Esercizio aerobico o di resistenza per migliorare il controllo glicemico e gli esiti della gravidanza nelle donne con diabete mellito gestazionale: una revisione sistematica

1. Introduction

Gestational diabetes (GDM) is defined as new onset maternal hyperglycemia in pregnancy that resolves after birth [ ]. Diabetes affects one in six pregnancies worldwide, of which the majority is GDM [ ]. The rate of GDM diagnosed in pregnancy has increased over the last decade, with predictions of further increases worldwide [ , ]. GDM is associated with increased maternal and fetal morbidity and mortality [ , ]. Women with GDM are at a significantly increased risk of developing type 2 diabetes (T2DM) in later life [ ]. Maintaining good glycaemic control can improve maternal and infant health outcomes [ ].

Exercise is considered to be an important component of the management of GDM [ , ]. Outside of pregnancy, structured exercise is effective in improving diabetic control in patients with insulin resistance [ , , ]. Pregnancy is a state of relative insulin resistance. Aerobic and resistance exercise have different metabolic effects and therefore both have a potential role in the treatment of insulin resistance seen in GDM [ , , ].

Many studies have looked at the effect of exercise in the prevention of hypertensive disorders of pregnancy [ , ], pelvic or back pain [ , , ], preterm birth [ , ], incontinence [ , ], and mental health disorders [ , ]. Most of the systematic reviews published on exercise in GDM have concentrated on exercise as a preventive measure to reduce the rate of GDM diagnosis [ , , ].

We hypothesise that the type of exercise (either strength, aerobic, or combination) has an impact on glycaemic control and therefore health outcomes for both the mother and fetus. The aim of this systematic review was to update and consolidate the evidence on the effect of exercise modality on both glycaemic control and obstetric outcomes in patients with GDM.

2. Materials and Methods

This systematic review of randomised controlled trials (RCTs) and cohort and case control studies was carried out following the protocol described in the Preferred Reporting Items for Systematic Reviews and Meta-Analysis (PRISMA) statement [ ]. Searches were initially performed from September to November 2019 and re-run in May 2021 prior to the final analyses so that further studies could be retrieved for inclusion. Electronic database searches were carried out using the following databases: Cochrane library, Embase, PubMed, Cochrane Central Register of Controlled Trial (CENTRAL), CINAHL, and Web of Science. Grey literature databases were also searched (Google Scholar and OpenGrey). Citation pearl indexing was also performed of previous systematic reviews, literature reviews, and guidelines. Studies were initially restricted to the last 15 years. There were no restrictions on language.

### 2.1. Data Sources and Search Strategy

Two reviewers (NK and CC) performed a search to identify RCTs that studied the effect of exercise (aerobic/resistance/any exercise) on glycaemic control or obstetric outcomes in pregnancies affected by GDM. The search strategy followed the PICO framework, using key words, free text, and MeSH terms as appropriate and combining Boolean operators of (AND/OR/NOT/quotation marks/brackets): Participants: women, pregnancy, gestational diabetes, hyperglycaemia, diabetes Intervention: exercise, aerobic, resistance Comparison: physical activity, aerobic, resistance, control Outcome: medication (insulin, metformin), glycaemic control, maternal outcome

Inclusion criteria were as follows: Randomised controlled trials Gestational diabetes mellitus Intervention of resistance exercise or aerobic exercise alone or in combination Comparator or control of either resistance, aerobic or no exercise

Exclusion criteria were: Review or opinion articles Studies without published results Studies involving women with pre-existing diabetes

The primary outcome was glycaemic control in women with GDM (defined as average blood glucose levels or use of insulin and insulin requirements where it was required to maintain euglycaemia). Secondary outcomes were obstetric outcomes including rate of caesarean birth, perineal trauma, or duration of labour.

### 2.2. Study Selection

All articles retrieved for review were imported to Endnote and duplicates were removed. Titles, abstracts, and full text were independently reviewed by two reviewers (N.K. and C.C.) with reference to a third reviewer (M.F.H.) if required. Inclusion and exclusion criteria are reported in .

