2023 Volume 70 Issue 8 Pages 777-786
We investigated the pathophysiology of the dawn phenomenon by examining the effects of changes in blood glucose levels from late night to early morning on various hormones in a group taking glargine BS and a group taking Lantus XR, with the goal of achieving better glycemic control. Patients with types 1 and 2 diabetes scheduled for inpatient education were divided into BS and XR groups. Blood glucose levels were tracked from 0:00 to 7:00, while blood samples were extracted at 3:00 and 7:00 to measure glucose levels and hormones related to the dawn phenomenon. Overall, we analyzed blood sample and intermittently scanned Continuous Glucose Monitoring data of 43 and 40 patients, respectively. From 0:00 to 7:00, the mean blood glucose was significantly lower in the BS group, although the fluctuation was similar (p < 0.0001). The BS group also exhibited significantly higher ∆ACTH (p = 0.0215) and ∆ cortisol (p = 0.0430) than the XR group. In the BS group, ∆Glu exhibited a significant negative correlation with ∆ACTH and ∆cortisol (p = 0.0491). Similar findings were not observed in the XR group. These results suggest that XR may be a better choice for long-acting insulin since it is less likely to induce cortisol secretion. Further, analysis of the dawn phenomenon and non-dawn phenomenon groups showed the mean CPR levels at 3:00 and 7:00 were significantly higher in the latter (p = 0.0135). This supports the conventional belief that appropriate basal insulin replacement therapy is a beneficial treatment for the dawn phenomenon.
CELLS IN THE HUMAN BODY function in coordination with each other. Various hormones are involved in intercellular communication. Homeostasis is maintained by the growth of organisms, and metabolic adjustment. Multiple hormones are involved in the maintenance of homeostasis. Hormones involved in glucose metabolism include insulin, cortisol, growth hormone (GH), catecholamines, and glucagon. Cortisol reduces glucose utilization by the peripheral tissue by antagonizing the insulin effect and promotes glucose output to the circulation by increasing the activity of glucose-6-phosphatase in the liver, leading to hyperglycemia [1, 2]. GH increases blood glucose levels by increasing hepatic glucose output and reducing cellular glucose consumption to maintain heart and skeletal muscle glycogen levels [3]. Catecholamines increase intracellular cAMP leading to increased phosphorylase activity, which subsequently leads to glycogenolysis resulting in increased blood glucose. Catecholamines inhibit the synthesis of muscle glycogen from glucose by promoting muscle glycogenolysis thereby increasing lactic acid production and hepatic glycogen resynthesis. Glucagon increases phosphorylase activity by promoting cAMP production and protein kinase activity by activating adenylate cyclases, leading to increased hepatic glycogenolysis and blood glucose levels. Furthermore, glucagon promotes gluconeogenesis from amino acids and lipids. In contrast, insulin has a hypoglycemic effect. Insulin promotes glucose uptake in the peripheral tissue by inducing translocation of glucose transporter 4 to the cell surface, thereby reducing blood glucose levels. Glycogen phosphorylase, a rate-limiting enzyme in glycogenolysis, is activated through phosphorylation by the cAMP-dependent protein kinase. Conversely, insulin reduces glycogenolysis by reducing intracellular cAMP levels [4, 5]. Thus, insulin and various insulin counter regulatory hormones are involved in regulating blood glucose [6]. In patients with diabetes, in addition to hyperglycemia due to insufficient insulin action, counter regulatory hormones can increase glucose levels to disrupt glucose homeostasis [7]. The condition in which glucose levels rise naturally from about 3:00 to morning due to abnormal secretion of counter regulatory hormones and irregular insulin secretion is called the dawn phenomenon [8]. Growth hormone is reported to be involved in diurnal fluctuations [9, 10]. In healthy states, insulin secretion is sufficient to suppress the rise in glucose caused by counter regulatory hormones. However, in type 1 diabetes, where endogenous insulin secretion has been depleted, the dawn phenomenon can be significant [6]. The dawn phenomenon is reported to not only contribute to early morning fasting blood glucose (FBG) levels, but also to the rise in glucose levels after breakfast, and thus affects glycemic control overall [11, 12]. Therefore, appropriate supplementation with long-acting insulin is considered an effective means of addressing the dawn phenomenon and glycemic control throughout the day.
