regulatory clinical-trial pregnancy dataset were appraised descriptively using design-appropriate domains because conventional cohort checklists do not fully apply. Systematic reviews and meta-analyses were assessed using all 16 items of A Measurement Tool to Assess Systematic Reviews-2 (AMSTAR-2)[3], covering the 7 core critical domains, without generating numerical scores.
A de novo meta-analysis was not performed because the studies varied substantially in design, exposure timing, GLP-1RA agent, indication, comparator group, outcome definition, and reporting structure. Instead, the findings were synthesized narratively and organized by inferential strength. Case reports and case series were treated as descriptive signal-generating evidence. Pharmacovigilance was interpreted as signal-detection evidence rather than risk-estimation evidence. Regulatory clinical-trial pregnancy data were considered an intermediate evidence category because they provided structured but post hoc information on accidental pregnancies in trials where pregnancy was an exclusion criterion. Comparative cohort, registry, and controlled studies were treated as the principal primary evidence for risk estimation, whereas non-comparative cohorts were interpreted descriptively. Systematic reviews and meta-analyses were treated as secondary evidence syntheses. The synthesis focused on whether consistent patterns emerged across evidence layers for major congenital malformations (MCMs), organ-specific anomalies, pregnancy loss, elective termination, stillbirth, preterm birth, HDP, GDM, gestational weight gain (GWG), fetal growth, neonatal complications, and long-term offspring outcomes. Particular attention was paid to comparator choice, exposure-window definition, GLP-1RA agent specificity, and confounding by obesity, diabetes severity, PCOS, infertility, hypertension, nephropathy, and other baseline maternal risk factors.
Forty-seven reports were included in the final evidence map. Forty-one reports contributed to the core evidence synthesis: 17 case reports[4-20]; 3 case series[21-23]; 1 pharmacovigilance study[24]; 1 regulatory clinical-trial pregnancy dataset[25]; 16 cohort, registry, and controlled studies[26-41]; and 3 systematic reviews and meta-analyses[42-44]. Six systematic reviews without meta-analysis were retained solely for contextual discussion and cross-reference checking[45-50]. The included evidence reflects a clear evolution in the field. Early evidence consisted of isolated case reports and small case series describing inadvertent exposure. Pharmacovigilance then provided signal-detection data. Regulatory clinical-trial pregnancy data contributed structured information from accidental pregnancies during drug-development programs. Larger population-based, teratology information service, EHR, and registry studies then provided comparative risk estimates. Finally, 3 systematic reviews and meta-analyses synthesized overlapping observational evidence.
Published case reports and case series described exposure to exenatide, liraglutide, dulaglutide, and semaglutide before conception or during pregnancy[4-23]. Most exposures were inadvertent and were discontinued after pregnancy was recognized, usually in the first trimester. A smaller number described longer exposure, including continuation into later pregnancy or lactation. The clinical backgrounds were generally high risk. Women commonly had T2D, obesity or morbid obesity, PCOS, infertility, hypertension, dyslipidemia, thyroid disease, previous pregnancy loss, advanced maternal age, obstructive sleep apnea, retinopathy, microalbuminuria, or poor glycemic control. These baseline features are important because they independently increase the probability of adverse pregnancy and neonatal outcomes.
Most reported pregnancies resulted in live birth, and no reproducible congenital malformation syndrome emerged. However, adverse outcomes were frequently reported and were heterogeneous. Maternal events included HDP, GDM, excessive GWG, preterm delivery, severe infection, and diabetic ketoacidosis in the setting of intercurrent illness. Fetal and neonatal outcomes included transient neonatal hypoglycemia, macrosomia, shoulder dystocia, respiratory distress syndrome, fetal growth restriction, low birth weight, NICU admission, polyhydramnios, mild urinary tract findings, atrial septal defect, micropenis, mild foot deformity, and isolated cardiac anomaly. However, the central limitation of these reports is the absence of a denominator and comparator. They cannot distinguish background metabolic risk from drug-associated risk. They also cannot estimate absolute risk, relative risk (RR), or causality. Follow-up was often short or incompletely reported, limiting interpretation of neurodevelopmental and long-term metabolic outcomes.
Nevertheless, these reports are clinically useful. They suggest that inadvertent early exposure does not invariably result in adverse fetal outcomes and has not produced a consistent pattern of malformations. They also support cautious reassurance after inadvertent exposure while emphasizing the limitations of uncontrolled evidence (Tables 1 and 2).
| Ref. | Country | GLP-1RA/dose | Exposure window/cessation timing | Maternal background/metabolic profile | Delivery/birth weight | Maternal outcome | Fetal/neonatal outcome | Follow-up | Interpretation |
| Williams et al[4], 2009 | United States | Exenatide, dose NR | 7 months preconception to 14 GW | 35 years; T2D, HTN, obesity, proliferative retinopathy, microalbuminuria; HbA1c 10.4% | LUCS, 38 weeks; 2835 g | Preeclampsia, intra-amniotic infection, postpartum uterine infection | Low Apgar, mild encephalopathy, neonatal sepsis | 3 months, uneventful | High-risk diabetes background; adverse neonatal outcome not clearly attributable to exenatide |
| Greco[5], 2015 | Italy | Liraglutide 1.8 mg/day | 2 years preconception to 12 GW | 37 years; T2D for 5 years, obesity, PCOS, infertility; BMI: 32.2 kg/m2 | LUCS, 37 weeks; 3790 g | Uneventful | Normal Apgar score; transient neonatal hypoglycemia | NR | Reassuring early liraglutide exposure; neonatal hypoglycemia likely diabetes-related |
| Ivanišević et al[6], 2018 | Croatia | Liraglutide 1.8 mg/day | Preconception to throughout pregnancy and lactation | 28 years; thyroidectomy for diffuse goiter/thyroid disease; uncontrolled T2D | LUCS, 39 weeks; 4220 g | Uneventful | Apgar 10/10; macrosomia | 2 years, uneventful | Unique prolonged exposure/Lactation case; macrosomia likely related to diabetes |
| Livadas et al[7], 2020 | Greece | Liraglutide 1.8 mg/day | 14 weeks preconception to 6-7 GW | 29 years; obesity with HAIR-AN syndrome, hirsutism, insulin resistance, acanthosis nigricans, severe PCOS | NR | Uneventful | Healthy baby | Uneventful | Reassuring inadvertent early exposure in severe PCOS/insulin-resistance phenotype |
| Burlina et al[8], 2023 | Italy | Dulaglutide 1.5 mg/week | 6 weeks preconception to 15 GW | 38 years; T2D for 11 years, obesity | LUCS, 37 weeks; 2860 g | Gestational hypertension; 15 kg weight gain | Cavum septum pellucidum at 20 weeks, resolved; neonatal jaundice | NR | No persistent structural anomaly; maternal metabolic risk remains major confounder |
| Alghamdi et al[9], 2023 | Saudi Arabia | Dulaglutide 1.5 mg/week | Preconception to 13 GW | 39 years; T2D for 19 years, obesity | LUCS, 38 weeks; 3100 g | Uneventful | Mild bilateral pyelectasis; Down syndrome excluded | NR; unplanned pregnancy | Mild urinary finding; no clear teratogenic pattern |
| Skov et al[10], 2023 | Denmark | Semaglutide 1 mg/week | 7 months preconception to 4 GW | 31 years; PCOS, uncomplicated myxedema | Vaginal delivery, 41 weeks + 5 days; 5230 g | 35 kg gestational weight gain | Macrosomia, shoulder dystocia, neonatal hypoglycemia | 6 months | Reassuring for malformation; macrosomia, shoulder dystocia, and neonatal hypoglycemia occurred in the setting of marked gestational weight gain and maternal metabolic risk |
| Wajtryt and Dedecjus[11], 2023 | Poland | Semaglutide, dose NR | 5 months preconception to 7 GW | 43 years; T2D; G5P4 | LUCS at term; 2750 g | Uneventful | None | NR | Reassuring but limited by sparse details |
| Patel and Healy[12], 2024 | United States | Dulaglutide 0.75 mg/week | Preconception to 15 GW | 41 years; T2D, PCOS, obesity | LUCS, 36 weeks; birth weight NR | Preeclampsia, preterm delivery | Polyhydramnios at 26 weeks, corrected by 32 weeks | NR | Maternal and fetal findings consistent with high-risk metabolic pregnancy |
| Molteni et al[13], 2024 | Italy | Dulaglutide 1.5 mg/week | 3 months preconception to 34 GW; continued during pregnancy | 42 years; G6P5; T2D for 11 years; obesity, dyslipidemia | LUCS, 38 weeks, breech; 3790 g | 5 kg weight gain; gestational hypertension | Polyhydramnios; fetal growth normal | NR | Important prolonged dulaglutide exposure; no malformation, but maternal HTN/polyhydramnios occurred |
| Okeke et al[14], 2024 | United States | Semaglutide 2 mg, single self-administered dose | Before pregnancy recognition; pregnancy detected at 6-7 GW | 34 years; G5P4; generalized anxiety, borderline personality, off-label weight loss, one previous preterm birth, three previous abortions | NR | NR | Hyperemesis gravidarum reported | NR | Limited interpretability; single-dose exposure and major psychiatric/obstetric background |
| Zhu et al[15], 2024 | China | Semaglutide, dose NR | Before pregnancy to 6 GW | 34 years; obesity | LUCS, 37 weeks; birth weight NR | NR | Micropenis, mild foot deformity | 6 months, normal growth | Isolated anomaly report; no causal inference possible |
| Abushanab et al[16], 2025 | Qatar | Liraglutide 3 mg/day | 8 months preconception to 7 GW | 29 years; G9P6; obesity, BMI 36 kg/m2; two previous preterm births, two abortions, four live births | LUCS, 37 weeks; 2330 g | Maternal weight gain 51 kg, from 80 kg to 131 kg | Low birth weight; Apgar 9/10 | 2 years, normal growth and neurodevelopment | Reassuring developmental follow-up; maternal weight rebound notable |
| Burlina et al[17], 2026 | Italy | Dulaglutide, dose NR | Preconception to 23 GW | 44 years; uncontrolled T2D | Vaginal delivery, birth weight | Maternal weight gain 9 kg | No obstetric complications or congenital anomalies reported | NR | Prolonged exposure; reassuring but no follow-up |
| Huang et al[18], 2025 | United States | Dulaglutide, dose NR | First-trimester exposure; exact GW NR | 35 years; T2D, HTN, hyperlipidemia | NR | Hospitalized for septic arthritis of left shoulder with euglycemic DKA | Elective termination | NR | Termination and severe maternal illness confound fetal inference |
| Ozdemir[19], 2025 | Greece | Liraglutide | Preconception to 3.5 GW | 20 years; obesity | LUCS, 34 weeks; 2700 g | Developed GDM | NICU for 1 week | 6 months, normal development | Early exposure; prematurity/GDM not clearly drug-related |
