Clinical Characteristics and Predictors of Glycemic Control During the First 24 Months After Diagnosis of Type 1 Diabetes

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for an opportunity to assess this manuscript, which addresses an important burden of diabetes care. Maintenance of good glycemic control remains one of the greatest challenges in some of the patients.  Identifying early predictors of further glycemic control may allow for  more individualized and earlier therapeutic approaches.

The manuscript is generally well-structured and communicates the scientific content effectively.

However, I would like to address some important issues:

1. In the Results, among other laboratory parameters, proinsulin levels are presented, however proinsulin was not mentioned in Materials and Methods section. Please clarify.
2. Presenting results and performing comparisons, authors compare BMI and creatinine levels between different age groups, despite the physiological differences of normal ranges at different age. In my opinion, to draw conclusions Authors should use either different norms for different age groups, or use BMI-SDS values instead just BMI.
3. Paragraph regarding “observed association between age at diagnosis and the number of positive autoantibodies” does not belong to results – as it tries to explain the finding it should be moved to the Discussion or to the imitations section.
4. There are some typographic errors, eg lack of the bracket [ (p=.034 ]
5. As mentioned before, in the analysis of the influence of BMI on the further glycemic control uses adult thresholds for diagnosing underweight, normal weight and obesity. Authors draw a statements that “average HbA1c levels over 24-months differed between BMI categories” and “BMI at diagnosis, particularly underweight, is associated with poorer long-term glycemic control” – however, as BMI ranges in children are substantially different, Authors should take it into consideration and re-evaluate this specific statistical model.
6. Paragraph 3.2.3. : the title “Adjusted analysis of predictors on HbA1c” is not informative enough, as previous paragraph refers to HbA1c at baseline, this one should clearly refer to the “further HbA1c” or “HbA1c in following months”
7. In the beginning of the Discussion Authors state that “younger age at diagnosis […] was the only independent predictor of PERSISTENTLY higher HbA1c” – however the time of the observation was only 24-months post-diagnosis. In my opinion, this statement is too strong and should be rephrased using eg. “early suboptimal glycemic control”.
8. Again, in the discussion, Authors analyze the effect of under- and overweight on glycemic control – while not adjusting BMI thresholds for patients’ age.
9. As the main finding of the study was that age at T1D onset was the strongest factor influencing glycemic control for the first 24-months after diagnosis, the Discussion section should include possible explanation why this particular factor makes such and impact.

Therefore, I recommend that the manuscript undergo major revision before accepting it for publication.

Author Response

First of all, we express our thanks to this reviewer for the careful evaluation of our manuscript. We are confident that the  amendments made approved the quality of the manuscript. 

See attached the point-to-point answers to the comments raised. 

 

Reviewer: Thank you for an opportunity to assess this manuscript, which addresses an important burden of diabetes care. Maintenance of good glycemic control remains one of the greatest challenges in some of the patients.  Identifying early predictors of further glycemic control may allow for more individualized and earlier therapeutic approaches.

The manuscript is generally well-structured and communicates the scientific content effectively.

We express our thanks for the careful review and highly appreciate the positive feedback. 

However, I would like to address some important issues:

Comment 1: In the Results, among other laboratory parameters, proinsulin levels are presented, however proinsulin was not mentioned in Materials and Methods section. Please clarify.

Response 1: We thank this reviewer for making us aware about this discrepancy. We adapted accordingly and added proinsulin to the methods section.

Comment 2: Presenting results and performing comparisons, authors compare BMI and creatinine levels between different age groups, despite the physiological differences of normal ranges at different age. In my opinion, to draw conclusions Authors should use either different norms for different age groups, or use BMI-SDS values instead just BMI.

Response 2: We thank the reviewer for this valuable comment and fully agree with the concern raised. As correctly pointed out, BMI is not an optimal parameter in pediatric populations due to the strong age- and sex-dependent physiological variations. Consequently, direct comparisons between pediatric and adult BMI values may lead to misleading interpretations if age-adjusted measures such as BMI-SDS are not used. In light of this, and to avoid inappropriate comparisons between age groups with different physiological reference ranges, we decided to remove the entire BMI comparison data from the manuscript. This change ensures that our analyses are not affected by potentially biased interpretations of BMI values across pediatric and adult populations. The manuscript has been revised accordingly.

 

Comment 3: Paragraph regarding “observed association between age at diagnosis and the number of positive autoantibodies” does not belong to results – as it tries to explain the finding it should be moved to the Discussion or to the imitations section.

Response 3: We agree that this paragraph better suits to the discussion section. We expanded the discussion in this matter underlining the limitation of the timely factor potentially influencing the number of available antibody panels.

 

Comment 4: There are some typographic errors, eg lack of the bracket [ (p=.034 ]

Response 4: We thank the reviewer for pointing out these typographic inconsistencies. The manuscript has been carefully checked again for similar irregularities throughout the entire text. Where necessary, corrections have been made in accordance with the formatting guidelines of MDPI. All corresponding changes have been implemented and highlighted in yellow in the revised manuscript to ensure transparency. We also removed the headlines from the graphs, as these are mentioned in the caption.