There are several types of long-acting insulin, including insulin glargine U-100 biosimilar (insulin glargine BS “Lilly”: BS) and glargine U-300 (Lantus XR: XR). The latter achieves a smaller surface area per unit dose by trebling the concentration of glargine, resulting in slower, flatter, and more sustained pharmacokinetics [13]. A multicenter, randomized, open-label, parallel-group trial of Japanese patients with type 1 and type 2 diabetes comparing glargine U-100 to XR observed significantly less nocturnal hypoglycemia with XR in both groups [14, 15]. These results indicate that the stable pharmacodynamic profile of XR enables it to reduce hypoglycemia, especially nocturnal hypoglycemia, and stabilize early morning FBG.
However, because in the above-mentioned previous studies, a self-administered blood glucose measuring device was used for intermittent blood glucose measurements, continuous changes in blood glucose levels and glycemic excursion were not evaluated. In addition, since counter hormones may contribute to changes in nocturnal blood glucose levels, not only pharmacodynamic profiles but also the effects of the drugs on counter hormones (e.g., XR-related factors contributing to reduced incidence of nocturnal hypoglycemia) should have been evaluated.
In the present study, we compared the changes in blood glucose levels from late night to early morning and examined the effects on various hormones between patients taking BS and XR. In addition, by measuring endocrine-related substances including blood glucose and insulin counter regulatory hormones from late night to early morning and observing the relationships between hormones and glucose changes, we hoped to clarify the pathophysiology of the dawn phenomenon and achieve better quality glycemic control.
This study was an open-label, randomized controlled trial. Patients scheduled for two weeks of inpatient education in the Department of Metabolism and Endocrinology at Chigasaki Municipal Hospital were divided into a BS group and an XR group.
This study performed treatment adjustment for diabetes during educational hospitalization and evaluated the presence or absence of complications of microangiopathy and macroangiopathy. In addition, group guidance was provided by physicians and co-medical staff to improve the patient understanding of the disease. Reduction of glucotoxicity by intensive insulin therapy was attempted especially during the first half of hospitalization. After confirming the reduction of glucotoxicity, the assigned long-acting insulin was administered by the attending physician while wearing a continuous glucose meter (Abbott FreeStyle Libre pro; America) to analyze changes in blood glucose from 0:00 to 7:00, which is morning FBG, and blood samples taken at 3:00 and 7:00.
The dose of the long-acting insulin analog was adjusted so that early morning fasting blood glucose was approximately at 80–126 mg/dL. However, the adjustment was at the discretion of the attending physician based on the patient’s insulin resistance, age, and progression of complications without specific criteria for titration methods. The doses of the long-acting insulin at the time of data collection were 12.1 ± 9.62 unit in the BS group, 18.3 ± 13.5 unit in the XR group.
The primary endpoints were intermittently scanned continuous glucose monitoring (isCGM) and blood glucose (Glu) levels at 3:00 and 7:00 (mean ± SD). The secondary endpoints were GH, cortisol, ACTH, catecholamines, and glucagon levels at 3:00 and 7:00. We registered 47 patients. Due to missing data, blood sample data were analyzed on 43 patients (29 men, 14 women, age 64.8 ± 16.6 years) and isCGM data on 40 patients (27 men, 13 women, age 66.7 ± 15.1 years). The study protocol was approved by the Medical Ethics Committee of the Chigasaki Municipal Hospital and carried out according to the Declaration of Helsinki. The patients were anonymized to protect their personal information.