| Jin et al[20], 2026 | China | Semaglutide 0.5 mg/week, off-label obesity use | 6 months preconception to 7 GW | 37 years; overweight, BMI: 27.73 kg/m2; chronic simple lichen planus; hypothyroidism | Vaginal delivery, 39.6 weeks; | 22 kg gestational weight gain; no GDM; no HTN or preeclampsia | Apgar 8/10 at 1 min, 10/10 at 5 min; fetal growth, NT and Doppler normal; healthy baby | 3 months, uneventful | Reassuring early semaglutide exposure; GDM/weight gain remain important |
| Ref. | Country | GLP-1RA/dose | Exposure window/cessation timing | Maternal background/metabolic profile | Delivery/birth weight | Maternal outcome | Fetal/neonatal outcome | Follow-up | Interpretation |
| Doğan et al[21], 2024 (case 1) | Turkey | Exenatide, dose NR | 13-17 GW for approximately 4 weeks | 40 years; morbid obesity, 139 kg; PCOS and T2D; BMI: > 35 kg/m2 | First pregnancy: 37 weeks; delivery mode NR | Uneventful | ASD (secundum): 4 mm | ASD closed spontaneously by 3 years | Isolated ASD; high-risk metabolic substrate; later spontaneous closure |
| Doğan et al[21], 2024 (case 2) | Turkey | Exenatide, dose NR | 3 months preconception to 13 GW | Same patient | Second pregnancy: 37.5 weeks | Uneventful | No congenital anomaly | 5 years, uneventful | Same mother later had unaffected pregnancy; argues against reproducible teratogenic pattern |
| Banerjee and Dasgupta[22], 2026 (case 1) | India | Dulaglutide 1.5 mg/week | 7 months preconception to 6-7 GW until pregnancy confirmation | 26 years; PCOS, morbid obesity, OSA, hypothyroidism, T2D, MASLD; BMI: 47.9 kg/m2 | NR; 2900 g | Uneventful | Healthy baby | NR | Reassuring early dulaglutide exposure in high-risk metabolic background |
| Banerjee and Dasgupta[22], 2026 (case 2) | India | Oral semaglutide 7 mg/day | 12 months preconception to 6 GW until pregnancy confirmation | 28 years; PCOS, obesity, T2D, infertility; BMI: 36.25 kg/m2 | NR; 2600 g | Uneventful | Healthy baby | NR | Reassuring early semaglutide exposure; sparse follow-up |
| Morton and He[23], 2026 (case 1) | Australia | Semaglutide, dose NR | Preconception to 4 GW | 33 years; G3P2; T2D and obesity; BMI: 35 kg/m2 | LUCS, 38.3 weeks; 3910 g | Uneventful | Healthy baby | NR | Reassuring |
| Morton and He[23], 2026 (case 2) | Australia | Semaglutide, dose NR | Preconception to 4 GW | 35 years; G1P0; T2D and obesity, BMI: 37 kg/m2; HTN, smoker, idiopathic intracranial hypertension | 38.1 weeks; 3288 g | Uneventful | Cardiac anomaly, not specified | NR | Isolated unspecified cardiac anomaly with multiple maternal confounders |
| Morton and He[23], 2026 (case 3) | Australia | Semaglutide, dose NR | Preconception to 6 GW | 37 years; G4P3; T2D, morbid obesity, BMI: 43 kg/m2; HTN | 38.4 weeks; 3484 g | Uneventful | Healthy baby | NR | Reassuring |
| Morton and He[23], 2026 (case 4) | Australia | Semaglutide, dose NR | Preconception to 7 GW | 31 years; G6P5; T2D and obesity, BMI: 40 kg/m2; Addison disease, OSA, previous neonatal death | LUCS, 35.6 weeks; 4800 g | Preterm delivery | Macrosomia, polyhydramnios, RDS | NR | Adverse metabolic/neonatal outcomes in very high-risk pregnancy |
| Morton and He[23], 2026 (case 5) | Australia | Semaglutide, dose NR | Preconception to 6 GW | 35 years; G3P2; T2D and obesity, BMI: 30 kg/m2 | Vaginal delivery, 31.6 weeks; 1035 g | Preeclampsia, preterm delivery | Fetal growth restriction | NR | Severe obstetric outcome; maternal diabetes/obesity confounding substantial |
| Morton and He[23], 2026 (case 6) | Australia | Semaglutide, dose NR | Preconception to 6 GW | 31 years; G3P2; T2D, obesity, BMI: 40 kg/m2; PCOS | LUCS, 38.2 weeks; 3950 g | Uneventful | Macrosomia | NR | Macrosomia likely metabolic background-related |
| Morton and He[23], 2026 (case 7) | Australia | Semaglutide, dose NR | Preconception to 6 GW | 35 years; G1P0; T2D, obesity, BMI: 30 kg/m2 | 36.4 weeks; 3360 g | Preterm delivery | Healthy baby | NR | Preterm but otherwise reassuring |
| Morton and He[23], 2026 (case 8) | Australia | Semaglutide, dose NR | Preconception to 6 GW | 39 years; G3P2; T2D, overweight, BMI: 29 kg/m2 | 34.5 weeks; 3896 g | Preterm delivery | Fetal distress, RDS | NR | Adverse neonatal respiratory outcome; background T2D and prematurity confound interpretation |
| Morton and He[23], 2026 (case 9) | Australia | Semaglutide, dose NR | Preconception to 6 GW | 27 years; G2P1; T2D, overweight, BMI: 29 kg/m2 | 36.5 weeks; 3080 g | Preterm delivery | Healthy baby | NR | Preterm but no anomaly |
| Morton and He[23], 2026 (case 10) | Australia | Semaglutide, dose NR | Preconception to 6 GW | 26 years; G1P0; T2D, overweight, BMI: 26 kg/m2 | 38.1 weeks; 3348 g | Uneventful | Healthy baby | NR | Reassuring |
| Morton and He[23], 2026 (case 11) | Australia | Semaglutide, dose NR | Preconception to 4 GW | 34 years; G1P0; T2D, obesity, BMI: 40 kg/m2 | 40 weeks; 3462 g | Uneventful | Healthy baby | NR | Reassuring |
| Morton and He[23], 2026 (case 12) | Australia | Semaglutide, dose NR | Preconception to 2 GW | 33 years; G1P0; T2D, obesity, BMI: 30 kg/m2 | 38.3 weeks; 3852 g | Uneventful | Macrosomia | NR | Macrosomia likely diabetes/obesity-related |
| Morton and He[23], 2026 (case 13) | Australia | Semaglutide, dose NR | Preconception to 6 GW | 35 years; G2P1; obesity, non-T2D | LUCS, 38.5 weeks; 3930 g | Uneventful | Macrosomia | NR | Reassuring for malformation; fetal size remains a metabolic issue |
One dedicated pharmacovigilance analysis evaluated GLP-1RA exposure in pregnancy using the United States Food and Drug Administration Adverse Event Reporting System (FAERS)[24]. This study identified 354 pregnancy-related cases comprising 1671 adverse drug reaction reports. Reports increased over time, consistent with expanding GLP-1RA use among women of reproductive age. The most frequently reported agents were liraglutide, semaglutide, and dulaglutide. At the system-organ-class level, reproductive and gastrointestinal events were prominent. Initial disproportionality analysis detected signals for pregnancy, puerperium, and perinatal conditions; congenital, familial, and genetic disorders; and gastrointestinal disorders. However, when the comparator was restricted to other pregnant women in FAERS, the congenital/familial/genetic disorder signal became negative. Pregnancy-related and gastrointestinal signals persisted. At the preferred-term level, spontaneous abortion, preeclampsia, and imminent abortion were the most notable signals. These findings are clinically important because they identify outcomes requiring careful monitoring in future com
The strengths of pharmacovigilance include broad real-world coverage, inclusion of multiple GLP-1RA agents, and the ability to detect early safety signals. The limitations are fundamental: Spontaneous reporting systems lack a true denominator, cannot estimate absolute risk or RR, are vulnerable to underreporting and stimulated reporting, may include duplicate reports, often lack precise exposure timing and dose, and cannot adjust adequately for obesity, diabetes severity, PCOS, infertility, hypertension, or concomitant medications. Therefore, pharmacovigilance should be inter
One study extracted pregnancy outcomes from regulatory clinical-trial documents submitted to regulatory agencies during GLP-1RA development programs[25]. This dataset is methodologically distinct from conventional cohorts. Pre
This evidence is valuable because it makes visible pregnancy outcomes that are often buried in regulatory documents rather than published trial reports. It is more structured than isolated case reports and less vulnerable to spontaneous-reporting bias than pharmacovigilance. However, it is not equivalent to a conventional comparative cohort. Pregnancy was not a planned exposure, data were post hoc, outcome reporting was incomplete, exposure timing and maternal characteristics were insufficiently detailed, and adjusted comparisons were not possible. The regulatory dataset should therefore be positioned as intermediate evidence: Stronger than isolated case reports for descriptive outcome capture, but weaker than comparator-based cohort studies for risk estimation. Table 3 summarizes the results from both pharmacovigilance and regulatory studies.
| Ref. | Evidence type | Data source | GLP-1RA exposure | Comparator | Exposure window/cessation timing | Maternal outcomes | Fetal/neonatal outcomes | Strength | Limitation | Interpretation |
| Zhou et al[24], 2025 | Pharmacovigilance signal-detection study | FAERS 19-year data | 354 pregnancy-related cases comprising 1671 ADR reports; liraglutide: 150; semaglutide: 106; dulaglutide: 52 | FAERS comparator; sensitivity analysis using other pregnant women in FAERS | Any time during pregnancy | ROR025 signals-spontaneous abortion: 32.24, preeclampsia: 5.06, Imminent abortion: 1.40, stillbirth: 1.15 | No MCM signal after sensitivity analysis using pregnant comparator | Broad real-world surveillance; multiple GLP-1RAs; used ROR and Bayesian IC; pregnancy-specific sensitivity analysis | No denominator; cannot estimate incidence or causality; reporting bias, duplicate reports, missing exposure timing, confounding by obesity/T2D/PCOS/HTN. No maternal baseline metabolic data available | Signal for spontaneous abortion and preeclampsia; hypothesis-generating, not risk-estimating |
| Parker et al[25], 2025 | Regulatory clinical-trial inadvertent pregnancy dataset | FDA/EMA regulatory trial documents before approval | 112 GLP-1RA-exposed pregnancies: Exenatide 4, lixisenatide 3, liraglutide 44, dulaglutide 7, albiglutide 6, semaglutide 41, tirzepatide 7 | 38 placebo/non-GLP-1RA comparator pregnancies | From pre-pregnancy reporting of incidental pregnancy and follow-up to delivery | 1 preeclampsia; 1 oligohydramnios; 1 low birth weight; 1 mild gestational hypertension/cholestasis; 48 healthy live births; 24 spontaneous abortions; 3 ectopic pregnancies; 14 elective terminations; 13% elective termination in both exposed and unexposed | Healthy live birth: 43% vs 34%; MCM: 1.8% vs 5.3%; stillbirth: 0.9% vs 5.3%; spontaneous abortion: 22% vs 13%; ectopic pregnancy: 2.7% vs 7.9%; nonviable pregnancies: 25.2% exposed vs 26.3% unexposed; P = NR for any outcomes | Structured regulatory clinical trial pregnancy data; class-wide exposure across several GLP-1RAs; some comparator pregnancies available | Pregnancy was an exclusion criterion; post hoc extraction; small numbers; incomplete outcomes; no adjustment for maternal baseline risk; not a conventional cohort | Reassuring against a large teratogenic signal, but should be interpreted as intermediate evidence, not definitive safety evidence |
Sixteen cohort, registry, and controlled human studies were identified[26-41]. Fifteen included a comparator or controlled design and provided the principal evidence for risk estimation; one small prospective cohort, Bolek et al[30], had no comparator and contributed only descriptive evidence. The direction of reported outcomes across the 15 comparative/controlled studies is summarized in Figure 2, while detailed findings from all 16 studies are presented in Tables 4, 5, and 6. Interpretation varied substantially by clinical indication, comparator group, exposure window, and adjustment for confounding.
