Metformin add-on continuous subcutaneous insulin infusion on precise insulin doses in patients with type 2 diabetes

To investigate whether metformin add-on to the continuous subcutaneous insulin infusion (Met + CSII) therapy leads to a significant reduction in insulin doses required by type 2 diabetes (T2D) patients to maintain glycemic control, and an improvement in glycemic variation (GV) compared to CSII only therapy. We analyzed data from our two randomized, controlled open-label trials. Newly diagnoses T2D patients were randomized assigned to receive either CSII therapy or Met + CSII therapy for 4 weeks. Subjects were subjected to a 4-day continuous glucose monitoring (CGM) at the endpoint. Insulin doses and GV profiles were analyzed. The primary endpoint was differences in insulin doses and GV between the two groups. A total of 188 subjects were admitted as inpatients. Subjects in metformin add-on therapy required significantly lower total, basal and bolus insulin doses than those of control group. CGM data showed that patients in Met + CSII group exhibited significant reduction in the 24-hr mean amplitude of glycemic excursions (MAGE), the standard deviation, and the coefficient of variation compared to those of control group. Our data suggest that metformin add-on to CSII therapy leads to a significant reduction in insulin doses required by T2D patients to control glycemic variations.

Glycemic profiles. Our CGM data showed that the hourly glucose concentrations in patients treated with metformin combined with CSII therapy were significantly lower than those of control group, especially from 0400 to 2300 o' clock (P < 0.05, respectively) ( Fig. 1). Furthermore, we also analyzed the 0000-0600, 0000-0300 and 0300-0600 hours glucose concentrations between the two groups at the conclusion of the study. Our data showed that patients treated with the metformin add-on CSII therapy had significantly lower glucose concentrations during 0300-0600 hours (5.84 ± 0.96 vs. 6.40 ± 1.45 mmol/L, P < 0.05), although patients had similar glucose concentrations during 0000-0600 and 0000-0300 hours (5.97 ± 1.07 vs. 6.40 ± 1.54 mmol/L and 6.10 ± 1.37 vs. 6.40 ± 1.78 mmol/L, P > 0.05, respectively) between the two groups.
We analyzed dawn phenomenon in this study, the incidence of dawn phenomenon was 3.1% (6 out of 195 subjects). The average magnitude of the dawn phenomenon was 2.60 ± 2.20 mmol/L, and the percentage of their nights do people display this phenomenon is 21.1%.

Items
Metformin + CSII CSII P Safety and tolerance. A total of 16 patients in metformin add-on CSII therapy group had mild digestive disorders. We also compared the risk of severe hypoglycemia (glucose level <3.9 mmol/L) between the two groups. Although Table 2 showed the incremental of area under the curve of glucose concentration less than 3.9 mmol/L was similar between the two groups. The hypoglycemic events in CSII group were statistically significant higher compared to those in Metformin add-on CSII group (17 vs. 7 patients, P = 0.03).