Inclusion criteriaPatients aged 20 years or older with type 1 and type 2 diabetes who consented to the purpose and content of the study and voluntarily gave written consent, were included. There was no upper limit on age as long as they had a cognitive function that allowed them to self-determine their participation in the study. One researcher (MH of Chigasaki Municipal Hospital) enrolled the participants and another researcher (medical staff from Yokohama City University Hospital) assigned participants to each group.
Exclusion criteriaPatients with adrenal or pituitary diseases, on long-term steroid therapy, with cancer, or with gestational diabetes were excluded.
Sample size calculationThe effect size was calculated as the change in the primary endpoint blood glucose level from 3:00 to 7:00. A significant effect size of 14 mg/dL (generally, a 0.5% decrease in HbA1c is considered a significant change, which corresponds to the mean glucose level of 14 mg/dL), SD of 15, significance level of 0.05, and power of 0.8; the sample size was N = 39. Considering a larger SD than expected, dropouts, and other factors, we set a target of about 25 patients for each group. Block randomization was also performed after adjusting for age (<70 years/≥70 years) and HbA1c at admission (<9%/≥9%) to prevent any discrepancies between the patients in both groups. Patients were randomly classified by a researcher from Yokohama City University Hospital who was blinded to the assignment. Computer-generated randomization systems were used in this trial. Patients were assigned to the groups in a 1:1 ratio.
Statistical analysisPatient characteristics and biochemical data were expressed as mean ± SD and examined with a paired t-test. The results were expressed as mean difference ± standard error. A p-value of <0.05 was considered statistically significant. In addition, we examined the correlations between the amount of change in blood glucose and the various hormones from 3:00 to 7:00. A correlation coefficient of ≥0.2 was considered a positive correlation and that of ≤–0.2, a negative correlation. JMP Pro 15 (SAS Institute Inc., Cary, NC, USA) was used for the statistical analysis.
Table 1 shows the blood test data and patient characteristics of 43 of the 47 patients who provided consent. Patients whose blood samples were not collected or those whose isCGM data were not accurately recorded were excluded. We analyzed the first blood sampling data after admission. There are no significant differences between the values for the BS and XR groups (Table 1). The isCGM assessments from 0:00 to 7:00 showed that the mean glucose was significantly lower in the BS group than in the XR group, although the fluctuations were similar (BS: 91.8 ± 8.46 mg/dL vs. XR: 100.7 ± 7.16 mg/dL, p < 0.0001) (Fig. 1). Further, the assessments of glucose and hormones levels from blood samples taken at 3:00 and 7:00 showed that glucose in the BS and XR groups at 3:00 was 100.8 ± 21.6 mg/dL and 109.3 ± 29.5 mg/dL, respectively. ∆Glu, which indicates the amount of change, was higher in the BS group but the difference was not significant (BS: 9.5 ± 19.1 mg/dL vs. XR: 3.74 ± 25.4 mg/dL, p = 0.4117) (Fig. 2). One patient in each group experienced hypoglycemia (Glu <70 mg/dL) at 3:00 or 7:00; both occurrences were hypoglycemia unawareness. ∆ACTH (BS: 16.7 ± 8.42 pg/mL vs. XR: 9.34 ± 11.4 pg/mL, p = 0.0215) and ∆cortisol (Cor) (BS: 7.13 ± 2.71 μg/dL vs. XR: 4.81 ± 4.29 μg/dL, p = 0.0430) were significantly higher in the BS group compared to the XR group (Fig. 2). Further, in the BS group, ∆Glu exhibited significant negative correlations with ∆ACTH (correlation coefficient –0.50, p = 0.0247), ∆Cor (correlation coefficient –0.45, p = 0.0491), and ∆GH (correlation coefficient –0.50, p = 0.0234), and a significant positive correlation with ∆noradrenaline (NA) (correlation coefficient 0.54, p = 0.015). These correlations were not observed in the XR group (Fig. 3).