| Ref. | Country/design | Exposed population | GLP-1RA | Exposure window/cessation timing | Comparator | Maternal outcomes | Fetal/neonatal outcomes | Strength | Limitation | Interpretation |
| Chou et al[26], 2026 | Taiwan (China); T2D cohort | 160 T2D; 19 obese among exposed | Not specified | 90 days before through 90 days after LMP | 606 PM T2D cohort on insulin | Preterm 34.6% vs 31.9%; P = NS | MCM: 0.6% vs 0.2%; stillbirth: 4.4% vs 2.1%; SGA: 2.6% vs 3%; all P = NS | Diabetes-specific active comparator; PM T2D cohort | Small exposed sample; agent unspecified; limited power for rare anomalies | Reassuring; risk not greater than insulin-exposed T2D pregnancies |
| Cesta et al[27], 2024 | Nordic countries/United States/Israel; T2D cohort | 938 T2D; 483 obese among exposed | Not specified | 180-365 days before LMP to end of first trimester (90-97 days) | 5078 T2D on insulin | NR | MCM: 8.2% vs 7.8%; cardiac anomalies: 3.2% vs 4.2%; all P = NS. First-trimester GLP-1RA: MCM: 6.7% vs 7.8%; cardiac anomalies: 1.6% vs 4.2%. All P = NS | Large multinational population-based dataset; clinically relevant diabetic comparator | Prescription-based exposure; limited glycemic/BMI details; maternal outcomes not central | Reassuring for major teratogenicity compared with insulin-treated diabetes |
| Hanif et al[28], 2025 | United States; T2D cohort | 1826 T2D | Not specified | 1 year before to 1 month after first trimester pregnancy diagnosis | 1826 T2D unexposed to GLP-1RA (mostly LSM and/or insulin) | Maternal all-cause mortality, gestational HTN, preeclampsia/eclampsia: All P = NS | Cardiac anomaly, kidney anomaly: All P = NS | Large PM T2D cohort | Fetal assessment limited to cardiac and kidney anomalies. Agent unspecified; broad exposure window; residual confounding | Reassuring; no increased risk compared with T2D controls |
| Lewin et al[29], 2026 | United States; T2D cohort, 3-year data | 35083 DM; (T2D: 34343; T1D: 740) | Not specified | Any time during pregnancy for any duration | 108510 DM; (T2D: 98628; T1D: 9882) not on GLP-1RA | PET 13.1% vs 11.9%, P < 0.0001; preterm 23.1% vs 20.8%, P < 0.0001; CD: 44.4% vs 38.9%, P < 0.0001 | First-trimester abortion/termination 9.8% vs 7.5%, P < 0.0001; MCM: 3.3% vs 3.3%, P = 0.63; genital malformations: 0.7% vs 0.6%, P = 0.02; urinary malformations: 0.5% vs 0.4%, P = 0.01 | Very large sample; able to detect rare system-specific signals | More severe baseline cohort; possible T1D admixture; strong confounding by disease severity; broad exposure definition | Reassuring for overall MCM but signals for PET, preterm, cesarean, abortion/termination, genital and urinary anomalies require cautious evaluation |
| Ref. | Country/design | Exposed population | GLP-1RA | Exposure window/cessation timing | Comparator | Maternal outcomes | Fetal/neonatal outcomes | Strength | Limitation | Interpretation |
| Bolek et al[30], 2026 | Slovakia; prospective cohort | 20 obese + PCOS + prediabetes | Semaglutide | Stopped ≥ 30 days before pregnancy | NR | Live birth 12/20 (60%); no reported maternal complication | Eutrophic neonates | Prospective; semaglutide-specific | Very small; no comparator; preconception discontinuation | Reassuring but not informative for continued pregnancy exposure |
| Bartelt et al[33], 2024 | United States; retrospective cohort, 7-year data | 1208 obese | Not specified | 90 days before conception to during pregnancy | Unexposed general population, n = 775478 | Preterm birth signal: 13% vs 10% | Higher NICU admission: 12% vs 8% | Large real-world cohort | General population comparator; high risk of confounding by obesity/metabolic disease | Not reassuring with general comparator. CHD and failure-to-thrive rates were similar by exposure timing; needs obesity-matched controls |
| Guo et al[31], 2025 | China; retrospective obese non-T2D cohort | 20 obese after exclusion of T2D | Semaglutide, titrated up to 2.4 mg/week | Used for 6 months; stopped 2 months before conception | 22 obese treated with LSM | GDM 20% vs 31.8%; CD similar; GWG: 9.16 kg vs 17.7 kg, P < 0.05 | No MCM; no NICU admission; no neonatal hypoglycemia; similar birth weight | Obesity comparator; T2D excluded; dose specified | Very small human study; preconception exposure only. Mostly animal/mechanistic emphasized study | Suggests improved GWG and GDM after preconception semaglutide; reassuring but underpowered |
| Li et al[32], 2022 | China; controlled preconception PCOS study | 57 overweight/obese, infertility, PCOS | Exenatide | 12-week pregestational exenatide | Obese PCOS/prediabetes; n = 60 on metformin | Spontaneous pregnancy rate by week 24: 29.2% vs 14.7%; total pregnancy by week 64: 79.2% vs 76%; pregnancy outcomes not significantly different | Miscarriage, live birth, preterm birth, GDM, gestational hypertension, macrosomia and birthweight not significantly different; one ASD reported in exenatide arm | Active comparator; preconception fertility context | Not designed as pregnancy-safety study; GLP-1RA was not continued as assigned therapy during organogenesis; metformin co-use and fertility-care surveillance confound interpretation | Early reassuring signal for preconception exenatide in PCOS |
| Ref. | Country/design | Exposed population | GLP-1RA | Exposure window/cessation timing | Comparator | Maternal outcomes | Fetal/neonatal outcomes | Strength | Limitation | Interpretation |
| Hviid et al[39], 2026 | Denmark; observational cohort | Liraglutide-exposed 295; semaglutide-exposed 234; T2D and obesity subgroups | Liraglutide and semaglutide vs unexposed | 8 weeks before to 8 weeks after LMP | Unexposed to liraglutide, n = 756107; unexposed to semaglutide, n = 310825 | Crude analyses-liraglutide: PET 9.2% vs 3.5%, P < 0.001. Preterm 12% vs 5%, P < 0.001; GDM 9.8% vs 4.3%, P < 0.001. Semaglutide: PET 10% vs 3.8%, P < 0.001; preterm 14% vs 5%, P < 0.001; GDM 8.1% vs 5.5%, P = 0.075. After propensity-score matching, the principal persistent association was preterm birth: Liraglutide aOR = 1.38 (95%CI: 1.01-1.90) and semaglutide aOR = 1.51 (95%CI: 1.06-2.14); this association was concentrated among women with pre-existing diabetes and was not observed in the obesity-only subgroup. Other crude differences were largely attenuated | Crude analyses-Liraglutide: LGA 18% vs 10%, P < 0.001. SGA 9.5% vs 10%, P = 0.80; placental weight increased, P < 0.001. Semaglutide: LGA 21% vs 10%, P < 0.001; SGA 11% vs 10%, P = 0.6; placental weight increased, P < 0.001 | National data; agent-specific analyses; propensity matching | Confounding by indication; mixed T2D/obesity; crude risks attenuated after matching | Risk higher for preterm in T2D-exposed cohorts but not obese cohort; most other differences attenuated after propensity matching |
| Dao et al[34], 2024 | Europe/Israel/United Kingdom; prospective cohort | 168 exposed; obese 117, T2D 46 | Liraglutide 99; semaglutide 51; dulaglutide 11; exenatide 7 | 147 women initiated GLP-1RA treatment before conception; treatment was stopped at a median GA of 5 weeks (IQR: 4-6; range: 2-40 weeks) | T2D unexposed, n = 156; obese unexposed, n = 163 | Live birth 59% vs 69% in T2D and 63% in obese; pregnancy loss including spontaneous abortion/stillbirth 23% vs 26% in T2D and 29% in obese; elective termination 18% vs 6% in T2D and 8% in obese | Preterm 8% vs 15.1% in T2D and 14.5% in obese; LGA 17.7% vs 23.4% in T2D and 6% in obese; SGA 8% vs 8.1% in T2D and 14.5% in obese; CD 46.4% vs 49.6% in T2D and 44.4% in obese; MCM: 3/117 vs 3/128 in T2D and 5/127 in obese | Prospective teratology service data; two relevant comparator groups | Small sample; referral bias; higher elective termination; limited power | Reassuring for preterm risk; increased termination; LGA risk higher than obese comparator but comparable to T2D comparator |
| Kolding et al[35], 2025 | Denmark; retrospective semaglutide cohort | 32; 27 on Ozempic, 5 on Wegovy; 22 also used insulin | Semaglutide | First trimester; 17 stopped in first trimester, and 14 continued throughout pregnancy | Insulin-exposed pregnancies without semaglutide, n = 547; no registered semaglutide or insulin, n = 103843 | Insulin comparator: Preterm 20% vs 23%, P = NS. Unexposed comparator: Preterm 20% vs 4.8%, P = 0.002 | Insulin comparator: MCM 9.4% vs 13.9%, P = NS; SGA 0% vs 4.1%, P = NS; LGA 20.7% vs 42.4%, P = 0.02; RDS 10.3% vs 8.9%, P = NS; neonatal hypoglycemia 10.3% vs 17%, P = NS; neonatal jaundice 17.2% vs 20.7%, P = NS. Unexposed comparator: MCM 9.4% vs 7.7%, P = NS; SGA 0% vs 9.5%, P = NS; LGA 20.7% vs 4.2%, P = 0.001; RDS 10.3% vs 2.4%, P = 0.024; neonatal hypoglycemia 10.3% vs 1.4%, P = 0.006; neonatal jaundice 17.2% vs 5.9%, P = 0.017 | Semaglutide-specific; two comparator groups | Very small exposed sample; high risk of bias; many on insulin; confounding by T2D severity | Apparent risk vs general/unexposed population but not vs insulin-treated T2D; comparator selection is critical |
| Imbroane et al[36], 2025 | United States; retrospective cohort | 4267; T2D 1604, obesity 3276 | Not specified; 426 tirzepatide cohort compared with 426 GLP-1RA cohort | 2 years before conception; sub-analysis 24 months, 12 months, 6 months and 3 months before conception | 1433709 unexposed unmatched; 4267 after matching | GDM 15.2% vs 18%; HDP 19.9% vs 22.8%; both P < 0.05. Reduced risk of both GDM and HDP | Preterm 3% vs 4.4%; CD 17.6% vs 19.7%; both P < 0.05; reduced risk of preterm and CD | Very large dataset; matching; timing sub-analyses; tirzepatide subgroup | Primarily preconception exposure; possible healthy-user and weight-loss confounding | Suggests antecedent GLP-1RA use may reduce coded GDM/HDP/preterm/CD; tirzepatide associated with lower CD risk |
| Pondugula et al[37], 2026 | United States; retrospective cohort, 10-year data | 243 total; T2D: 94, T1D: 9, obese 143; before pregnancy 84, during pregnancy 159 | Semaglutide 153; liraglutide 50; dulaglutide 27; tirzepatide 8 | 1 year before pregnancy to 90 days after conception | 375 unexposed: T2D 170, T1D 5, obese 200 | Lower odds of GWG below IOM recommendation with pregnancy exposure rather than before pregnancy. Obese cohort: GWG below IOM, 26% vs 38%; P < 0.01 but not significant in the diabetic cohort. Diabetic cohort: Preterm 28.4% vs 31.6%; HDP 38.8% vs 41.7%. Obese cohort: Preterm 13.6% vs 10%; HDP 19.3% vs 22%; GDM 16.4% vs 22.5% | Diabetic cohort: FGR 7.8% vs 8.6%; LGA 27% vs 32.8%; fetal death 3.9% vs 2.9%. Obese cohort: FGR 4.3% vs 6.5%; LGA 12.4% vs 11.5%; fetal death 2.9% vs 0.5% | Indication-stratified diabetic and obese cohorts; multiple agents | Small exposed sample; mixed T1D/T2D/obesity; low power for rare outcomes | No consistent significant adverse signal; GWG pattern hypothesis-generating; HDP findings require confirmation |
| Maya et al[38], 2025 | United States; retrospective pooled obstetric cohort, 9-year data | 655 total; GWG cohort 448; T2D 104, obese 344; obstetric cohort 566 | Semaglutide 202; liraglutide 118; dulaglutide 92; tirzepatide 29; others 5. Obstetric data: Semaglutide 247, liraglutide 145, tirzepatide 41 | 3 years before conception to 90 days after conception; sub-analysis 0-6 months before conception, n = 220; 90 days after conception, n = 70; | GWG analysis: 448 exposed vs 1344 matched unexposed; birthweight analysis: 442 vs 1326; obstetric outcome analysis: 566 vs 1698 | GWG: 13.7 kg vs 10.5 kg, +3.26 kg, P < 0.001. GWG lower in liraglutide than semaglutide, +2.08 kg vs +3.72 kg. Preterm: 17% vs 13%; liraglutide 13% vs 12%, semaglutide 14% vs 12%. HDP: 46% vs 36%; liraglutide 42% vs 33%; semaglutide 51% vs 38%. GDM: 20% vs 15%; liraglutide 24% vs 15%, semaglutide 17% vs 16% | CD: 46.4% vs 47%; liraglutide 41.3% vs 44.8%; semaglutide 48.5% vs 46.5%. LGA and SGA comparable | Agent and timing sub-analyses; evaluates GWG after GLP-1RA exposure/discontinuation | Exposure window extends up to 3 years; weight regain and discontinuation confounding | Prior GLP-1RA discontinuation associated with higher GWG and selected obstetric signals; residual confounding and discontinuation rebound likely. GWG higher, especially semaglutide subgroup; HDP higher in semaglutide than liraglutide; GDM higher in liraglutide than semaglutide |
| Broe et al[40], 2025 | Denmark; Danish population-based prescription safety-screening cohort | 55 liraglutide-exposed live births | Liraglutide | 0-30 days before to 91 days after LMP | Unexposed general population, n = 743276 | NR | MCM aOR = 3.52, 95%CI: 1.59-7.82 | National registry; liraglutide-specific | Screening study; live-birth-only outcome; small liraglutide sample; general-population comparator; incomplete disease-specific confounder control | Hypothesis-generating liraglutide signal requiring disease-specific validation; not definitive teratogenic evidence |
| Brown et al[41], 2026 | United States; observational cohort | 3572 | Liraglutide 903; semaglutide 1644; tirzepatide 461; dulaglutide 417; exenatide 128; albiglutide 15; lixisenatide 4 | 90 days before conception to first trimester | Continuation of further GLP-1RA dispensing into first trimester (n = 1433) vs no further dispensing after LMP among women with pre-LMP GLP-1RA dispensing (n = 1086) | Maternal outcomes not primary | Non-live birth: aRR = 1.09 (95%CI: 0.98-1.23); spontaneous abortion: aRR = 0.99 (95%CI: 0.88-1.13); elective termination: aRR = 1.78 (95%CI: 1.29-2.60). Among linked live births, prevalence ratios were 1.29 (95%CI: 0.82-2.06) for SGA, 1.08 (95%CI: 0.84-1.40) for LGA, and 1.21 (95%CI: 0.83-1.82) for MCM | Large cohort; clinically relevant continuer vs non-continuer comparison | Confounding by reason for continuation, disease severity, pregnancy intention and counseling; maternal outcomes NR | Reassuring against large fetal structural signal from continued dispensing; elective termination increase was likely influenced by pregnancy recognition, counseling, exposure anxiety, or pregnancy intention; no evidence supporting intentional continuation |
Diabetes-specific cohorts: Four diabetes-specific cohorts were broadly reassuring. Chou et al[26] found no significant increase in preterm birth, MCMs, stillbirth, or small for gestational age (SGA) birth among GLP-1RA-exposed women with T2D compared with insulin-treated controls. Cesta et al[27] provided the most informative diabetes-specific active-comparator evidence currently available. This multinational, population-based cohort used data from four Nordic countries, the United States MarketScan, and Israeli Maccabi Health Services. The principal periconceptional definition was at least one prescription fill from 90 days before the last menstrual period (LMP) through the end of the first trimester, with insulin as the active comparator; however, database-specific ascertainment windows ranged from 180-365 days before LMP through 90-97 days of pregnancy. Among women with T2D, 938 pregnancies were GLP-1RA-exposed and 5078 insulin-exposed. Compared with insulin, GLP-1RA exposure was not associated with any MCM [adjusted RR (aRR) = 0.95; 95% confidence interval (95%CI): 0.72-1.26] or cardiac malformation (aRR = 0.68; 95%CI: 0.42-1.12). A first-trimester-only sensitivity analysis was also reassuring for MCM (RR = 1.03; 95%CI: 0.73-1.47), although the number of exposures was smaller. Because the active comparator consisted of insulin-treated pregnancies, the estimate is more clinically relevant than comparisons with metabolically healthy women. However, it still does not fully balance diabetes duration, obesity, or glycemic control. The first-trimester-only sensitivity analysis narrows the biologically relevant window but remains based on dispensing rather than measured exposure.
