Impact of the Menstrual Cycle on Glycemic Control in Women with Type 1 Diabetes and the Potential Role of AHCL Systems

Abstract

Background/Objectives: Hormonal fluctuations during the menstrual cycle can affect glycemic control in women with type 1 diabetes (T1D), especially during the luteal phase, when increased insulin resistance may lead to prolonged hyperglycemia. Advanced Hybrid Closed-Loop (AHCL) systems could help manage these hormone-driven fluctuations. This study aimed to assess glycemic control across menstrual phases and explore the role of AHCL systems in counteracting the related glucose variability. Methods: A retrospective study was conducted including women with T1D and regular menstrual cycles (study group) and women on estroprogestin therapy (control group). Each group was subdivided by insulin delivery method (AHCL vs. non-AHCL). Glycemic metrics and insulin requirements were compared between the follicular and luteal phases, and between groups. Results: The study included 94 women (62 in the study group, 32 in the control group). In the study group, glycemic control worsened during the luteal phase, with increased average glucose, glycemic variability, and time above range > 250 mg/dL (+0.93%, p = 0.03) and reduced time in range 70–180 mg/dL. These changes were more pronounced among AHCL users, who also showed a significant increase in bolus insulin. No phase-related differences were observed in the control group or among non-AHCL users. Significantly higher insulin needs during the follicular phase were found in the study group compared with the controls. Conclusions: This study confirmed a worsening in glycemic control in women affected by T1D during the luteal phase of the menstrual cycle, suggesting a need for more tailored management. The clear efficacy of AHCL systems in counteracting hormone-related glycemic fluctuations has not been proved, highlighting the need for further research in larger, more homogeneous cohorts.

1. Introduction

The relationship between glycemic control and the menstrual cycle in young women with type 1 diabetes (T1D) has been studied extensively over the years [1,2]. A dominant role is played by sex hormones, which are known to have a significant influence on glucose homeostasis, although the pathophysiological mechanisms underlying these processes remain incompletely understood [2,3]. The introduction of tools such as continuous glucose monitoring (CGM) has represented a significant step forward in analyzing the effects of the different phases of the menstrual cycle on glycemic trends and insulin requirements in young women with T1D. Several studies have demonstrated, in patients with T1D, an increased frequency of hyperglycemia during the luteal phase and the periovulatory phase, compared with the follicular phase [4,5,6]. This condition is associated with reduced insulin sensitivity typical of these phases [6]. More recently, the effects of closed-loop insulin pumps on blood glucose levels throughout the menstrual cycle have been analyzed [7,8,9,10]. Even in these studies, although limited to small cohorts of patients, it was confirmed that glycemic control, as reflected by time in range (TIR), was better during the follicular phase compared with the luteal phase. However, only one of these studies reported a similar insulin requirement across the different menstrual phases in women who used closed-loop insulin delivery systems [8]. When specifically analyzing the impact of Advanced Hybrid Closed-Loop (AHCL) systems on glycemic control in women with T1D, it is highlighted that TIR improved across all phases of the menstrual cycle, but the performance of AHCL systems differed according to the phase. In particular, TIR became significantly higher during the early follicular phase than during the late luteal phase [7,9]. Thus, although the available findings suggest poorer glycemic control during the luteal phase, these results were derived from small, non-homogeneous cohorts, as evidenced by the different insulin regimens and devices used. Furthermore, there is limited evidence regarding glycemic fluctuations in women of reproductive age receiving contraceptive therapy who are not subject to physiological hormonal fluctuations and who could serve as a natural control group.

The aim of this study was to investigate the discrepancy in glycemic control between the luteal and the follicular phases in women of childbearing age with T1D using CGM. Furthermore, the impact of AHCL use on this outcome was explored.

2. Materials and Methods

2.1. Design of the Study

A retrospective study involving a cohort of women with T1D followed by the Diabetes Unit of IRCCS Istituto Giannina Gaslini (Genoa, Italy) and IRCCS Policlinico San Martino University hospital (Genoa, Italy) was conducted. The aim of the study was to compare glycemic control during the follicular and luteal phases in a study group of T1D women who had regular menstrual cycles and a control group of T1D women undergoing estroprogestin therapy (three weeks on therapy and one week off per month), since they were not exposed to physiological hormonal fluctuations but still simulated regular menstrual cycle timing.