| BS | XR | |
|---|---|---|
| N (male:female) | 20 (12:8) | 23 (17:6) |
| Age (y.o.) | 62.4 ± 17.1 | 67.0 ± 16.3 |
| Body mass index (kg/m2) | 24.4 ± 4.05 | 24.3 ± 4.98 |
| Systolic blood pressure (mmHg) | 127.1 ± 17.9 | 123.7 ± 15.9 |
| Diastolic blood pressure (mmHg) | 71.2 ± 9.76 | 70.1 ± 10.8 |
| Known duration of diabetes (year) | 13.4 ± 12.5 | 11.5 ± 11.5 |
| Family history of diabetes (%) | 50.0% | 52.2% |
| Past history of cardiovascular disease (%) | 5.0% | 8.7% |
| Number of patients with type 1 diabetes | 2 | 0 |
| Number of patients with hypoglycemia (<70 mg/dL) | 1 | 1 |
| HbA1c (%) | 10.8 ± 1.98 | 10.5 ± 2.38 |
| Cre (mg/dL) | 0.80 ± 0.46 | 0.78 ± 0.34 |
| UACR (mg/gCr) | 30.5 ± 49.5 | 82.6 ± 150.8 |
| eGFR (mL/min/1.73 m2) | 80.8 ± 26.5 | 80.1 ± 24.5 |
| AST (IU/L) | 27.7 ± 16.5 | 29.7 ± 26.8 |
| ALT (IU/L) | 36.9 ± 25.1 | 36.3 ± 40.0 |
| ChE (IU/L) | 342.8 ± 76.6 | 353.7 ± 104.0 |
| γGTP (IU/L) | 78.9 ± 175.7 | 51.7 ± 50.1 |
| TG (mg/dL) | 256.3 ± 443.6 | 178.3 ± 118.5 |
| HDL-Chol (mg/dL) | 55.1 ± 21.1 | 48.0 ± 9.36 |
| LDL-Chol (mg/dL) | 111.0 ± 37.5 | 157.5 ± 212.5 |
HbA1c: hemoglobin A1c; Cre: creatinine; UACR: urine albumin to creatinine ratio; eGFR: estimated glomerular filtration rate; AST: aspartate transaminase; ALT: alanine transaminase; ChE: cholinesterase; γ-GTP: γ-glutamyl transpeptidase; TG: triglyceride; HDL-Chol: high density lipoprotein cholesterol; LDL-Chol: low density lipoprotein cholesterol
Comparison of the changes in glucose and difference by isCGM from 0:00 to 7:00 between BS and XR groups.
Comparison of ∆Glu, ∆CPR, ∆ACTH, ∆Cor, ∆GH, ∆glucagon, ∆NA between BS group and XR group. ∆ is the change in each hormone from 3:00 to 7:00. Of the catecholamines, adrenaline (A) and dopamine (DA) are both under sensitivity.
(a) Correlations with ∆Glu and each ∆hormones in BS group, (b) Correlations with ∆Glu and each ∆hormones in XR group.
When examining all the patients by investigating the blood data of both groups collected at 3:00 and 7:00, among the insulin counter regulatory hormones, only ACTH and cortisol increased significantly from 3:00 to 7:00 (∆Cor: 5.89 ± 3.79 μg/dL, ∆ACTH: 12.8 ± 10.7 pg/mL) (Table 2). However, these did not correlate with the change in blood glucose during this period, and among the insulin counter regulatory hormones, only noradrenaline exhibited a significantly positive correlation with the change in blood glucose (correlation coefficient 0.40, p = 0.0087). A significantly positive correlation was also observed with blood CPR (correlation coefficient 0.33, p = 0.0302) (Fig. 4).