### 2.3. Data Extraction

Data extraction was independently performed by two researchers (N.K. and M.F.H.) using standardised electronic data extraction forms, saved on a shared drive after initial assessment, with reference to a third researcher (C.C.) as required. Missing data was to be requested from study authors. Data extraction was performed in accordance with the Cochrane Handbook for Systematic Review of Interventions [ ]. Extracted data items included study reference details, study context, study design, study population, data analysis and methods, and research findings.

### 2.4. Quality Assessment

Bias of randomised trials was assessed using the Cochrane Risk of Bias (RoB2) tool [ ]. This study was registered with PROSPERO (2020 CRD42020161454) [ ] and is reported following the PRISMA statement for systematic reviews [ ].

3. Results

### 3.1. Literature Search

Flow of studies through the stages of identification, screening, eligibility, and final inclusion is shown in . Fourteen randomised controlled studies were selected for inclusion in this systematic review [ , , , , , , , , , , , , , , ]. Two of the papers [ , ] reported different outcomes from one trial and for the purpose of this manuscript are referred to as one paper.

### 3.2. Characteristics

Characteristics of the fourteen studies are shown in with risk of bias summary in and results in . Study sizes ranged from six participants to 200 participants. In total, over 758 pregnant women (mother-baby pairs) are included in this systematic review. Length of intervention ranged from four to twelve weeks. Of the 14 papers analysed, five used aerobic exercise as the intervention, five used strength/flexibility-based exercise, two used a combined intervention, one did not specify the type of exercise used, and one compared strength or aerobic exercise with a control. Interventions ranged from supervised follow-up with a kinesiologist to cycling, aerobic exercises, walking, yoga, or combined aerobic and resistance exercises.

No randomised studies were identified that directly compared aerobic to resistance exercise. Study summaries are shown in . One study compared aerobic or resistance exercise to controls, but the numbers were small, and the study was reported as a pilot study within a conference abstract, with plans to continue the study [ ]. The most common comparison was “usual physical activity”.

### 3.3. Fasting Glucose

Seven of the studies included fasting glucose as a primary outcome. Two papers [ , ] showed an improvement in fasting glucose in the intervention group. Youngwanichseta et al. [ ] showed a statistically significant difference in fasting plasma glucose with intervention vs. control (post-test mean 83.39 ± 7.69 mmol/L vs. 87.85 mmol/L ± 7.94, p = 0.012). Jovanovic-Peterson [ ] et al. found a reduction in fasting glucose with intervention vs. control (70.1 mmol/L ± 6.6 vs. 87.6 mmol/L ± 6.2, p < 0.001). The remaining showed no difference in fasting glucose with exercise.

### 3.4. Postprandial Glucose

Seven of the papers looked at postprandial glucose as an outcome measure. Five papers [ , , , , ] showed an improvement in postprandial glucose results in the intervention group. In Bo’s [ ] study, postprandial glucose in the intervention group was 117.2 ± 16.5 mmol/L vs. 106.1 mmol/L ± 19.0 in the control group, giving an adjusted difference of minus 11.1 mmol/L (minus 16.1, minus 6.1, 95% CI, p < 0.001). In Halse’s [ ] paper, there was a significant difference in postprandial glucose at breakfast, with lower levels in the intervention group compared with the control, ( p = 0.046); postprandial glucose levels at dinner approached significance with lower levels in the exercise group, ( p = 0.054), compared with the control. There was no difference in postprandial levels at lunch, ( p = 0.312). Sklempe-Kokic et al. [ ] showed an improvement in postprandial blood glucose in the intervention vs. control (mean 4.66 mmol/L ± 0.46 vs. 5.3 mmol/L ± 0.47, p < 0.001). In Youngwanicheta’s [ ] paper, there was statistically significant difference in two hour postprandial glucose in the intervention vs. control (mean post-test glucose 103.67 mmol/L ± 9.93 vs. 114.36 mmol/L ± 10.15, p = 0.001). Brankston et al.’s [ ] study showed a significant reduction in pooled post-meal glucose levels in the diet plus exercise group compared to diet alone with no difference in individual breakfast, lunch, and supper postprandial glucose levels. Two studies ([ , ]) showed no difference in postprandial blood sugar.