2.2. Participants

According to our clinical records, a total of 143 female patients with T1D were identified as meeting the following inclusion criteria: age between 18 and 45 years, diagnosis of T1D for at least 12 months, use of a CGM device for at least 1 month, and regular menstrual cycles or estroprogestin medication use, with three weeks on therapy and one week off per month. The study was proposed to all identified patients, who all agreed to participate in screening for exclusion criteria, which considered their menstrual cycle duration and their disease history. Patients with amenorrhea (absence of menstrual cycles for at least 6 months), oligomenorrhea (menstrual cycles longer than 35 days), and polymenorrhea (menstrual cycles shorter than 21 days) were excluded. Additional exclusion criteria were pregnancy, symptoms of perimenopause (hot flushes, mood swings, sleep disturbances), previous diagnosis of polycystic ovary syndrome (PCOS), hyperprolactinemia, treatment with other hypoglycemic medications or corticosteroids during the period considered in the study, and continuous estroprogestin therapy (i.e., without 7-day off-therapy periods every 28 days), as this regimen leads to the suppression of menstrual cycles. Moreover, patients for whom glycemic data could not be downloaded were excluded. Applying the exclusion criteria, a total of 94 women were included in the study. A comprehensive flowchart of the recruitment process is shown in Figure 1.

2.3. Definition of Follicular and Luteal Phases

The follicular phase of the menstrual cycle exhibits marked variability, whereas the luteal phase demonstrates less variability and typically lasts approximately 14 days under physiological conditions [11]. Therefore, during the scheduled follow-up appointments in the outpatient clinic, patients were asked to report the dates of the first day of bleeding during their two previous consecutive menstrual cycles. The luteal phase was considered to begin 14 days before the onset of the second menstrual cycle and to end on its first day. The follicular phase was considered to span from the first day of the first menstrual cycle until the beginning of the luteal phase.

2.4. Outcomes

In accordance with the clinical practice protocols at both centers, patients’ weight was assessed. Moreover, the patients’ height was measured and then the body mass index (BMI) was calculated. The ambulatory glucose profile (AGP) data were downloaded using software provided by the CGM or AHCL system manufacturers (Glooko^®, Dexcom Clarity^®, Carelink^®, Libreview^®). Two different data downloads were performed, one for each phase of the menstrual cycle. As recommended by the Standard of Care for Diabetes 2025 [12], the following ambulatory glucose profile parameters were analyzed: time in range 70–180 mg/dL (TIR), time above range 180–250 mg/dL (TAR), time above range > 250 mg/dL (TAR > 250 mg/dL), time below range 54–70 mg/dL (TBR), time below range < 54 mg/dL (TBR < 54 mg/dL), coefficient of variation (CV), and average glucose. When available, the total daily insulin intake, the bolus insulin intake, and the basal insulin intake were considered and additionally computed according to body weight.

2.5. Statistical Analysis

A descriptive analysis was performed: continuous variables were reported as mean ± standard deviation (SD), median, and range, and categorical variables were summarized as absolute frequencies and relative frequencies. The analysis of continuous data across different time points in the study (follicular and luteal phases) was performed using non-parametric paired tests (Wilcoxon test). All p-values were calculated using two-tailed tests, with statistical significance set at a p-value of less than 0.05. Statistical analyses were conducted using SPSS software, version 20, for Windows (SPSS Inc., Chicago, IL, USA).

3. Results

3.1. Patients’ Characteristics

Of the 94 patients included in the study, 62 were not using estroprogestin medication (study group), while the remaining 32 were taking estroprogestin (control group). Among those included in the study group, 39 patients (62.9%) were administered insulin via an AHCL system, while 23 patients (37.1%) used other insulin delivery methods. In the control group, 26 patients (81.3%) used AHCL systems, while 6 patients (18.7%) used different insulin therapies. Of note, only two patients enrolled in the study group using AHCL had specific settings adjusted for menstrual cycle days. Neither the study group of non-AHCL users nor the entire control group included any women with dedicated settings for menstrual cycle days. The study group demonstrated a mean age of 25.24 ± 6.88 years, with no significant differences compared with the control group (25.41 ± 4.13). No statistically significant differences were observed between the two groups regarding weight, height, BMI, cycle duration, or duration of CGM/AHCL use. A detailed description of patients’ characteristics, the CGM devices used, and the insulin delivery methods is provided in

Table 1. Patients’ characteristics and CGM/AHCL usage.

