A Comparison of Glycemic Parameters and Their Relationship with C-Peptide and Proinsulin Levels During the Honeymoon and Non-honeymoon Periods in Children with Type 1 Diabetes Mellitus - a Cross-Sectional Study

Background : Currently, there is a lack of data relating to glycemic parameters and their relationship with C-peptide (CP) and proinsulin (PI) during the honeymoon period in type 1 diabetes mellitus (T1D). The aim of this study was to evaluate glycemic parameters in children with T1D who are in the honeymoon period using intermittently scanned continuous glucose monitoring systems (isCGMS) and to investigate any relationships between CP and PI levels. Methods: The study included 21 children who were in the honeymoon period and 31 children who were not. A cross-sectional, non-randomized study was performed. Demographic, clinical data were collected and 2 week- isCGMS data were retrieved. Results: The Serum CP showed a positive correlation with time-in-range in the honeymoon period (p:0.03), however PI showed no correlations with glycemic parameters in both periods. The Serum CP and PI levels and the PI:CP ratio were signicantly higher in the honeymoon group than in the non-honeymoon group. In the non-honeymoon group, the PI level was below 0.1 pmol/L (which is the detectable limit) in only 2 of the 17 cases as compared with none in the honeymoon group. Similarly, only 2 of the 17 children in the non-honeymoon group had CP levels of less than 0.2 nmol / L, although both had detectable PI levels. Overall time-in-range (3.9-10 mmol/L) was signicantly high in the honeymoon group. In contrast, the mean sensor glucose levels, time spent in hyperglycemia, and coecient of variation levels (32.2vs 40.5%) were signicantly low in the honeymoon group. Conclusions: While the CP levels did have an effect on glycemic parameters during the honeymoon period, there was no correlation between PI levels and glycemic parameters. In terms of glycemic parameters during the honeymoon the CP level had a signicant correlation TIR value. CP useful biochemical tool, while PI sensitive biochemical follow-up.


Background
The transient recovery period of beta-cell function, after initiation of insulin therapy in patients with newly diagnosed type 1 diabetes mellitus, is referred to as the 'honeymoon' or remission phase (RP). The clinical signi cance of this period lies in the maintenance of glycemic control characterized by a reduction in insulin requirements. Previous controlled trials on the natural course of beta cell function recovery reported the prevalence of remission as 60%, (1) with a duration ranging from 1 month to as long as 13 years (2). To achieve better glycemic control throughout the remission phase, several studies focused on the factors in uencing the natural course of remission, such as severity of presentation, age at diagnosis, gender, and effects of autoantibodies. Nevertheless, the glycemic parameters and variability in these parameters during the remission period remain unknown.
RP is de ned as an insulin dose-adjusted hemoglobin A1c (HbA1c) (IDAA1c) of equal to or less than 9, where IDAA1c is equal to the sum of HbA1c (% and mmol/mol) and 4 times the insulin dose (units/kg/day) (1). However, this de nition based on insulin doses and HbA1c is not su cient in terms of glycemic variability, insulin sensitivity, and episodes of hypoglycemia and hyperglycemia. Furthermore, there are no de nite cut-off values for C-peptide (CP) and proinsulin (PI) levels for the de nition of the honeymoon period in type 1 diabetes (T1D). However, the measurement of the CP levels for the de nition of the remission period has been established and is a laborious and expensive process. Additionally, an IDAA1C level of 9 has been shown to correspond to a substantially increased predicted stimulated C-peptide (CP) level of 300 pmol / l (3).
Normal insulin biosynthesis is a multi-step process, beginning with a pre-prohormone, pre-pro-insulin, which is then converted to proinsulin (PI). PI becomes incorporated into a new "immature" beta-granule, where it is subsequently cleaved into insulin and CP via prohormone convertases (PCSK1, PCSK2, and carboxypeptidase E) (4). It has been shown that the PI to CP ratio (PI:CP) increases in individuals at risk of diabetes and at the time of diagnosis of diabetes. On the other hand, Watkins et al. examined PI levels at the time of T1D diagnosis, shortly after diagnosis, and during the honeymoon period and found that the PI level was higher during the honeymoon period than that at the time of diagnosis (5). However, the importance of neither CP nor PI levels in clinical practice and their effects on glycemic parameters have been investigated in detail.
In the past, only the HbA1C level was taken into account for de ning glycemic control in diabetes. However, new glycemic parameters have emerged in the light of recent advances in diabetes technology. Currently, the de nition of honeymoon period or RP is still based only on HbA1C levels and insulin doses and there are few studies evaluating other glycemic parameters in this period. In the light of recent technological developments, this study aimed to evaluate the glycemic parameters in children with T1D who are in RP (honeymoon period) and compare them with those who are not in this phase using intermittently scanned continuous glucose monitoring systems (isCGMS). The study also aimed to investigate the relationships between PI and CI levels and glycemic parameters in order to draw comparisons between the honeymoon and non-honeymoon periods.