Its major strengths were restriction to pregestational T2D, use of insulin rather than a general-population comparator, multinational population-based data, harmonized cohort construction, exclusion of known teratogenic drug exposures and chromosomal anomalies, standardized malformation definitions, infant follow-up through 1 year in Nordic and Israeli data, and a first-trimester sensitivity analysis. Important limitations remained. Exposure was inferred from prescription fills rather than confirmed ingestion, adherence, dose, exact last dose, or discontinuation; a pre-LMP fill did not necessarily imply embryonic exposure. Insulin or metformin co-prescription was common, making persistent GLP-1RA exposure difficult to distinguish from early switching to pregnancy-compatible treatment. Residual confounding was likely because GLP-1RA users more often had obesity and PCOS, while glycemic data were available only in subsets of the United States and Israeli cohorts. Malformation analyses were largely restricted to live births, precluding reliable assessment of miscarriage, termination, stillbirth, or lethal anomalies, and estimates remained imprecise for rare organ-specific defects. The study also could not reliably identify early switching from GLP-1RA to insulin or metformin, and database-level heterogeneity means that the same nominal periconceptional label did not represent an identical biological exposure window across countries. These features should be explicit when counseling an individual who stopped treatment before LMP versus one who continued after conception. Thus, Cesta et al[27] argue against a large teratogenic effect compared with insulin-treated T2D but do not establish safety for planned or continued pregnancy use.
Hanif et al[28] used the TriNetX global electronic medical record (EMR)/EHR network to compare women with T2D and GLP-1RA exposure recorded from 1 year before to 1 month after first-trimester pregnancy diagnosis with unexposed women with T2D during the same window. After propensity matching, 1826 women remained in each cohort. Maternal all-cause mortality over 42 weeks, HDP, and fetal cardiac and kidney anomalies were comparable. These findings support the absence of an obvious short-term maternal or structural fetal signal in T2D. The analysis was notable for its large matched sample and clinically relevant T2D comparator, but it reported broad coded outcomes rather than detailed malformation phenotypes or agent-specific estimates. However, the broad EMR-based window did not confirm actual ingestion, agent, dose, adherence, duration, exact last dose, or discontinuation timing and could miss medication use outside the network. Exposure coding and incomplete capture introduced misclassification, and residual confounding persisted because exposed women had higher body mass index (BMI) and glycated hemoglobin (HbA1c) even after matching. Moreover, the exposure window extended far into preconception, so some women classified as exposed may have had little or no residual drug during organogenesis; conversely, a prescription near pregnancy diagnosis did not establish continued use. The study is therefore reassuring but limited and does not constitute definitive evidence of pregnancy safety.
Lewin et al[29] used Epic Cosmos to evaluate 35083 patients representing 40929 pregnancies with pregestational diabetes and documented GLP-1RA use during pregnancy; 92.6% had T2D. In crude comparisons, exposed pregnancies had higher rates of preeclampsia (13.1% vs 11.9%; P < 0.00001), CD (44.4% vs 38.9%; P < 0.001), first-trimester pregnancy loss or termination before 14 weeks (9.8% vs 7.5%), and preterm delivery at 22 weeks to < 37 weeks (23.1% vs 20.8%; P < 0.0001). These findings occurred in a group with substantially higher BMI, HbA1c, chronic hypertension, chronic kidney disease, and PCOS. The overall malformation rate was identical (3.3% vs 3.3%), with no difference in cardiac malformations (1.2% vs 1.2%), central nervous system or limb malformations, or orofacial clefts. Small absolute increases occurred in genital (0.7% vs 0.6%; P = 0.02) and urinary malformations (0.5% vs 0.4%; P = 0.01), generating an organ-specific signal without an overall MCM signal. The crude adverse obstetric differences were statistically detectable because of the very large sample size, but their absolute magnitudes were modest. Conversely, the identical total malformation rate provides reassurance against a broad structural signal while not excluding rare agent- or organ-specific effects.
No individual-level data were available, precluding multivariable regression, propensity matching, or adjustment for baseline differences. Exposure was defined broadly as GLP-1RA use at any time during pregnancy, without information on the agent, dose, duration, adherence, gestational timing, or discontinuation week. Pregnancy loss and termination were combined, preventing separation of spontaneous loss from elective termination. Congenital malformations were identified from International Classification of Diseases, 10th Revision (ICD-10) codes, which may misclassify anomalies and cannot consistently separate major from minor defects. The study also lacked stratification by diabetes type, glycemic severity, obesity class, hypertension, medication co-exposure, or agent. The unadjusted design is particularly important because even small baseline differences can materially affect renal, genital, hypertensive, and preterm outcomes in pregnancies complicated by diabetes. The small genital/urinary differences may therefore reflect pregestational diabetes severity, nephropathy, chronic hypertension, obesity, glycemic exposure, differential fetal imaging, or coding artifacts rather than direct toxicity or post-discontinuation metabolic destabilization. The finding warrants replication in validated congenital-anomaly registries with adjudicated defects and active diabetes comparators rather than dismissal or causal attribution. Lewin et al[29] is best viewed as a very large unadjusted signal-detection study: Reassuring for total malformations but insufficient to establish causality for the genital/urinary findings.
Overall, diabetes-specific evidence argues against a large teratogenic effect when metabolically relevant comparators are used. Renal/genitourinary and adverse obstetric signals nevertheless require confirmation in datasets with granular adjustment for HbA1c, diabetes duration, obesity severity, hypertension, nephropathy, and concomitant therapy (Table 4).
Obesity-, PCOS-, and fertility-focused studies: Four obesity-, PCOS-, or fertility-focused studies were generally more informative for preconception metabolic optimization than for direct fetal safety[30-33]. Bolek et al[30] and Guo et al[31] reported no MCM after semaglutide was stopped before conception, but Bolek et al[30] did not have a comparator and did not report detailed maternal or neonatal outcomes. The study by Guo et al[31] included 20 treated women and 22 lifestyle controls; preconception therapy was associated with lower GWG and improved metabolic measures, but it was not powered to assess congenital malformations or rare neonatal outcomes and focused substantially on mechanistic and animal data. Guo et al[31] also reported numerically lower GDM rates and no NICU admissions or neonatal hypog
Li et al[32] followed 160 overweight or obese Chinese women with infertility and PCOS randomized to 12 weeks of pregestational short-acting exenatide twice daily or metformin. Exenatide produced greater weight loss and improved insulin resistance, with a higher spontaneous pregnancy rate by week 24 (29.2% vs 14.7%; P = 0.03), but similar total pregnancy rates after assisted reproductive technology by week 64 (79.2% vs 76.0%; P = 0.65). Miscarriage, live birth, preterm delivery, GDM, HDP, fetal macrosomia, GWG, and neonatal birthweight did not differ. Because exenatide was stopped before active conception attempts, all women subsequently received metformin, and assisted reproductive technology followed failure of spontaneous conception, the study informs preconception optimization rather than safety during organogenesis. Fertility and assisted reproductive technology cohorts offer greater temporal precision because ovulation trigger, insemination, or embryo-transfer dates can be aligned with GLP-1RA discontinuation. The exact conception chronology is a distinctive strength: Unlike prescription databases, fertility treatment records can more closely locate the interval between final exposure and ovulation or embryo transfer. Nevertheless, the trial was designed primarily for metabolic and fertility outcomes, not rare fetal safety events. However, such cohorts remain confounded by PCOS, infertility diagnosis, obesity, insulin resistance, metformin co-use, and intensive surveillance. Bartelt et al[33], an Epic Research report, found higher NICU admission after glucagon-like peptide-1 medication prescription within 90 days before pregnancy among women without diabetes/GDM, while congenital heart defect and failure-to-thrive diagnoses were similar. The general-population comparator, incomplete adjustment for obesity/metabolic severity, and non-peer-reviewed format make this finding hypothesis-generating. In addition, medication orders in Epic do not confirm dis
Mixed diabetes/obesity cohorts: Mixed-indication studies included diabetes, obesity, PCOS, and infertility and were especially sensitive to comparator choice. Dao et al[34], in a prospective teratology information service cohort, found no increase in major birth defects or pregnancy loss compared with diabetic and overweight/obesity comparators. Elective termination was more frequent, potentially reflecting pregnancy intention, counseling practice, or exposure anxiety; preterm birth was lower than in both comparator groups; and large for gestational age (LGA) birth was higher than in obese controls but similar to diabetic controls. Its prospective design and clinically relevant comparators make it useful for post-exposure counseling. Kolding et al[35] reported 32 semaglutide-exposed pregnancies, 14 of which involved continued exposure throughout pregnancy. Adverse neonatal outcomes were higher than in a general unexposed population but not clearly higher than in insulin-treated diabetes controls, again demonstrating strong comparator dependence. Dao et al[34] also reported descriptive differences in termination, preterm birth, and LGA rather than uniformly adjusted causal estimates, while the small Kolding et al[35] sample produced wide uncertainty despite agent-specific exposure information.
Imbroane et al[36] used the United States Collaborative Network in TriNetX and 1:1 propensity matching for age, race, ethnicity, T2D, overweight/obesity, hypertension, and prediabetes. In the principal analysis, 4267 women with an antecedent GLP-1RA prescription within 24 months before pregnancy were matched to 4267 controls. Prescription was associated with lower odds of GDM [odds ratio (OR) = 0.81; 95%CI: 0.72-0.91], HDP (OR = 0.84; 95%CI: 0.76-0.94), preterm delivery (OR = 0.68; 95%CI: 0.54-0.85), and CD (OR = 0.89; 95%CI: 0.87-0.97). A 3-month-window analysis included 1350 matched pairs. Lower odds persisted for GDM (OR = 0.82; 95%CI: 0.67-1.00), HDP (OR = 0.81; 95%CI: 0.67-0.97), gestational hypertension (OR = 0.71; 95%CI: 0.55-0.91), and CD (OR = 0.71; 95%CI: 0.58-0.87), whereas preeclampsia and preterm delivery were no longer significant. Thus, lower rates of coded GDM and hypertensive disorders were more consistent across windows than lower rates of preterm delivery. The 3-month findings were important because they tested whether associations persisted when the exposure-history window was narrowed to the period around conception. Their attenuation for preterm delivery and preeclampsia demonstrates that apparently favorable results were not uniform across reasonable exposure definitions. In a post hoc comparison of 426 tirzepatide- and 426 GLP-1RA-prescribed women, tirzepatide was associated with lower coded cesarean risk (OR = 0.55; 95%CI: 0.37-0.83), with no other significant differences.