	Study Group	Control Group *	p Value
Number	62	32
Age (years)	25.24 ± 6.88	25.41 ± 4.13	0.248
Weight (kg)	65.32 ± 11.88	61.20 ± 11.06	0.068
Height (m)	1.64 ± 0.06	1.63 ± 0.82	0.477
Body mass index (kg/m²)	24.15 ± 4.02	23.00 ± 3.38	0.133
Cycle lenght (days)	28.84 ± 3.05	28.13 ± 1.83	0.228
CGM			0.890
Dexcom G6	37 (59.7%)	21 (65.6%)
Guardian 4	20 (32.3%)	9 (28.1%)
Dexcom one	1 (1.6%)	0 (0%)
Guardian 3	1 (1.6%)	1 (3.1%)
Freestyle libre 2	3 (4.8%)	1 (3.1%)
Insulin delivery method			0.219
Minimed 780	21 (33.0%)	10 (31.3%)
Tandem Control-IQ	18 (29.0%)	15 (46.9%)
Omnipod dash	13 (21.0%)	3 (9.4%)
Omnipod 5	0 (0%)	1 (3.1%)
Glucomen day pump	1 (1.6%)	1 (3.1%)
MDI	9 (14.5)	2 (6.3%)
CGM/AHCL use duration (months)	32.95 ± 18.71	39.97 ± 24.10	0.128

* Patients on estroprogestin medication. Continuous variables are expressed as mean ± standard deviation, while categorical variables are expressed as count and frequency.

.

3.2. Comparison of Glycemic Control Between the Follicular and Luteal Phases

The differences in glycemic control between the luteal phase and the follicular phase were assessed in the study group, indicating poorer glycemic management in the luteal phase, as indicated by TIR (−1.73% in the luteal phase), TAR (+1.08% in the luteal phase), TAR > 250 mg/dL (+0.93% in the luteal phase), average glucose, and CV. However, statistical significance (p = 0.03) was only achieved for TAR > 250 mg/dL, indicating a longer duration of severe hyperglycemia in the luteal phase compared to the follicular phase (

Table 2. Comparison of glycemic control between the follicular and luteal phases in the study group.

Study Group (n = 62)
	Follicular Phase	Luteal Phase	p Value
TIR	66.02 ± 13.24	64.29 ± 12.88	0.07
TAR	23.05 ± 8.32	24.13 ± 8.18	0.16
TAR > 250 mg/dL	8.05 ± 6.78	8.98 ± 6.99	0.03
TBR	2.06 ± 2.19	1.95 ± 2.34	0.48
TBR < 54 mg/dL	0.63 ± 0.75	0.66 ± 0.90	0.82
Average glucose	159.26 ± 19.60	161.66 ± 20.08	0.08
CV	34.82 ± 5.49	35.18 ± 5.46	0.48
Total daily insulin requirement (U/kg/day)	0.64 ± 0.17	0.66 ± 0.17	0.13
Basal insulin requirement (U/kg/day)	0.30 ± 0.83	0.30 ± 0.82	0.69
Bolus insulin requirement (U/kg/day)	0.34 ± 0.12	0.35 ± 0.12	0.07

Continuous variables are expressed as mean ± standard deviation. Data regarding insulin intake were available for 47 patients. Bold means statistically significant.

). In contrast, no statistically significant difference in insulin requirement was identified (data available only for 47 patients), although a trend towards increased total daily insulin (TDI) requirements (expressed in U/kg/day) in the luteal phase was observed (Table 2).

As expected, no differences in glycemic control and variability or in insulin intake were observed when comparing the luteal phase and the follicular phase in the control group (

Table 3. Comparison of glycemic control between the follicular and luteal phases in the control group.