Subjects
The study was conducted between March 2018 and 2019 and included 21 children with T1D who were in RP (the honeymoon group) and 31 children with T1D who had been diagnosed at least 2 years before the enrollment and who were not in RP (the non-honeymoon group). The inclusion criteria required an age of 5 to 18 years and being under multiple-dose insulin injection therapy and isCGMS. The exclusion criteria were the presence of concomitant diseases that could in uence metabolic control, the use of an insulin infusion pump and having HbA1C level above 9% (75 mmol/mol) (A consort ow diagram is showed in gure 1).

Study design
A cross-sectional, non-randomized study was performed. The study protocol was approved by the Koç University Committee on Human Research (reference number:2018.022.IRB1.004). Written consent was obtained from the parents along with assent from the adolescents as required by the local institutional review board regulations.
A partial remission phase was de ned as an insulin dose adjusted HbA1c (IDAA1c) of equal to or less than 9 (1). Demographic, socioeconomic, and clinical data were collected from medical records and from interviews with the participants and parents. Also, 2 weeks' isCGMS data were retrieved using the Free-Style Libre software in the outpatient pediatric endocrinology clinic. In addition to the software report, the raw data were used for statistical analysis and assessed according to continuous glucose monitoring systems (CGMS) consensus report (6). Random (mostly non-fasting) serum samples were collected for measurement of CP and PI during the routine follow-up visits.

Laboratory analyses
For the measurement of CP and PI, samples were collected into dry tubes and centrifuged promptly (3500 g) for 10 minutes at +4 °C. The sera were separated in aliquots and frozen immediately at −80 °C. Serum CP and PI levels were determined by competitive ELISA using commercial kits (DRG Instruments GmBH, Germany). Intra-and inter-CVs were 6.54% and 9.33% for CP, and 4.3% and 6.8% for PI, respectively. The detection limit was 0.1 pmol/L for PI (7). HbA1c levels were analyzed immediately after collecting blood samples into K 2 -EDTA-added tubes on an ADAMS A1c Lite HA-8380V analyzer (Arkray) using the HPLC (Reversed-phase cation-exchange liquid chromatography) method. Sustained levels of CP were considered to be consistent readings greater than 0.2 nmol/l (8).

Statistical Analyses
The sample size calculation was made using OpenEpi Statistical Software version 3.01 (9). Type 1 error of the study is α = 0,05, the power of the study is 1-β = 0,80, and the number of samples was calculated as 18 per group. A total of at least 40 people had to be included in the study, 20 per group, with an appendix of approximately ten percent loss. The other statistical analyses were conducted using SPSS statistical software version 22 (USA). Descriptive statistics included means, standard deviation (SD), median, interquartile range, and proportions when appropriate. Correlations between CP, PI and glycemic parameters were computed using Spearman's correlation coe cient. For continuous variables, mean ± standard deviation and median values were used for variables with and without normal distribution, respectively. For continuous variables, the Student's t-test was used for normally distributed data and the Mann-Whitney U-test was used for data without normal distribution.