The study suggests that antecedent GLP-1RA prescription may identify or contribute to better preconception cardiometabolic health. However, a prescription within 24 months or 3 months does not establish confirmed use, duration, adherence, dose, weight change, exact discontinuation, or biological exposure near conception. The 24-month analysis may primarily capture a history of access to and response to metabolic treatment. By contrast, the 3-month window is closer in time to conception but still cannot distinguish cessation months before conception from continued use into early pregnancy. ICD-10-based covariates and outcomes were susceptible to misclassification; indication, pregnancy loss, termination, and post-conception exposure were unavailable; and the mixed high-risk cohort included substantial T2D, obesity, hypertension, and prior obstetric morbidity. In particular, inclusion of women with pre-existing T2D complicates interpretation of coded incident GDM. Favorable associations may reflect healthier-user effects, weight reduction before pregnancy, more intensive medical care, or residual selection rather than a pharmacologic effect on pregnancy. The favorable associations, including the small post hoc tirzepatide comparison, are therefore hypothesis-generating rather than evidence of causal obstetric benefit.
Pondugula et al[37] studied 243 exposed pregnancies: 103 treated for pregestational diabetes and 140 for weight management. Among exposed women, 34.6% had pre-pregnancy-only exposure, and 65.4% had evidence of exposure during pregnancy. Semaglutide was the last GLP-1RA in 63.0%, followed by liraglutide (20.6%), dulaglutide (11.1%), tirzepatide (3.3%), and exenatide (2.1%). Comparators comprised 175 women with pregestational diabetes receiving non-GLP-1RA therapy and 200 women in BMI-stratified weight-management groups. In the diabetes cohort, GLP-1RA expo
By separating indication and timing, the study suggests that discontinuation may alter gestational weight trajectory, particularly after weight-management treatment. Still, it does not demonstrate that GLP-1RAs reduce HDP or broader obstetric risk. The lower probability of inadequate GWG should not be misread as a proven benefit because rebound gain may move some women from inadequate toward recommended gain while also increasing total weight in others. The clinical implication is individualized nutritional and weight monitoring after withdrawal, not continuation of therapy. Limitations included the retrospective single-center design, modest sample size, exclusion of early losses and induced abortions without delivery admission, missing glycemic and preconception weight-change data, possible ICD-10 misclassification, aggregation of agents with different potency and half-lives, incomplete capture of external or compounded products, and control selection over a narrower time period than the exposed cohort. The findings remain hypothesis-generating.
Maya et al[38] shifted the focus from fetal structural safety to the metabolic consequences of stopping GLP-1RAs. Exposure was defined by at least one EMR medication order from 3 years before conception to 90 days after conception; 655 pregnancies met this definition. After exclusions and matching, the GWG analysis included 448 exposed and 1344 matched unexposed pregnancies, the birthweight analysis included 442 and 1326, and the obstetric analysis included 566 and 1698, respectively. Semaglutide was most frequently ordered, followed by liraglutide and dulaglutide. Mean GWG was 13.7 kg vs 10.5 kg (difference, +3.26 kg), and excessive GWG was more frequent after exposure/discontinuation (65% vs 49%; RR = 1.32). Preterm delivery (17% vs 13%; RR = 1.34), GDM (20% vs 15%; RR = 1.30), and HDP (46% vs 36%; RR = 1.29) were also higher. Exploratory agent-specific analyses suggested GDM after liraglutide discontinuation and HDP after semaglutide discontinuation. Still, these were comparisons with separate matched control groups rather than a direct head-to-head comparison. Maya et al[38] is therefore best interpreted as a discontinuation-trajectory study suggesting rebound GWG with downstream metabolic and obstetric risks, rather than as evidence of toxicity from continued exposure. Matching used pre-pregnancy rather than pre-treatment BMI; medication orders did not confirm ingestion, dose, or exact cessation; and residual confounding remained. Importantly, the medication-order definition included women with remote preconception exposure, and the analysis estimated outcomes among treated pregnancies rather than all potentially eligible pregnancies. Greater GWG may partly represent regain after successful preconception weight loss, an effect not fully captured by matching on weight immediately before pregnancy. Nevertheless, the study identifies post-discontinuation weight and metabolic management as a key component of transition to pregnancy-compatible care.
Hviid et al[39] identified 529 Danish pregnancies with prescription redemption for liraglutide (n = 295) or semaglutide (n = 234) within 8 weeks before or after LMP. In crude analyses, preeclampsia, GDM, preterm birth, and LGA were more frequent. Still, most differences attenuated after propensity matching for maternal age, BMI, smoking, region, education, parity, pre-existing diabetes, and temporal factors. No consistent adjusted association remained for preeclampsia, GDM, birthweight, placental weight, SGA, or LGA. The persistent adjusted signal was preterm birth: Liraglutide aOR = 1.38 (95%CI: 1.01-1.90) and semaglutide aOR = 1.51 (95%CI: 1.06-2.14). This association was confined to women with pre-existing diabetes—liraglutide aOR = 1.70 (95%CI: 1.17-2.48) and semaglutide aOR = 1.84 (95%CI: 1.24-2.71)—and was absent among women without diabetes: Liraglutide aOR = 1.01 (95%CI: 0.58-1.76) and semaglutide aOR = 0.71 (95%CI: 0.30-1.70). The diabetes-stratified analysis is central: The overall preterm association would appear drug-related if diabetes status were ignored, whereas the null estimates among women without diabetes support substantial confoun
Nationwide linkage, agent-specific analyses, a biologically relevant exposure window, and stratification by diabetes status make Hviid et al[39] one of the more informative registry studies and suggest that the preterm signal is largely related to underlying diabetes. Limitations include redemption rather than confirmed use; unknown dose and discontinuation; possible residual confounding by diabetes severity, glycemic control, weight trajectory, hypertension, and fertility treatment; and inclusion only of pregnancies resulting in delivery, which precluded assessment of miscarriage or termination. The delivery-only design can also introduce selection if early losses differ between groups, and the relatively recent introduction of semaglutide for obesity limited obesity-specific follow-up. Semaglutide weight-management use occurred mainly late in the study period, and the findings may not generalize to tirzepatide or other agents. Thus, the absence of an adjusted signal outside preterm birth is reassuring but not definitive.
Broe et al[40], in a Novo Nordisk-funded analysis, reported an increased MCM rate with liraglutide vs a general-population comparator (aOR = 3.52; 95%CI: 1.59-7.82). The liraglutide group included only 55 pregnancies, whereas the unexposed comparator exceeded 0.7 million, limiting causal inference. Brown et al[41] addressed a more clinically relevant question: Among 3572 pregnancies with a GLP-1RA dispensation within 90 days before LMP, what occurred when dispensing continued into the first trimester vs when no further dispensing occurred after LMP? Semaglutide accounted for 1644 exposures (46.0%), liraglutide 903 (25.3%), tirzepatide 461 (12.9%), dulaglutide 417 (11.7%), exenatide 128 (3.6%), albiglutide 15 (0.4%), and lixisenatide 4 (0.1%); 41.1% had T2D. The target-trial emulation used cloning, censoring, and weighting to align treatment classification with follow-up and reduce immortal-time bias. The analysis directly compared 2 strategies among prevalent users rather than comparing users with the general population, thereby reducing—but not eliminating—confounding by indication. The grace-period strategy was clinically realistic because women may not recognize pregnancy immediately and dispensing behavior does not change instantaneously at LMP. The weighted risk of non-live birth was similar (aRR = 1.09; 95%CI: 0.98-1.23), as was spontaneous abortion (aRR = 0.99; 95%CI: 0.88-1.13), whereas elective termination was more frequent with continuation (aRR = 1.78; 95%CI: 1.29-2.60). Among linked live births with adequate infant enrollment, continuation was not definitively associated with SGA (RR = 1.29; 95%CI: 0.82-2.06), LGA (RR = 1.08; 95%CI: 0.84-1.40), or MCM (RR = 1.21; 95%CI: 0.83-1.82); cardiovascular malformations were similar (2.4% vs 2.5%). Fetal-abnormality codes were not more frequent among elective terminations, suggesting that the higher termination rate may reflect pregnancy intention, delayed recognition, exposure anxiety, or counseling rather than a clear anomaly signal. The distinction between continuation and non-continuation is operational rather than pharmacokinetic: A non-continuer could still carry residual semaglutide or tirzepatide from a pre-LMP dose, whereas a continuer may not have injected every dispensed dose.
The continuation-vs-non-continuation design, contemporary agents, clinically relevant comparator, and target-trial methods reduce confounding by indication and immortal-time bias. Nevertheless, dispensing does not prove ingestion, dose, adherence, exact last injection, or biological exposure. Non-continuers may retain drug supply or residual systemic concentrations from pre-LMP dispensing, particularly with long-acting agents; therefore, the contrast represents further dispensing vs no further dispensing rather than biologically exposed vs unexposed fetuses. HbA1c, obesity severity, preconception weight loss, pregnancy intention, timing of pregnancy recognition, and reasons for continuation were unavailable. Live-birth analyses required infant linkage and adequate enrollment, malformations were identified from claims algorithms rather than validated anomaly registries, and CIs for SGA and MCM remained wide. The higher elective-termination association is especially vulnerable to pregnancy intention, recognition, counseling, and exposure anxiety and should not be construed as evidence of fetal toxicity. Conversely, wide CIs mean that modest increases in rare structural outcomes cannot yet be excluded. Brown et al[41] is reassuring against a large fetal signal from inadvertent early continuation but does not support planned treatment during pregnancy.
Across database-based cohorts, exposure was usually inferred from prescriptions, medication orders, dispensations, or redemptions rather than measured administration. Agent-specific dose, adherence, exact last-dose timing, maternal concentrations, and post-discontinuation glycemic or weight trajectories were commonly unavailable. Comparator choice was equally influential: Insulin-treated T2D, obesity-matched, and general-population controls estimate different clinical contrasts. These recurring limitations are therefore best considered collectively rather than interpreted as evidence that every observed adverse or favorable association reflects a direct drug effect. This heterogeneity also explains why the studies cannot be collapsed into a single binary category of exposed vs unexposed pregnancies. A remote antecedent prescription, discontinuation before conception, discontinuation at pregnancy recognition, and dispensing throughout the first trimester represent different clinical and embryologic questions.
Taken together, mixed cohorts show that apparent risk depends heavily on the choice of comparator and the definition of exposure. Major-malformation signals generally attenuate with metabolically relevant comparators, whereas elective termination, preterm birth, HDP, GWG, and fetal-growth outcomes remain incompletely resolved (Table 6).
Three systematic reviews and meta-analyses synthesized human pregnancy outcomes: Kodali et al[42], Armstrong et al[43], and Hakim et al[44] (Table 7). Kodali et al[42] pooled 4 comparative observational studies comprising 36963 pregnancies. The exposed population was older and differed materially from comparators: 73% vs 21% had overweight/obesity, whereas 49% vs 91% had pre-existing T2D. Only 21% of exposed women had periconceptional or first-trimester exposure; most exposure was preconceptional. GLP-1RA exposure was associated with lower odds of adverse maternal outcomes (OR = 0.72; 95%CI: 0.66-0.79; P < 0.00001) and preterm birth (OR = 0.66; 95%CI: 0.53-0.82; P = 0.0001). MCM were not increased vs other antidiabetic agents (OR = 1.08; 95%CI: 0.86-1.37; P = 0.51) or insulin alone (OR = 1.00; 95%CI: 0.77-1.28; P = 0.98), and overall adverse fetal outcomes were also not significantly increased (OR = 0.91; 95%CI: 0.72-1.16; P = 0.44). These apparently favorable maternal findings require caution because only 4 heterogeneous observational studies were included, exposure windows and comparators varied, baseline metabolic risk was markedly imbalanced, and most exposure preceded conception. Residual confounding by indication, obesity severity, glycemic control, treatment access, and health-system factors could explain the apparent benefit. The review also contained a text-forest-plot inconsistency for adverse fetal outcomes and did not adequately separate remote preconceptional exposure from exposure during organogenesis. It therefore provides an early pooled signal against a large increase in MCM, but not evidence that GLP-1RAs improve pregnancy outcomes or are safe for planned pregnancy use.