Control Group (n = 32)
	Follicular Phase	Luteal Phase	p Value
TIR	69.16 ± 9.66	69.13 ± 9.61	0.96
TAR	21.31 ± 6.82	21.56 ± 5.75	0.86
TAR > 250 mg/dL	7.06 ± 4.41	7.25 ± 5.57	0.84
TBR	1.84 ± 1.71	1.72 ± 1.46	0.43
TBR < 54 mg/dL	0.66 ± 1.04	0.34 ± 0.55	0.07
Average Glucose	155.78 ± 16.65	156.25 ± 15.74	0.95
CV	35.09 ± 3.89	34.64 ± 4.04	0.46
Total daily insulin requirement (U/kg/day)	0.74 ± 0.20	0.75 ± 0.21	0.47
Basal insulin requirement (U/kg/day)	0.32 ± 0.10	0.33 ± 0.94	0.33
Bolus insulin requirement (U/kg/day)	0.42 ± 0.13	0.42 ± 0.14	0.52

Continuous variables are expressed as mean ± standard deviation. The data on insulin intake were available for 28 patients.

).

Subsequently, the study group and the control group were divided into two subgroups to assess the influence of AHCL system utilization on glycemic control and variability. Among the 39 patients in the study group using AHCL, glycemic control in the luteal phase was observed to be poorer than in the follicular phase, as suggested by a significant decrease in TIR (−2.71% in the luteal phase; p = 0.03). Accordingly, there was a notable increase in TAR, TAR > 250 mg/dL, and average glucose during the luteal phase. Furthermore, the bolus insulin requirement showed a significant increase in the luteal phase compared with the follicular phase (+0.02 U/Kg/day, p = 0.02), as shown in

Table 4. Comparison of glycemic control between the follicular and luteal phases in the study group in patients using AHCL.

Study Group, AHCL Users (n = 39)
	Follicular Phase	Luteal Phase	p Value
TIR	69.74 ± 10.79	67.03 ± 11.57	0.03
TAR	21.64 ± 7.58	23.74 ± 8.19	0.03
TAR > 250 mg/dL	6.26 ± 4.78	7.56 ± 5.23	0.02
TBR	1.59 ± 1.43	1.28 ± 0.10	0.06
TBR < 54 mg/dL	0.46 ± 0.68	0.38 ± 0.54	0.63
Average glucose	155.72 ± 15.36	159.90 ± 17.11	0.04
CV	33.82 ± 5.57	33.63 ± 4.60	0.72
Total daily insulin requirement (U/kg/day)	0.65 ± 0.18	0.68 ± 0.18	0.08
Basal insulin requirement (U/kg/day)	0.30 ± 0.08	0.30 ± 0.08	0.49
Bolus insulin requirement (U/kg/day)	0.35 ± 0.12	0.37 ± 0.12	0.02

Continuous variables are expressed as mean ± standard deviation. The data on insulin intake were available for 38 patients. Bold means statistically significant.

.

When comparing the luteal phase to the follicular phase in the patients using AHCL among the control group, no appreciable differences in insulin intake, glycemic control, or glycemic variability were observed, as expected (Supplementary Table S1). Similarly, no significant differences in glycemic control or insulin requirement were observed between the two phases among patients who were not using AHCL, both in the study group (

Table 5. Comparison of glycemic control between the follicular and luteal phases in the study group in patients not using AHCL.

Study Group, Non-AHCL Users (n = 23)
	Follicular Phase	Luteal Phase	p Value
TIR	59.70 ± 14.78	59.65 ± 13.89	0.89
TAR	25.43 ± 9.12	24.78 ± 8.30	0.81
TAR > 250 mg/dL	11.09 ± 8.52	11.39 ± 8.87	0.58
TBR	2.87 ± 2.96	3.09 ± 3.36	0.56
TBR < 54 mg/dL	0.91 ± 0.79	1.13 ± 1.18	0.37
Average glucose	165.26 ± 24.46	164.65 ± 24.45	0.79
CV	36.52 ± 5.01	37.81 ± 5.89	0.08
Total daily insulin requirement (U/kg/day)	0.58 ± 0.13	0.58 ± 0.11	0.99
Basal insulin requirement (U/kg/day)	0.30 ± 0.11	0.30 ± 0.10	0.99
Bolus insulin requirement (U/kg/day)	0.29 ± 0.13	0.28 ± 0.12	0.84

Continuous variables are expressed as mean ± standard deviation. The data on insulin intake were available for 9 patients.