| Total | |||||
|---|---|---|---|---|---|
| 3:00 | 7:00 | Average | ∆ | p value | |
| Glu (mg/dL) | 105.3 ± 26.2 | 111.7 ± 25.4 | 108.5 ± 23.2 | 6.42 ± 22.6 | 0.07 |
| CPR (ng/mL) | 0.86 ± 0.58 | 0.96 ± 0.57 | 0.91 ± 0.54 | 0.10 ± 0.39 | 0.1043 |
| Cor (μg/dL) | 7.05 ± 3.20 | 12.9 ± 3.44 | 10.0 ± 2.73 | 5.89 ± 3.79 | <0.0001 |
| ACTH (pg/mL) | 19.5 ± 12.8 | 32.3 ± 13.1 | 25.9 ± 11.8 | 12.8 ± 10.7 | <0.0001 |
| GH (ng/mL) | 1.30 ± 1.40 | 1.13 ± 1.15 | 1.21 ± 0.98 | –0.17 ± 1.65 | 0.5133 |
| Glucagon (mg/dL) | 17.5 ± 12.6 | 18.8 ± 11.6 | 18.1 ± 11.1 | 1.26 ± 9.69 | 0.3982 |
| A (pg/mL) | Under sensitivity | Under sensitivity | |||
| NA (pg/mL) | 0.21 ± 0.14 | 0.25 ± 0.18 | 0.23 ± 0.15 | 0.04 ± 0.12 | 0.0507 |
| DA (pg/mL) | Under sensitivity | Under sensitivity |
Glu: glucose; CPR: C-peptide; Cor: cortisol; ACTH: adrenocorticotropic hormone; GH: growth hormone; A: adrenaline; NA: noradrenaline; DA: dopamine
Correlations with ∆Glu and each ∆hormones in all the subjects.
When the patients were divided into a dawn phenomenon (increase in blood glucose from 3:00 to 7:00: ∆Glu >zero) and a non-dawn phenomenon groups (decrease in blood glucose over the same period: ∆Glu ≤zero), ∆CPR was significantly higher in the former (dawn phenomenon: 0.21 ± 0.41 ng/mL vs. non-dawn phenomenon: –0.09 ± 0.26 ng/mL, p = 0.0135), though mean CPR was significantly higher in the latter (dawn phenomenon: 0.78 ± 0.44 ng/mL vs. non-dawn phenomenon: 1.14 ± 0.63 ng/mL, p = 0.0135). None of the other insulin counter regulatory hormones exhibited significant differences between the groups in the amount of change or mean values (Fig. 5).
(a) Comparison of ∆Glu, ∆CPR, ∆ACTH, ∆Cor, ∆GH, ∆glucagon, ∆NA between non-dawn phenomenon (NDP) group and dawn phenomenon (DP) group. (b) Comparison of mean Glu, mean CPR, mean ACTH, mean Cor, mean GH, mean glucagon, mean NA between NDP group and DP group. Each mean concentration was calculated from the blood sampling results at 3:00 and 7:00.
The dawn phenomenon is observed not only in patients with type 1 diabetes but also among patients with type 2 diabetes and people with normal glucose tolerance [16-18]. However, in people with normal glucose tolerance, nocturnal blood glucose and serum insulin levels remain relatively flat, with insulin secretion increasing temporarily around dawn [16]. This rise in insulin secretion suppresses gluconeogenesis in the liver to prevent blood glucose levels from increasing [16, 17]. Insulin sensitivity is higher from 0:00 to 3:00 than from 4:00 to 8:00. A comparison of neutral protamine hagedorn (NPH), continuous subcutaneous insulin infusion (CSII), and long-acting insulin showed no inappropriate decline in blood glucose near 3:00 or rise around dawn in the latter two compared to the former [19]. In a study of patients with type 1 diabetes, not only was the dawn phenomenon more common with NPH than with CSII and long-acting insulin but also the levels of insulin-like growth factor-binding protein-1 (IGFBP-1), which is an indirect indicator of GH secretion, were higher [20].