| Ref. | Search window/databases | Included evidence | Exposed pregnancies | Exposure window | Main maternal outcomes | Main fetal/neonatal outcomes | Strength | Limitation | Interpretation |
| Kodali et al[42], 2026 | PubMed and Google Scholar; January 2010-April 2025 | 4 observational studies: 3 retrospective cohorts from North America/Europe and 1 prospective cohort from Europe/Middle East | 5405 exposed for fetal composite analysis; total pregnancies 36963 | During or before pregnancy; only 21% periconceptional/first trimester | Adverse maternal outcomes lower: OR = 0.72, 95%CI: 0.66-0.79, P < 0.00001. Preterm birth lower: OR = 0.66, 95%CI: 0.53-0.82, P = 0.0001 | Major congenital anomalies not increased: OR = 1.08, 95%CI: 0.86-1.37, P = 0.51; insulin-only comparator OR = 1.00, 95%CI: 0.77-1.28, P = 0.98. Overall adverse fetal outcomes: Text reports OR = 0.91, 95%CI: 0.72-1.16, P = 0.44; forest plot reports OR = 0.59, 95%CI: 0.36-0.96, P = 0.03 with I2 = 87% | Focused SRMA; includes comparator cohorts; highlights lack of MCM signal | Only 4 studies; no formal risk-of-bias table reported; text/figure inconsistency for adverse fetal outcome; sparse outcomes; study dominance; heterogeneous exposure/comparators | Reassuring for MCM, but apparent maternal/preterm benefit should be interpreted as observational association rather than drug benefit |
| Armstrong et al[43], 2025 | PubMed, EMBASE, Cochrane; up to June 2025 | Comparative studies; table/review reports inconsistently describe 6-9 studies; includes cohort/RCT-derived data | 8761 exposed among 182269 participants | Pregnancy/preconception; first-trimester subgroup reported | GDM: OR = 0.86, 95%CI: 0.56-1.32, P = 0.49. Miscarriage: OR = 1.16, 95%CI: 0.82-1.65, P = 0.41. Live birth: OR = 0.72, 95%CI: 0.50-1.06, P = 0.10 | Major birth defects: OR = 0.97, 95%CI: 0.77-1.22, P = 0.79; CHD: OR = 0.63, 95%CI: 0.42-0.95, P = 0.03; preterm delivery: OR = 1.66, 95%CI: 0.23-12.23, P = 0.62, I2 = 98%; stillbirth: OR = 0.29, 95%CI: 0.07-1.20, P = 0.09 | Used ROBINS-I/RoB2; leave-one-out sensitivity; first-trimester subgroup analyses; publication bias not assessed because fewer than 10 studies | Internal inconsistency in study count; pooled heterogeneous evidence including Parker regulatory data and preconception fertility RCT; high heterogeneity for preterm birth; unexpected protective CHD signal likely confounded | Supports absence of major malformation signal; not sufficient to infer fetal protection or pregnancy safety |
| Hakim et al[44], 2026 | PubMed, EMBASE, Web of Science; through Nov 2025 | 22 studies in systematic review; 10 comparative observational studies in meta-analysis; case reports/series and pharmacovigilance summarized narratively | 49395 exposed pregnancies overall; 43579 exposed in comparative observational MA as reported | 3 months before LMP through pregnancy | Preterm delivery OR = 1.11, 95%CI: 0.84-1.47, I2 = 46.1%; live birth OR = 0.63, 95%CI: 0.37-1.09; spontaneous pregnancy loss OR = 1.13, 95%CI: 0.68-1.88; elective termination OR = 2.39, 95%CI: 0.32-17.68; HDP OR = 0.86, 95%CI: 0.52-1.43; preeclampsia OR = 1.55, 95%CI: 0.57-4.19; CD OR = 1.42, 95%CI: 0.55-3.64 | MCM: OR = 1.19, 95%CI: 0.72-1.98, I2 = 58.5%; CHD: OR = 0.88, 95%CI: 0.53-1.44, I2 = 47.7%; renal malformations OR = 1.23, 95%CI: 1.09-1.39, I2 = 0%, largely driven by one large cohort; T2D-only MCM: OR = 1.01, 95%CI: 0.98-1.06; semaglutide-only live birth: OR = 0.57, 95%CI: 0.35-0.95 | Largest and most comprehensive synthesis; separates comparative MA from narrative noncomparative evidence; comparator-specific and agent-specific analyses; Joanna Briggs Institute appraisal | Residual confounding by disease severity; prescription/dispensing exposure misclassification; heterogeneous comparators; publication-bias assessment was limited by the small number of studies contributing to individual outcomes; no clear leave-one-out for renal signal | Strongest current synthesis: Cautiously reassuring for overall MCM, but renal signal and semaglutide live-birth finding require cautious, granular evaluation |
Armstrong et al[43] pooled 6 comparative studies of preconceptional or pregnancy exposure. No significant differences were found for fetal growth restriction/SGA, birthweight, live birth, MCM, miscarriage, preterm delivery, stillbirth, or GDM. Lower odds of congenital heart defects were reported, including in the first-trimester subgroup (OR = 0.62; 95%CI: 0.41-0.95; P = 0.03). This apparent protective association is biologically uncertain and vulnerable to residual confounding, comparator imbalance, and study-selection effects. The synthesis combined heterogeneous indications, exposure windows, data sources, and comparators; it also included regulatory clinical-trial pregnancy data, which are methodologically distinct from routine care cohorts. Substantial heterogeneity in some outcomes, especially preterm delivery, and inconsistencies in the reported search period and number of included studies further limit interpretation. Armstrong et al[43] therefore support the absence of a consistent major malformation or miscarriage signal, while the congenital heart defect finding should be considered hypothesis-generating rather than causal.
Hakim et al[44] provided the broadest secondary synthesis, pooling 10 comparative observational studies. Periconceptional exposure was defined as the period from 3 months before LMP through pregnancy. No significant associations were found for overall MCM (OR = 1.19; 95%CI: 0.72-1.98; I2 = 58.5%), cardiac malformations (OR = 0.88; 95%CI: 0.53-1.44; I2 = 47.7%), LGA (OR = 1.23; 95%CI: 0.59-2.55; I2 = 63.3%), or SGA (OR = 0.88; 95%CI: 0.54-1.45; I2 = 0.0%). Pregnancy and delivery outcomes were likewise inconclusive: Preterm delivery (OR = 1.11; 95%CI: 0.84-1.47; I2 = 46.1%), live birth (OR = 0.63; 95%CI: 0.37-1.09; I2 = 0.0%), spontaneous pregnancy loss (OR = 1.13; 95%CI: 0.68-1.88; I2 = 0.0%), elective termination (OR = 2.39; 95%CI: 0.32-17.68), HDP (OR = 0.86; 95%CI: 0.52-1.43; I2 = 79.7%), preeclampsia (OR = 1.55; 95%CI: 0.57-4.19; I2 = 99.4%), and CD (OR = 1.42; 95%CI: 0.55-3.64; I2 = 52.4%). The principal statistically significant finding was renal malformation (OR = 1.23; 95%CI: 1.09-1.39; I2 = 0.0%), but this was largely driven by the large, unadjusted Lewin cohort, which had substantial baseline imbalance. A semaglutide-specific live-birth signal was also interpreted cautiously because of obesity and glycemic confounding, heterogeneous comparators, and potentially anxiety-driven terminations.
Hakim et al[44] is the best available secondary synthesis because of its broader database search, larger evidence base, explicit exposure-window definition, separation of comparative studies from case reports/series, random-effects methods, comparator-specific sensitivity analyses, and integration of non-pooled outcomes. Nevertheless, the underlying evidence remained observational and susceptible to confounding by indication, diabetes severity, obesity, hypertension, infertility, PCOS, and comparator choice. Exposure was commonly inferred from prescription or dispensing records rather than confirmed ingestion, dose, adherence, or exact discontinuation. Heterogeneity was moderate to very high for several obstetric outcomes, rare organ-specific anomalies remained underpowered, and long-term offspring neurodevelopmental and cardiometabolic outcomes were unavailable. The renal signal therefore requires confirmation in validated anomaly registries. Hakim et al[44] should be regarded as the most robust current synthesis, but not as definitive evidence of pregnancy safety.
Across all 3 meta-analyses, the most consistent finding was the absence of a reproducible overall MCM signal. Confidence remains limited because the primary studies overlap, are observational, use heterogeneous comparators and exposure windows, and frequently rely on prescription-, dispensing-, or EHR-based exposure definitions (Table 7).
Six systematic reviews without meta-analysis were retained solely for contextualization and cross-checking of references[45-50]. Earlier reviews were dominated by concerns about animal reproductive toxicity, pharmacologic caution, and isolated case reports; later reviews incorporated emerging human cohorts and generally found no consistent major teratogenic signal, while continuing to recommend avoidance during pregnancy. Because these reviews contributed no independent patient- or cohort-level data and substantially overlapped with the primary studies already extracted, treating them as core evidence would risk double-counting and inflating the apparent evidence base.
Quality appraisal was performed to contextualize evidence rather than determine eligibility. The JBI case-report appraisal showed that most reports described maternal background and exposure timing reasonably well. Still, many had limited follow-up, incomplete diagnostic detail, short infant outcome reporting, and substantial confounding by maternal metabolic disease. Case series were limited by small sample size, uncertain consecutive inclusion, and limited long-term follow-up. The JBI cohort appraisal showed that studies using active metabolic comparators, particularly insulin-treated T2D or obesity-matched comparator groups, were generally more informative than studies using general-population comparators. Common cohort limitations included prescription-based exposure ascertainment, incomplete adherence data, broad exposure windows, residual confounding by disease severity, limited data on glycemic status and BMI, and incomplete adjustment for PCOS, infertility, hypertension, nephropathy, and concomitant medication use. The pharmacovigilance study was appraised as useful for signal detection but limited in causal inference because of the absence of a denominator, reporting bias, and an inability to adjust for confounding. The regulatory clinical-trial pregnancy dataset was appraised as useful descriptive evidence but limited for causal inference because pregnancies were accidental, outcome reporting was incomplete, and adjusted comparisons were not possible.
The AMSTAR-2 critical-domain appraisal identified Hakim et al[44] as the strongest available meta-analysis because it used the broadest search, included the largest evidence base, separated comparative and non-comparative evidence, and reported study-level appraisal. Armstrong et al[43] was methodologically useful but limited by heterogeneous pooling and inclusion of regulatory trial data alongside observational cohorts. Kodali et al[42] was clinically relevant but limited by the small number of studies, sparse outcomes, unclear detailed risk-of-bias reporting, and internal inconsistency. Supplementary Tables 2-7 summarize the quality appraisal of all the studies included in this scoping review.
Human evidence on GLP-1RA exposure before conception and during early pregnancy has expanded rapidly but remains methodologically uneven. Across case reports, pharmacovigilance, regulatory trial pregnancy data, comparative cohorts, registries, and meta-analyses, no consistent overall increase in MCM has emerged after inadvertent preconceptional or early-pregnancy exposure. This supports cautious reassurance after accidental exposure but does not establish safety for planned or continued use during pregnancy. The evidence is best interpreted as a layered body of information: Descriptive reports provide clinical detail, pharmacovigilance detects signals, regulatory data capture unintended trial pregnancies, comparative studies estimate risk but remain vulnerable to confounding and exposure misclassification, and meta-analyses inherit the limitations and overlap of the underlying observational studies. Thus, current human data argue against a large or consistent teratogenic signal after inadvertent early exposure, rather than proving pregnancy safety. The relative inferential role of each evidence layer, the central clinical interpretation, and research priorities are summarized in Figure 3.
Uncertainty arises mainly from heterogeneous exposure windows, prescription-based ascertainment, and major differences in baseline metabolic risk. Preconceptional use, discontinuation after pregnancy recognition, and continued pregnancy exposure are biologically distinct but are often grouped (Figure 4). Prescription, dispensing, redemption, or EMR do not reliably establish ingestion, adherence, dose escalation, exact last dose, treatment duration, or discontinuation timing. In addition, most exposed women have T2D, obesity, PCOS, infertility-associated metabolic dysfun
A pharmacokinetic-embryologic framework is useful when long-acting incretin-based therapies are discontinued after pregnancy is recognized. Semaglutide and tirzepatide do not disappear immediately: Their terminal half-lives are approximately 1 week and 5 days, respectively. If discontinued around gestational age (GA) weeks 4-5, residual maternal exposure may overlap with early organogenesis, broadly GA weeks 5-10 or post-fertilization weeks 3-8. With first-order elimination, semaglutide declines more gradually and tirzepatide more rapidly; residual levels therefore fall progre
Maternal serum concentration cannot be equated directly with embryonic exposure. Semaglutide and tirzepatide are modified peptides of approximately 4.1 kDa and 4.8 kDa and are about 99% albumin-bound; liraglutide is approximately 3.7 kDa, whereas dulaglutide is approximately 63 kDa because of its fusion-protein structure. These properties would be expected to limit passive placental transfer compared with small, lipophilic, non-protein-bound drugs, but direct data on human placental transfer during early organogenesis are lacking. Figure 5 should therefore be interpreted as a theoretical maternal washout framework, not a fetal-dose model. Existing studies rarely report measured drug concentrations, exact last-dose dates, dose-escalation histories, placental transfer, or agent-specific washout intervals. They cannot define dose-response or concentration-response thresholds for malformation risk. Small organ-specific signals must consequently be considered alongside diabetes severity, obesity, nephropathy, comparator choice, exposure misclassification, and post-discontinuation metabolic instability.