) and in the control group (Supplementary Table S2).

3.3. Comparison Between the Study Group and the Control Group

During the follicular phase, no differences in glycemic control were observed between the study and control groups. However, patients on estroprogestin therapy showed higher total daily insulin (TDI) and bolus insulin requirements (p = 0.017 and p = 0.007, respectively). These findings were also observed in the luteal phase, although not reaching statistical significance (TDI, p = 0.051; bolus insulin, p = 0.049), likely due to increased insulin intake in the study group during this phase (Table 2), while insulin intake remained stable in the control group (Table 3). Moreover, in the luteal phase, the study group demonstrated more severe hypoglycemia compared to the control group (p = 0.044).

When the analysis was restricted to patients with AHCL, the study group showed more bolus insulin requirements compared to the control group in the follicular phase (p = 0.036), but no differences were observed in the other variables analyzed. Similarly, no differences were demonstrated in the two groups during the luteal phase. As for patients not using AHCL devices, the study group demonstrated higher insulin intake both for total and bolus doses (p = 0.033 and p = 0.052, respectively) compared to the control group. Similarly, the same results were confirmed in the luteal phase (TDI p = 0.021; bolus insulin requirement p = 0.052) (Supplementary Table S2).

4. Discussion

To our knowledge, this is the largest study designed to investigate, as the primary outcome, the impact of the menstrual cycle on glycemic control in women of childbearing age with T1D who use CGM devices. The evidence indicates that menstrual cycle phases significantly affect insulin sensitivity, with the luteal phase often associated with increased insulin resistance compared to the follicular phase. In 1992, Widom et al. reported that a subgroup of women with T1D exhibited a decline in glucose control during the luteal phase, and this was correlated with elevated serum estradiol levels, suggesting that estradiol may play a key role in modulating insulin sensitivity in this population [1]. Interestingly, Sacerdote and Bleicher demonstrated that the exacerbation of hyperglycemia during the luteal phase could be attenuated by the administration of oral contraceptives, further highlighting the hormonal substrate behind these metabolic fluctuations [13]. More recently, Brown et al. suggested that the cyclic hormonal variations in both estradiol and progesterone are related to glycemic variability across the menstrual cycle in women with T1D [6]. With the introduction of glucose monitoring systems, these findings were further reinforced, since glucose trends can be analyzed more precisely using CGM reports [4,5,6].

In our cohort, the presence of a worse glycemic control during the luteal phase was demonstrated in the study group (women not taking estroprogestin therapy). Interestingly, when the analysis was restricted to patients using AHCL, this finding was confirmed, showing not only a lower TIR but also a higher TAR, TAR > 250 mg/dL, average glucose, and bolus insulin requirement. Notably, all the patients included in this evaluation used the most recently released AHCL devices with the latest available updates during the study period. It must be mentioned that only two patients in this cohort used a specific profile for AHCL settings, designed to improve the management of glycemic control throughout the different phases of the menstrual cycle, and it was used a couple of days before and throughout the whole vaginal bleeding period, as reported by the patients. Thus, as indicated by our findings, the AHCL algorithms seemed not flexible enough to adapt to the fast changes in insulin sensitivity secondary to the hormonal fluctuations that occur throughout menstrual cycles. This is in line, for instance, with a recent publication by Monroy and colleagues [14], which reported higher average glucose and total insulin requirements during the late luteal phase compared to the early follicular phase in 12 women using Minimed 780G^® throughout three consecutive menstrual cycles. In clinical practice, if possible for the type of algorithm, the activation of a dedicated profile in the luteal phase with more aggressive settings than the usual profile should be encouraged to tackle the increased insulin resistance observed in this phase.

On the other hand, we did not find significant differences in glycemic control or insulin requirement when comparing the follicular and luteal phases in patients from the study group not using AHCL. Overall, as expected, the women not using AHCL still showed worse glycemic control than those using AHCL.

Significant changes in glycemic control or insulin requirement during the menstrual cycle among patients taking estroprogestin therapy (control group) were not observed.