### 3.5. Average Glucose

Five studies reported average glucose levels as an outcome. One study, by Halse [ ], showed an improvement in mean postprandial glucose levels in the exercise group compared with the control ( p = 0.004). The remaining four studies showed no difference [ , , , ].

### 3.6. HbA1C

Five papers recorded HbA1C as an outcome with four of these ([ , , , ]) showing a reduction in HbA1C with exercise intervention. In Bo’s [ ] paper, HbA1C reduced from 4.9% ± 0.4 in the non-exercise group to 4.6% ± 0.5 in the exercise group with an adjusted difference of minus 0.3, p < 0.001. In Youngwanichseta’s [ ] paper, there was a significant difference in post-test HbA1C with intervention vs. control (mean HbA1C 5.23% ± 0.22 vs. 5.68% ± 0.38, p = 0.03). Qadir [ ] showed a significant difference in average HbA1c between the intervention and control groups (4.9% vs. 5.38%, p = 0.04). The paper by Ramos [ ] is a conference abstract reporting the preliminary results of a pilot study. Post intervention mean HbA1c in the study group ( n = 2) was 5.5% ± 0.4 compared with 6.3% ± 4.0 in the control group ( n = 4). Halse [ ] showed was no difference in HbA1C post study between intervention and control.

### 3.7. Insulin Use

Six papers studied the need for insulin treatment as an outcome, with one paper showing a reduction in the number of women requiring insulin with exercise intervention. There was a significant difference in De Barros’s paper [ ] in the number of women who required insulin: 21.9% in the exercise group (7/32) vs. 56.3% (18/32) in the control group, p = 0.005. Five studies showed no difference [ , , , , ].

Four papers reported changes in insulin dose. One paper (Brankston [ ]) showed a reduction in insulin dose requirements with exercise intervention (diet plus exercise) vs. diet alone (0.22 units/kg ± 0.2 vs. 0.48 units/kg ± 0.3, p < 0.05). Halse’s paper [ ] reported a clinical difference with a lower mean dose of insulin in the intervention (7 iu ± 1) group compared with the control (13 iu ± 1); the numbers of participants were too small to draw a statistically meaningful conclusion. Adam [ ] and deBarros [ ] showed no difference in mean dose of insulin between exercise intervention and control. Four studies reported latency to insulin requirement as an outcome. One paper [ ] showed a statistically significant delay in starting insulin with intervention vs. control (3.71 weeks ± 3.1 vs. 1.11 weeks ± 0.8, 0 < 0.05), with three studies showing no difference [ , , ].

### 3.8. Maternal Hypoglycaemia

There were no reports of maternal hypoglycaemia in the 14 papers analysed.

### 3.9. Caesarean Section

Six papers reported rate of Caesarean section (CS) as an outcome. There was a significant difference in Awad’s [ ] paper: the rate of CS was higher in the control group (63.4% ( n = 19/30) compared with 16% ( n = 5/30) in the exercise group, ( p = 0.001) . Five studies showed no difference in CS rates between the exercise intervention and control groups [ , , , ].

### 3.10. Induction of Labour Rates and Labour Duration

Halse [ ] was the only paper to report on induction of labour with no difference shown between intervention (55%) and control (42%), p > 0.05. They were also the only group to report on duration of labour with no difference between exercise intervention and control (445 min ± 309 vs. 348 min ± 187, p > 0.05)

### 3.11. Other Outcomes

None of the studies included reported on the use of Metformin, perineal trauma (third- or fourth-degree tears), or shoulder dystocia as an outcome.

4. Discussion

There were no well-designed RCTs identified comparing aerobic and strength exercise in the management of GDM and therefore we are unable to comment on our primary outcome, but we were able to provide a narrative review for glycaemic control parameters and maternal outcomes related to exercise.