Large registry and EHR/EMR studies should be understood as prescription-, redemption-, dispensing-, or medication-order-based exposure studies unless ingestion is confirmed. Broad windows extending from months or years before conception into early pregnancy may include remote users with little biological exposure during organogenesis, while omitting treatment interruption, compounded or off-platform supply, dose escalation, and exact discontinuation. Combining long-acting semaglutide or tirzepatide with shorter-acting liraglutide or exenatide further obscures the fetal exposure window and may dilute, exaggerate, or misdirect risk estimates.
Comparator selection strongly influences the apparent direction of risk. General-population controls often yield higher crude adverse-outcome rates among exposed women because GLP-1RA users commonly have obesity, T2D, hyper
The literature often combines three different clinical questions. First, preconception metabolic optimization involves using a GLP-1RA or related incretin-based therapy before conception, followed by discontinuation. Potential benefits include lower weight, improved insulin resistance and glycemia, improved ovulatory function, and reduced cardio
Recent studies extend the question beyond congenital malformations to the metabolic consequences of stopping treatment. Maya et al[38] and Pondugula et al[37] suggest that women entering pregnancy after pharmacologic weight loss or improved glycemic control may experience a return of appetite, rebound weight gain, altered GWG, or worsening glycemic control after discontinuation, potentially contributing to GDM, HDP, or preterm birth. Counseling should therefore address both fetal exposure and maternal metabolic stabilization. A transition plan should record the specific agent, last dose, GA, indication, pre-treatment and current BMI, recent HbA1c and glucose profile, renal status, blood pressure, and current therapy. Women with pregestational diabetes may need prompt insulin-based management, while those treated for obesity require structured nutrition, activity, GWG surveillance, glycemic monitoring, and timely pregnancy-compatible therapy when indicated.
Pregnancy-loss outcomes require careful separation. Spontaneous abortion, elective termination, ectopic pregnancy, stillbirth, and ongoing pregnancy are biologically and clinically distinct; combining them into broad composites obscures interpretation. Elective termination is particularly influenced by pregnancy intention, counseling, anxiety about teratogenicity, and referral bias and should not automatically be regarded as a toxicity signal. Likewise, spontaneous-abortion signals from pharmacovigilance remain hypothesis-generating because obesity, diabetes, PCOS, infertility, advanced maternal age, and poor glycemic control independently increase miscarriage risk.
Although the absence of an overall MCM signal is reassuring, renal, urinary, and genital signals in some large databases require surveillance. Their absolute differences are small, and exposed populations are metabolically imbalanced; residual confounding, coding artifacts, differential fetal surveillance, diabetes severity, hypertension, nephropathy, or a true association remain possible. These findings must also be considered in the context of diabetic embryopathy. Early dysglycemia, glycemic variability, oxidative stress, and metabolic instability can affect cardiac, caudal, renal, and uro
The exposure landscape increasingly includes compounded semaglutide and tirzepatide, off-label use, and emerging dual- or triple-incretin agonists. Tirzepatide is a dual glucose-dependent insulinotropic polypeptide/GLP-1RA, while investigational agents such as retatrutide also include glucagon-receptor agonism; the broader term incretin-based therapy is therefore more accurate when these agents are discussed together. Compounded products may be incom
After inadvertent preconceptional or early-pregnancy exposure, current evidence supports cautious reassurance rather than automatic pregnancy termination. Treatment should be discontinued when pregnancy is recognized, and maternal metabolic care should be transitioned promptly to pregnancy-compatible therapy. Women planning pregnancy may benefit from GLP-1RA treatment before conception when clinically indicated. Still, they should receive contraception and reproductive counseling, including possible restoration of fertility with weight loss, planned discontinuation, and transition to safer therapy. This is particularly important for long-acting agents whose exposure persists after the last dose. Intentional continuation during pregnancy is not supported outside research settings or formal pregnancy-safety surveillance. Figure 6 summarizes the implications of this scoping review.
Research should move from broad reassurance toward precise risk estimation. Prospective exposure registries are needed for semaglutide, tirzepatide, liraglutide, dulaglutide, and other agents to capture indication, dose, duration, adherence, last dose, washout interval, exposure window, weight trajectory, HbA1c, diabetes duration, hypertension, renal disease, PCOS, infertility treatment, and concomitant medications. Large active-comparator studies should emulate target trials where possible. Diabetes cohorts should use metabolically similar women receiving insulin or other pregnancy-com
This review maps the full human evidence spectrum rather than restricting synthesis to meta-analyzable cohorts. It separates case reports, pharmacovigilance, regulatory trial pregnancy data, comparative cohorts, and meta-analyses according to inferential value; distinguishes diabetes-specific, obesity-focused, and mixed-indication evidence; and applies JBI and AMSTAR-2 tools to contextualize methodological quality.
The review also has limitations. Google Scholar was used as a supplementary, high-sensitivity source, but its dynamic, relevance-ranked results are less reproducible than those of conventional bibliographic databases. Screening the first 500 records yielded more than 3 times as many records as the PubMed search but was not equivalent to a formal EMBASE or Cochrane Library search; conference abstracts, European pharmacovigilance reports, trial records, or registry analyses may therefore have been missed. Some included studies had overlapping populations, broad exposure definitions, incomplete adjustment, and limited outcome validation. The scoping design was appropriate for mapping heterogeneous evidence but did not yield a new pooled estimate; the rapidly evolving field will require periodic updates.
Human evidence on GLP-1RA exposure before conception and during early pregnancy is cautiously reassuring, with no consistent overall MCM signal. However, the evidence remains observational, heterogeneous, and strongly influenced by comparator choice, exposure-window definition, and confounding from diabetes, obesity, PCOS, infertility, and cardiometabolic disease severity. Available data support reassurance and structured follow-up after accidental early exposure but not planned or continued use during pregnancy. The clinical challenge extends beyond teratogenicity to safe discontinuation and metabolic transition, including prevention of rebound GWG, GDM, HDP, and preterm birth. Until more definitive data are available, incretin-based therapies should be prescribed to reproductive-age women with explicit contraception and preconception counseling, discontinued before planned conception according to agent-specific washout considerations or when pregnancy is recognized, and replaced with pregnancy-compatible metabolic therapy when needed.
| 1. | Tricco AC, Lillie E, Zarin W, O'Brien KK, Colquhoun H, Levac D, Moher D, Peters MDJ, Horsley T, Weeks L, Hempel S, Akl EA, Chang C, McGowan J, Stewart L, Hartling L, Aldcroft A, Wilson MG, Garritty C, Lewin S, Godfrey CM, Macdonald MT, Langlois EV, Soares-Weiser K, Moriarty J, Clifford T, Tunçalp Ö, Straus SE. PRISMA Extension for Scoping Reviews (PRISMA-ScR): Checklist and Explanation. Ann Intern Med. 2018;169:467-473. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 31615] [Cited by in RCA: 26070] [Article Influence: 3258.8] [Reference Citation Analysis (6)] |
| 2. | Moola S, Munn Z, Tufanaru C, Aromataris E, Sears K, Sfetcu R, Currie M, Lisy K, Qureshi R, Mattis P, Mu PF. Systematic reviews of etiology and risk. [cited 23 June 2026]. Available from: https://jbi-global-wiki.refined.site/space/MANUAL/355598596/7.+Systematic+reviews+of+etiology+and+risk. |
| 3. | Shea BJ, Reeves BC, Wells G, Thuku M, Hamel C, Moran J, Moher D, Tugwell P, Welch V, Kristjansson E, Henry DA. AMSTAR 2: a critical appraisal tool for systematic reviews that include randomised or non-randomised studies of healthcare interventions, or both. BMJ. 2017;358:j4008. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 7487] [Cited by in RCA: 7098] [Article Influence: 788.7] [Reference Citation Analysis (12)] |
| 4. | Williams J, Pomeroy NE, Pop-Busui R, Lash R, Douyon L, Chames M, Wyckoff J. Case Report: Exenatide Use During Pregnancy. Endocrinologist. 2009;19:119-121. [DOI] [Full Text] |
| 5. | Greco D. Normal pregnancy outcome after first-trimester exposure to liraglutide in a woman with Type 2 diabetes. Diabet Med. 2015;32:e29-e30. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 8] [Cited by in RCA: 43] [Article Influence: 3.9] [Reference Citation Analysis (0)] |
| 6. | Ivanišević M, Herman M, Horvatiček M, Lovrenčić Vučić M, Đelmiš J. Pregnancy outcome and liraglutide levels in serum and umbilical vein blood of a woman with type 2 diabetes. A case report. Gynaecol Perinatol. 2018;27:70-72. |
| 7. | Livadas S, Androulakis I, Angelopoulos N, Lytras A, Papagiannopoulos F, Kassi G. Liraglutide administration improves hormonal/metabolic profile and reproductive features in women with HAIR-AN syndrome. Endocrinol Diabetes Metab Case Rep. 2020;2020:19-0150. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 11] [Cited by in RCA: 11] [Article Influence: 1.8] [Reference Citation Analysis (1)] |
| 8. | Burlina S, Dalfrà MG, Caprino R, Lapolla A. A case report on use of dulaglutide during the first weeks of pregnancy in woman affected by type 2 diabetes mellitus. Acta Diabetol. 2023;60:137-138. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 19] [Reference Citation Analysis (0)] |
| 9. | Alghamdi A, Alsaeddi A, Malki H, Alsaedi A. A Case Report of a Pregnant Woman With Type 2 Diabetes Mellitus Using Dulaglutide During the First Trimester of Pregnancy. Cureus. 2023;15:e44644. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 8] [Reference Citation Analysis (0)] |
| 10. | Skov K, Mandic IN, Nyborg KM. Semaglutide and pregnancy. Int J Gynaecol Obstet. 2023;163:699-700. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 27] [Reference Citation Analysis (0)] |
| 11. | Wajtryt O, Dedecjus M. Pregnancy after exposure to GLP-1 receptor agonists. A case report and literature review. Endocr Abstr. 2023;90:EP532. [DOI] [Full Text] |
| 12. | Patel B, Healy A. Positive Pregnancy Outcome following First Trimester Dulaglutide Exposure in a Women with Type 2 Diabetes and PCOS: A Case Report. J Diabet Res Rev Rep. 2024;6:1-4. [DOI] [Full Text] |
| 13. | Molteni M, Lodigiani S, Teliti M, Rotondi M, Guazzoni V. Use of dulaglutide in a pregnant woman with type 2 diabetes until third trimester of pregnancy. Acta Diabetol. 2024;61:1491-1494. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 9] [Reference Citation Analysis (0)] |
| 14. | Okeke IG, Camarda AR, Okeke R, Chaughtai S. Semaglutide-induced Hyperemesis Gravidarum. JCEM Case Rep. 2024;2:luad167. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 7] [Reference Citation Analysis (0)] |
| 15. | Zhu XM, Huang QY, Chen Y, Zhou Y, Yang CH. Normal pregnancy outcome after first-trimester exposure to semaglutide in an overweight woman: a case report. Clin Res Commun. 2024;7:13. [DOI] [Full Text] |
| 16. | Abushanab D, Mohammed HR, Abdul Rouf PV, Thomas B, Al-Badriyeh D, Elkassem W, Al-Mohannadi H, Hanssens Y, Khan S, Mohammed S, Al Hail M. Exposure to High-Dose Liraglutide in a Pregnant Woman With Obesity. O G Open. 2025;2:e077. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 5] [Reference Citation Analysis (0)] |