Discussion
We analyzed data from our two randomized, controlled open-label trials on newly diagnoses T2D patients randomized assigned to receive either CSII therapy or metformin add-on to CSII therapy. Our data demonstrated that a metformin add-on to the CSII therapy can significantly reduce the total daily insulin doses to maintain euglycemic control and facilitate further improvements in glycemic variations. We also found that patients treated with the combined metformin and CSII therapy needed a shorter treatment period to achieve euglycemic control than those in the control (CSII) group at the end of the study, and that combination therapy resulted in no increase of hypoglycemia. Metformin, a family number of biguanide, has become established as the first-line treatment in management of T2D. In combination with insulin in treatment of T2D, the use of metformin leads to the increase in insulin sensitivity and the reduction of daily insulin doses [22][23][24] . In this study, we observed that patients in the metformin add-on CSII therapy achieved euglycemic control in significantly shorter times, and with lower daily insulin doses than those in control group. Our results were in agreement with a previous study reporting that metformin combined with CSII therapy needed less insulin titration time to achieve glycemic control than the control group,    and 90% patients achieved euglycemic control by month 3 26 . Our data indicated that patients treated with a metformin add-on CSII therapy showed significant improvement in β-cell function as measured by HOMA-B and Insulinogenic Index. Although these are indirect measurements of beta cell function, they suggest a reason for the lower FPG and fasting insulin concentration found in the treated patients. In addition, patients treated with metformin combined with NPH therapy for 12-week had significantly decrease in MBG, MAGE, SDBG, and LAGE from baseline 27 . CGM provides glucose concentrations every 5 min, which may be an option for determining 24-hr glycemic profiles. The increasing use of CGM leading to the better understanding of glycemic variability 31 , and promoting the emerging of treatment aimed at avoiding glucose fluctuations in diabetic patients 32 . A number of metrics obtained from CGM could be used to describe the GV in diabetic patients [14][15][16] . Using CGM data, clinical researchers and clinicians could efficiently evaluate the quality of the glycemic control which might be important for decision-making 20 . In this study, our CGM data revealed that patients in the metformin add-on CSII therapy group had further improvement in MAGE, SD, and CV compared with those in CSII alone therapy group. Studies have demonstrated that there was a high degree of correlation between SD and MAGE 13,18 . The percent CV displays well the interpretation of GV 13,18,19 . In addition, we observed that patients in the metformin add-on CSII group had lower glycemic variations during 0300-0600 o' clock compared with those in CSII alone group. In this study, the incidence of dawn phenomenon was 3.1% in all subjects. Our observation was in agreement of a previous study reporting that the dawn phenomenon affecting nearly 3% of T2D population 33 . However, our data could not address the underlying mechanisms of metformin decreasing the dawn phenomenon.
A trial conducted in subjects with type 1 diabetes indicated that a metformin add-on CSII therapy resulted in an improvement in blood glucose control and insulin requirements 21 . Of most importance, metformin combination with CSII is a safe therapy for the treatment of type 1 diabetes 21 . We previously observed that older (>60 years) men with newly diagnosed T2D had a high incidence of nocturnal low glucose before and during intensive insulin therapy, which indicates that special attention should be paid to prevention of hypoglycemia in elderly male patients when treated with intensive insulin therapy using an insulin pump 30 . In this study, we used insulin pump 712E (Medtronic Incorporated, Northridge, USA) delivered insulin in both groups, our data showed that patients treated with the metformin add-on CSII therapy had significantly decreased incidences of hypoglycemia (blood glucose <3.9 mmol/L). Although, we did not observe the long-run effect of metformin add-on CSII therapy on T2D glycemic control, the significant improvement in MAGE, SD, and %CV compared with control group indicating that metformin add-on CSII therapy may be the better option aimed at reducing glycemic fluctuations in T2D.
In the current study, we observed that drug naïve diabetic patients treated with metformin combination with CSII therapy experienced a significant reduction in the total daily insulin doses, bolus insulin doses, and basal insulin doses to maintain euglycemic control, the shorter times to achieve euglycemic control, the improvement in glycemic variation and associated with no increase of hypoglycemia, which indicates that metformin combined with CSII therapy can be considered a safe and valuable treatment option. However, the study patient population was limited to the Nanjing area in China, therefore, the situation might not be the same for other geographical regions or populations. The Chinese population had thrifty gene 34 , the high portion intake of carbohydrates and highly different life-style 35 . Moreover, Asian T2D populations have the lower BMI and smaller waist circumferences compared to the Western participants 36 . The better response to acarbose 37 , DPP-4 inhibitor 38 , GLP-1 receptor agonists 39 , and SGLT-2 inhibitor therapy 40 . These therapies have shown potential improvements in glucose fluctuation 37,38,40 and lower insulin dose requirement by patients with T2D to maintain euglycemic control in Chinese populations 38 .
Our study has several limitations. Firstly, the study only observed patients population in Jiangsu Provence in China, so the situation might not be applicable to patients of other populations. Secondly, we did not observe the long-term effects of metformin add-on CSII therapy on glycemic control in patients with newly diagnosed T2D. Thirdly, we did not demonstrate the underline mechanisms of older men patients who had a higher risk of nocturnal hypoglycemia when treated using insulin intensive therapy. Lastly, a major limitation of study that performed in the inpatient setting with meals provided, therefore limited applicability to the real world.
In conclusion, our data showed that patients with newly diagnosed T2D may take advantage of the metformin add-on CSII therapy in requiring less insulin doses, and having available other methods of glycemic treatments in a hospital setting.