The isCGM results of the present study showed that nocturnal glucose levels were significantly lower in the BS group than in the XR group, although the fluctuation patterns were similar to each other. One reason for this could be the degree of insulin unit adjustment in each group. However, in the BS group, ∆Glu from 3:00 to 7:00 exhibited negative correlations with ∆ACTH, ∆Cor, and ∆GH, which suggest that the physiological defense response to nocturnal hypoglycemia increased the secretion of hormones that increase blood glucose, which may have reduced glucose fluctuations in the BS group. Cortisol secretion is reported to affect the circadian rhythm of glucose levels [21], and thus there is concern that excess secretion of cortisol at night could lead to an increase in glucose levels after breakfast. In the present study, a significant increase in ∆Cor was observed in the BS group, but not in the XR group. Appropriate post-breakfast glucose management is thought to contribute substantially to improving diurnal glucose fluctuations, and thus HbA1c levels [11, 22]. Based on the results of this study, XR is considered a better choice for long-acting insulin because it stabilizes nocturnal glucose levels and improves post-breakfast glucose levels.
There have been various reports on the dawn phenomenon, with many finding that nocturnal GH secretion is most strongly associated with the dawn phenomenon in patients with type 1 diabetes. Boyle et al. found that GH increases insulin clearance by examining changes in blood glucose and insulin levels in patients with insulin-dependent type 1 diabetes with growth hormone deficiency [23]. In addition, Campbell et al. compared a group that received growth hormone after taking somatostatin (which suppresses multiple hormones that raise glucose levels) to a group that received glucagon also after taking somatostatin and found that glucose levels increased only in the former [24]. Moreover, Kovisto et al. and Atiea et al. reported significant correlations between increased plasma glucose at night and increased cortisol levels. The former confirmed that the use of metyrapone reduced early morning FBG [25, 26]. A study by Atiea et al. on patients with non-insulin-dependent diabetes found that nocturnal blood cortisol levels increased significantly from 4:00 to 9:00 but significant changes were not observed in insulin, CPR, glucagon, GH, or catecholamines [27]. Kerner et al. also found that insulin clearance was associated with the dawn phenomenon but did not determine which hormones affect clearance [28].
While the present and previous studies also observed increases in blood ACTH and cortisol levels from late night to dawn, the present study found no correlation with blood glucose levels, which indicates this single factor is not necessarily directly linked to the dawn phenomenon. Neither did the present study find a significant correlation between GH and the dawn phenomenon, which has been described in various other studies [9, 23]. Although there have been relatively few reports on the correlation between the dawn phenomenon and noradrenaline, Bolli et al. mentioned an association between the dawn phenomenon and catecholamines [17]. In the present study, this had the strongest correlation with glucose levels and the insulin counter regulatory hormones, suggesting its involvement in the dawn phenomenon.
Oral hypoglycemic agents are insufficient to treat dawn phenomenon and that appropriate supplementation of basal insulin is effective [11, 19, 20, 29]. In the present study, blood CPR levels were significantly higher in the non-dawn phenomenon group than that in the dawn phenomenon group, corroborating previous findings that appropriate basal insulin replacement is a beneficial treatment for the dawn phenomenon.
Because dose determination of long-acting insulin analogs and reduction of glucotoxicity were at the clinical discretion of the attending physician, and owing to the lack of criteria, the results of changes in blood glucose levels at midnight and hormone levels might have been biased. In addition, stress induced by blood sampling at 3:00 and the special environment of hospitalization may have affected the secretion of hormones such as cortisol and catecholamines. Furthermore, it has been reported that GH levels peak before 3:00 [7]. Therefore, blood sampling at midnight should have also been considered, taking into account the degree of invasion. It is also reported that poor sleep quality is directly associated with the dawn phenomenon in patients with diabetes [30]. Thus, changes in nocturnal blood glucose levels observed in this study may not represent the actual changes.
The authors thank the medical staff of Chigasaki Municipal Hospital for performing blood sampling tests and centrifuging the samples at 3:00 and 7:00. We would like to thank Editage (www.editage.com) for English language editing.
Yasuo Terauchi received honoraria from Eli Lilly Japan and Sanofi. Masanori Hasebe, Shinobu Satoh, Kohei Ito and Haruka Tamura declare no conflict of interest.
This trial was registered on September 13, 2021 (Reg No: UMIN000045464).