| 17. | Burlina S, Cioccoloni D, Romanello G, Lombardi S. Dulaglutide plus dapaglifozin use until the second trimester of pregnancy in a woman with type 2 diabetes. Acta Diabetol. 2026;63:343-345. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 2] [Reference Citation Analysis (0)] |
| 18. | Huang A, Lu J, Pancotto N, Sekhon GS, Gossen IM, Reyes JM, Lazarescu R. Unmasking Euglycemic Diabetic Ketoacidosis: The Interplay of Pregnancy, Sepsis, and Glucagon-Like Peptide 1 Analog. Cureus. 2025;17:e78193. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 2] [Reference Citation Analysis (0)] |
| 19. | Ozdemir H. Safety of liraglutide during pregnancy: a case report. Poster 49. 36th ENTIS Conference, ENTIS2025 conference; 2025 Sep 11-14; Thessaloniki, Greece. |
| 20. | Jin YC, Wu YH, Ma XS, Shi W, Wu ML, Hu QQ. Semaglutide exposure in early pregnancy and pregnancy outcomes: A case report and review of literature. J Gynecol Obstet Hum Reprod. 2026;55:103133. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 2] [Reference Citation Analysis (0)] |
| 21. | Doğan ŞE, Kuşkonmaz ŞM, Koc G, Aypar E, Çulha C. Case Series: Exposure to Glucagon-like Peptide-1 Receptor Agonist in the First Trimester of Pregnancy in Two Siblings. Endocr Metab Immune Disord Drug Targets. 2024;24:1237-1239. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 1] [Cited by in RCA: 19] [Article Influence: 9.5] [Reference Citation Analysis (0)] |
| 22. | Banerjee M, Dasgupta S. Combined Glucagon-Like Peptide-1 Receptor Agonist (GLP-1RA) and Sodium-Glucose Cotransporter 2 Inhibitor (SGLT2i) Therapy to Restore Fertility in Patients With Obesity, Polycystic Ovary Syndrome, and Incident Type 2 Diabetes. Cureus. 2026;18:e102358. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 1] [Reference Citation Analysis (0)] |
| 23. | Morton A, He J. Pregnancy outcomes following first trimester exposure to semaglutide. Obstet Med. 2026;19:104-107. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 14] [Cited by in RCA: 12] [Article Influence: 12.0] [Reference Citation Analysis (0)] |
| 24. | Zhou J, Wei Z, Lai W, Liu M, Wu X. The safety profile of usage of glucagon-like peptide-1 receptor agonists in pregnancy: A pharmacovigilance analysis based on the Food and Drug Administration Adverse Event Reporting System. Br J Clin Pharmacol. 2025;91:1272-1280. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 1] [Cited by in RCA: 15] [Article Influence: 15.0] [Reference Citation Analysis (0)] |
| 25. | Parker CH, Slattery C, Brennan DJ, le Roux CW. Glucagon-like peptide 1 (GLP-1) receptor agonists' use during pregnancy: Safety data from regulatory clinical trials. Diabetes Obes Metab. 2025;27:4102-4108. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 27] [Cited by in RCA: 17] [Article Influence: 17.0] [Reference Citation Analysis (0)] |
| 26. | Chou YC, Weng SH, Cheng FS, Tseng CH, Hu HY, Liu CH. Unintentional periconceptional exposure to glucagon-like peptide-1 receptor agonists and adverse pregnancy outcomes: A nationwide cohort study in Taiwan. Diabetes Obes Metab. 2026;28:1420-1430. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 8] [Cited by in RCA: 6] [Article Influence: 6.0] [Reference Citation Analysis (0)] |
| 27. | Cesta CE, Rotem R, Bateman BT, Chodick G, Cohen JM, Furu K, Gissler M, Huybrechts KF, Kjerpeseth LJ, Leinonen MK, Pazzagli L, Zoega H, Seely EW, Patorno E, Hernández-Díaz S. Safety of GLP-1 Receptor Agonists and Other Second-Line Antidiabetics in Early Pregnancy. JAMA Intern Med. 2024;184:144-152. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 71] [Cited by in RCA: 136] [Article Influence: 68.0] [Reference Citation Analysis (0)] |
| 28. | Hanif M, Hays AG, Nagarajan JS, Sah SP, Weinstock RS, Taub CC. Efficacy and Safety of Glucagon-like Peptide-1 Receptor Agonist Use During the First Trimester in Pregnant Women With Type 2 Diabetes. Am J Cardiol. 2025;246:10-13. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 16] [Cited by in RCA: 14] [Article Influence: 14.0] [Reference Citation Analysis (0)] |
| 29. | Lewin Z, Snow S, Lee JA, Wilkie G. Glucagon-Like Peptide-1 Receptor Agonists among Pregnancies with Pregestational Diabetes and Its Relationship with Congenital Malformations. Am J Perinatol. 2026;43:984-988. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 1] [Cited by in RCA: 5] [Article Influence: 5.0] [Reference Citation Analysis (0)] |
| 30. | Bolek T, Turňová P, Janošova S, Péč MJ, Ságová I, Nagy N, Jurica J, Mokáň M, Samoš M. Effect of semaglutide with metformin for weight loss and fertility in polycystic ovary syndrome (PCOS) patients with obesity: A pilot prospective study. Clin Nutr ESPEN. 2026;71:102885. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 4] [Reference Citation Analysis (0)] |
| 31. | Guo H, Jing Y, Zhang Y, Song L, Wu W, Wang J, Wang M, Niu X, Wang M, Pan X, Wang T, Cui W, Sun B, Wang N. Prepregnancy GLP-1RA use improves maternal lipid metabolism via liver-secreted FGF21 during pregnancy in HFD-fed dams. Obesity (Silver Spring). 2025;33:1504-1517. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 7] [Reference Citation Analysis (0)] |
| 32. | Li R, Mai T, Zheng S, Zhang Y. Effect of metformin and exenatide on pregnancy rate and pregnancy outcomes in overweight or obese infertility PCOS women: long-term follow-up of an RCT. Arch Gynecol Obstet. 2022;306:1711-1721. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 35] [Reference Citation Analysis (0)] |
| 33. | Bartelt K, Stamp T, Gracianette M, Deckert J. Increased Risk of NICU Stays for Babies Born to Mothers Prescribed GLP-1 Medications in the 90 Days Before Pregnancy. Jun 20, 2024 [cited 23 June 2026]. Available from: https://epicresearch.org/articles/increased-risk-of-nicu-stays-for-babies-born-to-mothers-prescribed-glp-1-medications-in-the-90-days-before-pregnancy. |
| 34. | Dao K, Shechtman S, Weber-Schoendorfer C, Diav-Citrin O, Murad RH, Berlin M, Hazan A, Richardson JL, Eleftheriou G, Rousson V, Diezi L, Haefliger D, Simões-Wüst AP, Addor MC, Baud D, Lamine F, Panchaud A, Buclin T, Girardin FR, Winterfeld U. Use of GLP1 receptor agonists in early pregnancy and reproductive safety: a multicentre, observational, prospective cohort study based on the databases of six Teratology Information Services. BMJ Open. 2024;14:e083550. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 2] [Cited by in RCA: 81] [Article Influence: 40.5] [Reference Citation Analysis (0)] |
| 35. | Kolding L, Henriksen JN, Sædder EA, Ovesen PG, Pedersen LH. Pregnancy Outcomes After Semaglutide Exposure. Basic Clin Pharmacol Toxicol. 2025;136:e70021. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 18] [Reference Citation Analysis (0)] |
| 36. | Imbroane MR, LeMoine F, Nau CT. Preconception glucagon-like peptide-1 receptor agonist use associated with decreased risk of adverse obstetrical outcomes. Am J Obstet Gynecol. 2025;233:116.e1-116.e7. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 44] [Cited by in RCA: 36] [Article Influence: 36.0] [Reference Citation Analysis (0)] |
| 37. | Pondugula N, Culhane JF, Lundsberg LS, Partridge C, Merriam AA. Gestational Weight Gain and Hypertensive Disorders of Pregnancy With Prepregnancy and Early Pregnancy Glucagon-Like Peptide-1 Receptor Agonist Exposure. Obstet Gynecol. 2026;147:293-302. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 6] [Cited by in RCA: 14] [Article Influence: 14.0] [Reference Citation Analysis (0)] |
| 38. | Maya J, Pant D, Fu Y, James K, Batlle C, Hsu S, Soria-Contreras DC, Shook LL, Mow C, Hivert MF, Thaweethai T, Powe CE. Gestational Weight Gain and Pregnancy Outcomes After GLP-1 Receptor Agonist Discontinuation. JAMA. 2025;334:2186-2196. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 12] [Cited by in RCA: 40] [Article Influence: 40.0] [Reference Citation Analysis (0)] |
| 39. | Hviid KVR, Banasik K, Mortensen LH, Madsbad S, Strandberg-Larsen K, Geiker NRW, Westergaard D, Nielsen HS. Periconceptional GLP-1 receptor agonist exposure and obstetric outcomes: a Danish nationwide cohort study. Hum Reprod Open. 2026;2026:hoag015. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 6] [Cited by in RCA: 3] [Article Influence: 3.0] [Reference Citation Analysis (0)] |
| 40. | Broe A, Pottegård A, Munk-Olsen T, Hallas J, Bliddal M, Petersen I, Damkier P. Prescription Drugs in Pregnancy and Congenital Malformations: A Population-Based Safety Screening Study. Pharmacoepidemiol Drug Saf. 2025;34:e70211. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 3] [Cited by in RCA: 3] [Article Influence: 3.0] [Reference Citation Analysis (0)] |
| 41. | Brown JP, Huybrechts KF, Straub L, Patorno E, Seely EW, Bateman BT, Hernández-Díaz S. Continuing Glucagon-Like Peptide-1 Receptor Agonists Into the First Trimester of Pregnancy and Pregnancy Outcomes: A Target Trial Emulation Study Using Claims Information. Ann Intern Med. 2026;179:929-937. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in RCA: 1] [Reference Citation Analysis (0)] |
| 42. | Kodali LSM, Metlock FE, Nriagu BN, Brazile TL, Kwapong YA, Dastmalchi LN, Spitz JA, Sharma G. GLP-1 Receptor Agonist Exposure During Pregnancy: A Systematic Review and Meta-Analysis of Adverse Pregnancy Outcomes. JACC Adv. 2026;5:102649. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 1] [Cited by in RCA: 3] [Article Influence: 3.0] [Reference Citation Analysis (0)] |
| 43. | Armstrong BBS, Servidoni ACP, Martin GCC, Machado GF, Amador WFO, Yousif A. Evaluation of GLP-1 receptor agonists in obstetrics and perinatal outcomes: A systematic review and meta-analysis. J Gynecol Obstet Hum Reprod. 2025;54:103046. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 8] [Reference Citation Analysis (0)] |
| 44. | Hakim J, Rajesh D, Tello JA. Neonatal and obstetric outcomes following periconceptional exposure to glucagon-like peptide-1 receptor agonists: a systematic review and meta-analysis. Am J Obstet Gynecol. 2026;S0002-9378(26)00222. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 4] [Reference Citation Analysis (0)] |
| 45. | Muller DRP, Stenvers DJ, Malekzadeh A, Holleman F, Painter RC, Siegelaar SE. Effects of GLP-1 agonists and SGLT2 inhibitors during pregnancy and lactation on offspring outcomes: a systematic review of the evidence. Front Endocrinol (Lausanne). 2023;14:1215356. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 65] [Cited by in RCA: 128] [Article Influence: 42.7] [Reference Citation Analysis (1)] |
| 46. | Minis E, Stanford FC, Mahalingaiah S. Glucagon-like peptide-1 receptor agonists and safety in the preconception period. Curr Opin Endocrinol Diabetes Obes. 2023;30:273-279. [RCA] [PubMed] [DOI] [Full Text] [Full Text (PDF)] [Cited by in Crossref: 13] [Cited by in RCA: 25] [Article Influence: 8.3] [Reference Citation Analysis (0)] |
| 47. | Kuitunen I. Exposure to GLP-1 receptor agonists in early pregnancy did not increase the risk of congenital malformations. Acta Paediatr. 2024;113:2682-2683. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 4] [Reference Citation Analysis (0)] |
| 48. | Drummond RF, Seif KE, Reece EA. Glucagon-like peptide-1 receptor agonist use in pregnancy: a review. Am J Obstet Gynecol. 2025;232:17-25. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 81] [Cited by in RCA: 64] [Article Influence: 64.0] [Reference Citation Analysis (0)] |
| 49. | Mandal L, Andersen LU, Luef BM, Tanvig MH, Vinter CA. Impact of semaglutide exposure on fetal and neonatal outcomes in pregnant women: a systematic review. Eur J Obstet Gynecol Reprod Biol. 2026;317:114836. [RCA] [PubMed] [DOI] [Full Text] [Cited by in RCA: 5] [Reference Citation Analysis (0)] |
| 50. | Ozbek L, Shah E, Al-Shiab R, Inal A, Guldan M, Afsar B, Covic A, Kanbay M. Safety of GLP-1 and Dual GLP-1/GIP Receptor Agonists in Preconception, Pregnancy, and Lactation: A Systematic Review of Maternal, Fetal, and Neonatal Outcomes. Diabetes Obes Metab. 2026;28:4503-4528. [RCA] [PubMed] [DOI] [Full Text] [Cited by in Crossref: 5] [Cited by in RCA: 6] [Article Influence: 6.0] [Reference Citation Analysis (0)] |