Materials and Methods
The two studies were both randomized, controlled open-label trials. Between April 2012 and June 2016, we recruited a total of 188 patients with newly diagnosed T2D from Nanjing First Hospital in China. The inclusion criteria were (1) patients aged between >18 and <80; (2), HbA1c values >9.0% at diagnosis. Patients were excluded from analysis if they had serum creatinine levels ≥1.5 mg/dL (males), ≥1.4 mg/dL (females) or abnormal creatinine clearance, known hypersensitivity to metformin, acute or chronic metabolic acidosis, including diabetic ketoacidosis, with or without coma.
The study protocols and patient consent forms were approved by the Institutional Ethical Committee of Nanjing First Hospital. All patients gave written informed consent to participate. The methods were carried out in accordance with the Declaration of Helsinki guidelines, including any relevant details.
The trials included a screening period to collect baseline parameter values, a 2± weeks treatment period, and a CGM period. During the study period, all subjects were admitted as inpatients. Computer-generated random orders were prepared by the data-coordinating center and distributed to each patient to determine the patient's treatment assignments. After the baseline parameters were assessed, subjects underwent oral glucose tolerance tests (OGTTs) using 75 g of glucose (dissolved in 200 ml water) before and after treatment. Serum samples were obtained before and 30, and 120 min after oral administration for glucose, insulin, C-peptide, and HbA 1c determination. Plasma insulin was determined using an insulin radioimmunoassay kit (Beijing Technology Company, Beijing, China). HbA 1c was measured by a DiaSTAT HbA 1c analyzer (Bio-Rad, Hercules, CA). C-peptide and glucose concentrations were measured centrally at the central laboratory in Nanjing First Hospital, Nanjing Medical University.
Enrolled subjects were then randomized either metformin add-on CSII or CSII only treatment. Metformin (Bristol-Myers Squibb, USA) was administered 500 mg thrice-daily. If the patients were unable to tolerate the minimum metformin dose (1000 mg/day), they were excluded from this study. The total daily insulin (Aspart, Novo Nordisk, Bagsvaerd, Denmark) doses were 0.4 IU/kg which was given in two injection modes: 1/3 of total daily dose was equally given as boluses within three meals, the remaining insulin was given as basal dose. Investigators titrated insulin doses on an individual-patient basis using the algorithm as before described 39 . When euglycemic control was achieved, treatments would remain unchanged for another 3 days. The treatment period required to achieve euglycemic control (the fasting capillary blood glucose was less than 6.1 mmol/L and capillary blood glucose at 2-hr after each of three meals was less than 8.0 mmol/L), were recorded for each subjects 41,42 . Changes in insulin doses following treatment were also analyzed.
All patients were subjected to 4-day retrospective CGM (Sof-sensor, CGMS-Gold, Medtronic Incorporated, Northridge, USA) in hospital by the specialist nurse after the completion of treatment. The CGM was performed by the study nurses, and the CGM data were saved by the investigator, as described previously 9,37,43 . During the CGM period, all subjects were instructed to maintain their usual physical activity and received meals consisting of a total daily caloric intake of 25 kcal/kg. The ratio of carbohydrate, proteins and fats was 55%, 17% and 28%, respectively.
The 24-hr mean glucose (MG), the SD, the CV, the MAGE, the incremental area under curve (AUC) of hyperglycemia (glucose >10.0 mmol/L) and hypoglycemia (glucose <3.9 mmol/L), the 2-hr postprandial after each meals, and the hourly BG were recorded and calculated. The β-cell function and insulin resistance were assessed by the homoeostasis model assessment B (HOMA-B) and HOMA-IR 42,44 . Matsuda Index 45 and Insulinogenic Index 46 were calculated as previously described.
The primary endpoints were differences in insulin doses and glycemic variations between groups at the completion of treatments. The times taken by the patients to achieve euglycemic control, the total daily insulin doses, the basal insulin doses, the bolus insulin doses, the hourly mean glucose concentrations, the 24-h BG and the AUC of hypoglycemia and hyperglycemia at the endpoint were also analyzed.
Statistical analysis. All data were normal distribution and were presented as the means ± SD. Statistical significance was determined by one-way analysis of variance (ANOVA), or two-way ANOVA for repeated measurements for the group comparisons, followed by Bonferroni-Dunn post hoc test. P < 0.05 was considered to be statistically significant. All of the statistical analyses were performed using the Statistical Product and Services Solutions (SPSS) package (Version 11.5, SPSS, Science, Chicago, USA).