Results
The two groups were similar in terms of age and gender. As expected, the mean diabetes duration, daily insulin dose, HbA1c, IDDA1c, estimated HbA1c levels were signi cantly lower in the honeymoon group compared with the non-honeymoon group.
However, the Serum CP and PI levels and the PI:CP ratio were signi cantly higher in the honeymoon group than those in the non-honeymoon group. The demographic characteristics and laboratory ndings of the two groups are given in Table 1. A total of 17 children in each group agreed to give serum samples for the measurement of PI and CP. In the non-honeymoon group, PI level was below 0.1 pmol/L (which is the detectable limit) in only 2 of the 17 cases as compared with none in the honeymoon group. Similarly, only 2 of the 17 children in the non-honeymoon group had CP levels of less than 0.2 nmol / L, although both had detectable PI levels.
When the glycemic parameters were compared in the two groups (Table 2), overall time-in-range, day-time and night-time levels were signi cantly higher in the honeymoon group. In contrast, the mean sensor glucose levels (overall, day-time and night-time), the mean time spent in level 1 and level 2 hyperglycemia overall, day-time and night-time , the median time spent in level 1 hypoglycemia (<3.9 mmol/L) day-time, and the mean coe cient of variation (CV) and SD levels were all signi cantly lower in the honeymoon group.
However, the median time spent in level 1 hypoglycemia (<3.9 mmol/L) overall and night-time were similar in the two groups.
In correlation analyses (Table 3), the serum CP showed a positive correlation with time-in-range (3.9-10 mmol/L) in the honeymoon group ( gure 2), along with inverse correlations with the mean sensor glucose, mean sensor glucose SD and the median time spent in level 1 hyperglycemia. In the non-honeymoon group, serum CP was positively correlated with the time spent in level 1 hypoglycemia and level 1 hyperglycemia. However, serum PI showed no correlations with glycemic parameters in both groups ( gure 3).