Clearly in contrast to our findings, a study conducted by Levy et al. on data derived from 96 menstrual cycles from 16 women using Control-IQ^® did not find any significant difference in glycemic control or insulin requirement between the menstrual cycle phases. Differently from our study, this work introduced a third phase of the menstrual cycle, referred to as the menstrual phase and defined as the days surrounding vaginal bleeding. Moreover, the luteal phase corresponded to the nine days preceding the onset of bleeding. These differences in the definition of the menstrual cycle phases may explain the different findings reported by our study. Additionally, Levy and colleagues investigated a lower number of patients for a higher number of menstrual cycles compared to our study. Therefore, this study mainly investigated the intraindividual variability rather than the interindividual variability [7]. The different results obtained might suggest that there is a specific subgroup of women who undergo a severe worsening in insulin sensitivity in the luteal phase that might not be effectively tackled by the AHCL system. In a recent publication, 15 females aged 14–21 years underwent three consecutive months of using an open-loop system and then switched to three consecutive months of AHCL therapy with MiniMed 780G^® using a glycemic target of 100 mg/dL and an active insulin duration of 2 h. The luteal and follicular phases were defined according to hormonal fluctuations evaluated through blood tests; the periovulatory phase was not considered for this analysis. As expected, during the open-loop period, glycemic control was poorer in the luteal phase compared to the follicular phase, even though insulin delivery was increased. During the AHCL period, no differences in glycemic control were observed between the two phases. However, an increase in insulin total daily dose, in auto-basal and automatic correction, and in insulin delivery was reported in the luteal phase compared to the follicular phase. The aggressive settings used for the AHCL period may have contributed to the discrepancy observed in this study compared to ours [8]. This hypothesis is supported by another study [9], which included 13 females and used a similar design, where patients were switched from a pre-low glucose suspend (PLGS) therapy with MiniMed 640G^® to an AHCL device, namely MiniMed 780G^®. However, unlike the previously mentioned study, a glucose target of 110 mg/dL and an active insulin time of 3 h were set upon AHCL initiation. During the AHCL period, a lower TIR, despite an increase in total, basal, and bolus insulin requirements, was observed in the luteal phase compared to the follicular phase, suggesting the importance of using aggressive AHCL settings to effectively counteract the increased insulin resistance associated with the luteal phase. A prospective study on the effects of physical exercise in women with T1D [10] showed that glycemic control was poorer during the luteal phase compared to the follicular phase. However, this finding was not confirmed when the analysis focused only on patients undergoing AHCL therapy. The women involved in this study performed more than 30 min per day of physical exercise, and this might have had an impact on reducing the insulin sensitivity fluctuations across the different phases of the menstrual cycle.

In our study, no differences in glycemic control were demonstrated when women taking estroprogestin therapy were compared to women not receiving this treatment, despite the first group demonstrating a higher insulin intake. This result is in line with recent findings suggesting that only high-dose combined oral contraceptives have a potential mild negative effect on glucose regulation, whereas more recent low-dose contraceptive pills have no major impact on glucose metabolism [15].

In this study, there are limitations that warrant consideration. First, the retrospective and observational design limited the accuracy and completeness of the information collected. Then, the classification of menstrual cycle phases relied on self-reported data, which may limit the accuracy of the classification of the two phases. Additionally, heterogeneity in AHCL system types and settings limited the ability to detect significant differences between groups and prevented definitive conclusions on their role in glycemic regulation across menstrual cycle phases. Moreover, the small number of patients observed might limit the generalizability of our findings. Finally, the analysis was based on a single menstrual cycle per participant, thus limiting the capturing of intraindividual variability in glycemic control over multiple cycles.

5. Conclusions

This study confirms that glycemic control in women with T1D deteriorates during the luteal phase, suggesting that a more aggressive glycemic management strategy may be required during this period to counteract the increased risk of hyperglycemia. In the context of AHCL systems, this study contributes to the multifaceted body of literature, with inconsistent findings regarding their effectiveness in reducing luteal phase glycemic worsening. This might be explained by a lack of consistency in defining the menstrual cycle phases, by the substantial heterogeneity in the study populations (different ages or physical exercise habits), and by the variability in AHCL settings. Larger prospective and homogeneous studies are needed to evaluate the ability of AHCL to adapt in response to the increased insulin resistance observed in the luteal phase.