Pregnancy is a state of relative insulin resistance. Exercise has its insulin sensitising effects by increasing GLUT-4, increasing sensitivity GLUT-4 to insulin and increased glycogen synthase [ ]. By this mechanism, uptake of glucose into muscles is increased. Aerobic and resistance exercise have different metabolic effects and therefore both have a potential role in the treatment of states of insulin resistance such as that seen in GDM. As skeletal muscle is the largest mass of insulin sensitive tissue, an increase in muscle mass through resistance training is associated with improved glycaemic control [ , ]. Aerobic exercise reduces visceral obesity which improves insulin sensitivity [ ]. Aerobic exercise has been shown to be more beneficial in modulating insulin resistance and inflammatory cytokines in obese patients with T2DM when compared with resistance exercise [ ]; however, it has been proposed that both methods likely incur benefit through their different mechanisms of action [ ].

Previously, women had been advised against moderate exercise in pregnancy due to concerns about the risk of harm to the fetus including growth restriction and preterm birth. A systematic review [ ] of over 2000 women found that in singleton, uncomplicated pregnancies, moderate exercise did not increase the risk of preterm birth or growth restriction. Other systematic reviews and meta-analyses have supported this finding [ , ], including showing a reduction in the rate of caesarean birth and gestational weight gain with exercise [ ]. The American College of Obstetricians and Gynecologists (ACOG) [ ] recommend 30 min or more of moderate exercise on most days for women without medical or obstetric risks. In one study [ ], over half of women believed that weight-based exercise was unsafe in pregnancy. A lack of awareness of exercise in pregnancy guidelines among physicians has previously been shown [ ]. As this review shows, there were no adverse effects reported in women with GDM, i.e., a pregnancy that may be considered “high risk”. A meta-analysis of women with risk factors such as obesity, hypertension, and GDM showed no adverse effect on the fetus with moderate intensity exercise [ ]. An RCT largely comprised of previously sedentary women showed no increase in maternal or neonatal adverse effects with combined aerobic and resistance training [ ]. Studies looking at the role of exercise in preventing GDM have shown conflicting results, with some studies showing a reduced incidence [ ] and others showing no impact [ , ].

Of the studies analysed, eight out of the 14 showed some benefit with exercise intervention on parameters of glycaemic control, particularly lower fasting glucose [ , ], postprandial glucose [ , , , , , , ], HbA1c [ , , , ], reduced need for insulin, and increased latency to starting insulin [ , ]. Different makers of glycaemic control were used which makes it difficult to compare results. Of the seven studies that included resistance training either alone or in combination, four of these demonstrated improved glycaemic control with intervention [ , , , ]. Two of these studies involving strength-based exercise as an intervention showed a reduction in the need for insulin or delayed the onset of starting insulin. The need for insulin treatment in GDM is an important clinical indicator of the degree of hyperglycaemia. Hyperglycaemia in pregnancy is associated with adverse outcomes [ ] and pregnant women who require insulin as treatment are at a higher risk of adverse outcomes [ ]. The impact of resistance exercise on eliminating the need for or delaying starting insulin in GDM is an important finding and more research supporting this would be helpful in advising patients.

Of the five studies using aerobic exercise alone as an intervention, three [ , , ] of these showed some improvement in glycaemic control. Of note, none of the studies included showed any negative effect of exercise. There is insufficient evidence to recommend one modality of exercise over another for the management of GDM and in practice, a combination of aerobic and strength is likely to show the most overall benefit in pregnancy as has been shown in the management of Type 2 DM [ ].

The potential positive effect of exercise in GDM pregnancies on obstetric outcomes is largely theoretical modelling of the benefits of exercise in reducing hyperglycaemia. The studies presented have had mixed reports with many studies showing no difference with intervention. One study [ ] showed a reduction in the CS rate among women with insulin controlled GDM who participated in a combined aerobic and strength exercise regime compared with standard care. When considering the global impact of GDM, it is notable that there are relatively few studies published in exercise as a treatment. Exercise is generally a free or low-cost intervention and is safe with few, if any, potential adverse effects and can be modified to all degrees of fitness and physical ability. Of the studies presented here, they appear to have been powered to look at variation in glycaemic control but were perhaps not large enough to report on maternal outcomes.