Discussion
Recent developments in continuous and intermittently scanned continuous glucose monitoring systems provide not only more e cient regulation of type 1 diabetes treatment, but also new information on glycemic parameters from the time of diagnosis. In this study, by using intermittently scanned glucose monitoring systems, the glycemic parameters of children with type 1 diabetes who are in RP and those who are not in RP were compared. As expected, the glycemic parameters in the RP were found to be closer to the target levels but were far from those of non-diabetic subjects (10). It was found that while CP levels did have an effect on glycemic parameters during the honeymoon period, there was no correlation between PI levels and glycemic parameters. In terms of glycemic parameters during the honeymoon period, the CP level had a signi cant correlation with the TIR value. CP seems to be a useful biochemical tool, while PI was not a sensitive biochemical parameter.
New parameters are needed to de ne the honeymoon or preserved beta-cell reserve period. Although several studies (11)(12)(13) focused on the honeymoon period and investigated honeymoon-related factors, data related to the glycemic parameters during the honeymoon period are limited. Meng et al. examined the relationship between blood glucose uctuations during various phases of diabetes and oxidative stress and showed that the mean glucose, glucose SD, the mean amplitude of glycemic excursions (MAGE) and incremental area under the blood glucose curve (IAUC) levels during the honeymoon period were lower than those during the acute metabolic disturbance and long-standing phases (14). Similarly, in our study, the mean sensor glucose level, SD value, CV value, time-in-range, time in hyperglycemia were found to be lower in the honeymoon group, suggesting a better metabolic control ( Table 2). While the mean CV of 32.3% was lower than the target value of 36% in the CGMS consensus, (6) the mean SD value was higher than that found in a recent study analyzing the CGMS data of healthy children (9). In addition, time spent in level 1 hypoglycemia was found to be similar in the honeymoon vs. nonhoneymoon groups, indicating that glycemic control in RP is restricted.
CP is a useful and widely used method of assessing pancreatic beta-cell function. The formula for IDAA1c was derived using a higher CP cut-off value of 300 pmol/L. Therefore, although it previously played a role in de ning the honeymoon phase, (3), it is no longer used to de ne the honeymoon phase. Venous blood CP levels can be measured in the random, fasting, or stimulated state. Random samples are taken at any time during the day without consideration of recent food intake, whereas fasting samples are taken after an 8 to 10 hours fast (8). In the present study, serum CP levels were measured in the random state, and as expected the mean CP level was signi cantly higher in the honeymoon group than in the non-honeymoon group. Although CP is a widely accepted biochemical parameter for pancreatic beta-cell reserve, data on the association between CP levels and glycemic parameters are very limited. In the present study, the presence of a correlation between CP levels and glycemic parameters was evaluated. In the honeymoon group, the mean CP level was inversely correlated with the mean sensor glucose and SD glucose whereas it positively correlated with TIR. These ndings support that CP might be a relevant biochemical parameter in the honeymoon phase. However, the presence of a positive correlation between CP and time in both level 1 hypoglycemia and level 1 hyperglycemia in the non-honeymoon group suggests that CP is not a reliable parameter when it is low.
Recent studies reported an increased risk of developing 5-year T1D in antibody-positive rst-degree relatives having an increased PI / CP ratio (15). Similarly, the PI/CP ratio was found to be higher at the time of diagnosis of diabetes compared with the control group (16). Although PI is used as a beta-cell stress marker, two recent studies showed that PI secretion was maintained for a long time and that the level of CP was still detectable in individuals with T1D (17,18). In our study, both the PI level and PI / CP ratio were signi cantly higher in the honeymoon group. In the non-honeymoon group, however, two children had CP levels below the measurable limit despite detectable PI levels and two children had PI levels below the measurable limit despite measurable CP levels. Due to the small size of the patient groups, it is di cult to comment on the differences in the maintenance of CP and PI secretions. In this study, no signi cant correlation was detected between proinsulin levels and glycemic parameters in any group, suggesting that proinsulin is not as clinically reliable as CP in the management of diabetes.
One of the main limitations of our study was the random measurement of CP and PI levels. These measures may have been affected by the degree of fasting, which was not taken into account because we recruited patients during scheduled outpatient follow-up visits. Measurements of stimulated CP and PI secretions using tests such as a mixed-meal tolerance test may be bene cial, as well. However, this study design would require more inconvenience to the participants and their families. Moreover, Watkins et al conducted a research by random sampling for the CP and PI measurement in order to evaluate the β-cell function in persons with T1D (5) and Leighton et al in their review article had stated that venous blood CP levels can be measured in the random, fasting, or stimulated state (8). Another limitation was the lack of simultaneous blood glucose measurements with the use of isCGMS which does not require calibration. The accuracy of isCGMS was reported to be lower during hypoglycemia than during euglycemia and hyperglycemia (19). However, time spent in level 1 hypoglycemia was not high and similar in both groups. Despite these limitations, we feel that this pilot study is important as it is the rst study to evaluate glycemic parameters in the honeymoon period with the use of isCGMS data and to examine the relationship between these parameters and Beta-cell reserve markers CP and PI. There is no doubt that a future study with larger patient numbers would show more generalizable results.
In conclusion, although the glycemic pro le during the honeymoon period was better than that in the non-honeymoon group, the glycemic variability during this period was not as low as expected. It is also important to continue efforts to improve and maintain metabolic control during the honeymoon period. Subcutaneous insulin infusion or automated insulin delivery systems may help to achieve better glycemic control in the remission phase as well. CP seems to be a useful biochemical tool, while PI was not a sensitive biochemical parameter. Further studies are needed to determine if PI might prove to be a useful parameter in clinical follow-up.