A core outcome set (COS) for studies of GDM was published in 2020 [ ] with 14 outcomes reported in the final set. All the papers included in this systematic review were published before the development of a core outcome set and much variation is seen in the outcomes measured; this reiterates the need for such a COS for consistency in reporting outcomes in publications.

Several systematic reviews have been published on the topic of exercise for treatment for GDM. Cremona [ ] looked at the effect of exercise modality on markers of insulin sensitivity in women with or at risk of GDM. They concluded that exercise sessions three times per week of either aerobic or strength training targeting major muscle groups could improve glycaemic control. Women with a high body mass index (BMI) at risk of GDM would also benefit; however, women with previous GDM pregnancies and a normal BMI do not appear to reduce their risk of GDM through exercise intervention. Huang [ ] studied the effects of difference exercise modalities on glycaemic control alone but did not include five studies included in this review [ , , , , ]. They performed a meta-analysis of nine RCTs involving 618 women which demonstrated high heterogenicity. The results showed that aerobic exercise reduced the fasting blood glucose, postprandial blood glucose, and HbA1C in patients with GDM compared to conventional treatment. They included Youngwanichestha et al.’s [ ] study as an aerobic intervention although yoga is not generally considered aerobic as the intention is not to increase the heart rate. The dosage of insulin was reduced in the resistance exercise group compared with conventional treatment; however, this was based on two studies [ , ]. The combination of aerobic and strength exercise compared with conventional treatment reduced postprandial glucose levels, but this was based on only one study [ ]. Brown [ ] performed a systematic review of 11 studies involving 638 women and concluded that the evidence is poorly reported and confounded by the variety of exercise interventions. Harrison [ ] studied eight studies with self-monitored postprandial glucose as the primary outcome and concluded that exercise performed at moderate intensity a minimum of three times a week appears to be effective in managing GDM, although again this review was limited by the limited number of studies, heterogenicity in study design, and reporting of outcomes. Bgengski [ ] included the same eight studies in their systematic review but only studied the effect of exercise on fasting plasma glucose as their primary outcome, concluding there was no difference between exercise and physical activity counselling compared with standard care on fasting plasma glucose levels. There was no difference on secondary outcomes which were macrosomia, preterm birth, CS, GA at delivery, and birth weight. Allehandan [ ] included eight studies in their systematic review and concluded that diet plus exercise for women with GDM lowered fasting and postprandial glucose levels compared with diet alone.

Only two of the systematic reviews in this area [ , ] included both maternal outcomes as well as glycaemic control. While glycaemic control is an important predictor of morbidity in GDM, it is crucial that obstetric and neonatal outcomes are reported, as ultimately this is the information that will be used to counsel patients and inform guidelines. Some abstracts from the grey literature included in this analysis have not been reported in previous systematic reviews. Given the lack of large RCTs, it is important that smaller studies such as those published at conferences be included in systematic reviews to allow a complete review of the available literature.

### Strengths and Limitations

A strength of this study is the inclusion of studies not included in previous systematic reviews and inclusion of obstetric outcomes which have often been excluded previously. As with previous systematic reviews, it is limited by the small number of studies and the lack of large RCTs. We could not identify a study that directly compared strength training with aerobic exercise in the management of GDM and therefore cannot answer our primary research question.

5. Conclusions

Studies on exercise for the management of GDM have shown mixed effects on maternal glycaemic control, with disappointingly no apparent impact on pregnancy outcomes.

Despite exercise being integral to GDM management, the evidence bases for medium to long term benefit remains lacking. This is an important topic for discussion that reveals a paucity of data to guide healthcare providers in recommendations on exercise modality for the management of GDM. Further well designed large RCTs using a core outcome set are needed to determine the most efficient way to use exercise to treat GDM. Ideally, future studies would continue into the postpartum period to determine the effect of resistance or aerobic exercise on long term progression to T2DM following a pregnancy with GDM.

The development of a COS for diabetes will improve reporting in studies on the role of exercise in GDM. The heterogenicity of reported studies make it difficult to make specific recommendations on the optimum exercise regime. Given the different effects aerobic and strength training have on glucose metabolism, it is plausible that a combination of both modalities is useful in controlling GDM.