 

Comment 5: As mentioned before in the analysis of the influence of BMI on the further glycemic control uses adult thresholds for diagnosing underweight, normal weight and obesity. Authors draw a statements that “average HbA1c levels over 24-months differed between BMI categories” and “BMI at diagnosis, particularly underweight, is associated with poorer long-term glycemic control” – however, as BMI ranges in children are substantially different, Authors should take it into consideration and re-evaluate this specific statistical model.

Response 5: We again thank the reviewer for this important comment and fully agree with the concern raised. As mentioned in our response to the previous comment regarding BMI, we acknowledge that BMI thresholds differ substantially between pediatric and adult populations, and that applying adult BMI categories to children may lead to biased interpretations.

 

Comment 6: Paragraph 3.2.3. : the title “Adjusted analysis of predictors on HbA1c” is not informative enough, as previous paragraph refers to HbA1c at baseline, this one should clearly refer to he “further HbA1c” or “HbA1c in following months”

Response 6: We thank the reviewer for this helpful comment. As correctly noted, the previous paragraph refers to HbA1c at baseline, while this section was intended to address HbA1c during follow-up. During the revision process, we also identified that the corresponding paragraph originated from an earlier preliminary version of the manuscript. As part of the revision, this section—including the paragraph describing the adjusted model—has been removed from the manuscript. The text has been revised accordingly to avoid potential confusion.

 

Comment 7: In the beginning of the Discussion Authors state that “younger age at diagnosis […] was the only independent predictor of PERSISTENTLY higher HbA1c” – however the time of the observation was only 24-months post-diagnosis. In my opinion, this statement is too strong and should be rephrased sing eg. “early suboptimal glycemic control”.

Response 7: We thank the reviewer for this important comment and agree that a follow-up period of 24 months cannot be considered long-term glycemic control. In response to this suggestion, we have revised the wording throughout the manuscript to avoid the term “long-term” and instead use more appropriate phrasing reflecting the 24-month observation period.

Accordingly, the respective statements in the discussion have been rephrased to better reflect early glycemic outcomes after diagnosis, and the title of the manuscript has been adjusted as well. All relevant changes have been implemented in the revised version of the manuscript.

 

Comment 8: Again, in the discussion, Authors analyze the effect of under- and overweight on glycemic control – while not adjusting BMI thresholds for patients’ age.

Response 8: We also removed BMI related content from the discussion section.

 

Comment 9: As the main finding of the study was that age at T1D onset was the strongest factor influencing glycemic control for the first 24-months after diagnosis, the Discussion section should include possible explanation why this particular factor makes such and impact.

Response 9: We agree with the reviewer that our main finding requires a more detailed explanation. Thus, we adapted this paragraph in the discussion section as follows:

First, early-onset T1D is frequently associated with a more aggressive autoimmune process and more rapid β-cell destruction. Younger children tend to have lower residual C-peptide levels, reflecting diminished endogenous insulin secretion and reduced capacity to buffer glycemic excursions. In the Diabetes Control and Complications Trial Research Group (1993), preservation of C-peptide was strongly associated with improved glycemic outcomes and reduced complications [15]. Greenbaum et al. (2012) demonstrated that younger age at diagnosis is linked to lower stimulated C-peptide levels [16]. Similarly, Barker et al. (2014) reported age-dependent decline in β-cell function in new-onset T1D [17]. Registry analyses from the Type 1 Diabetes Exchange further show that children diagnosed at younger ages have higher mean HbA_1c values compared with young adults [18].

Second, developmental physiology complicates glycemic management. Young children exhibit unpredictable food intake, variable activity, and heightened insulin sensitivity, increasing hypoglycemia risk. Fear of hypoglycemia among caregivers has been associated with higher HbA_1c [19]. During puberty, increased growth hormone and sex steroids induce insulin resistance, contributing to deterioration in glycemic control [20].

Third, psychosocial factors are critical. Glycemic outcomes in childhood are strongly influenced by family functioning and parental involvement. Greater family conflict isassociated with higher HbA_1c [21]. Parental stress and depressive symptoms correlate with poorer adherence and metabolic control. Survey analyses indicated that psychiatric stress among patients and their caregivers is directly associated with younger age at diabetes diagnosis, supporting our observation that earlier onset may be linked to poorer glycemic control [22,23].

Reviewer 2 Report

Comments and Suggestions for Authors

The authors performed a retrospective study on data sets of anthropometric and metabolic parameters of T1D patients collected over a ~20-year period and concluded that age at diagnosis was the only predictive factor of dysregulated glycemic control during the first 24 months after T1D onset.  Specifically, patients diagnosed with T1D before age 18 consistently showed higher HbA1c levels over 24 months, indicating a significant risk for poor long-term glycemic control.

The manuscript is well written, well organized, and is founded on sound statistical analyses of retrospective data sets that were well described and performed.

The following are a few minor concerns that if addressed should improve the manuscript.

1. In Fig. 1, the legend should state that the data are "baseline values” or from the “Index visit” or at “diagnosis ".
2. With respect to section “2. Predictors of HbA1C progression over 24 months with linear mixed model”, the manuscript should include most other longitudinal HbA1c data sets of other grouping (i.e., factors or predictors), similar to that in Fig .2 showing longitudinal data with respect to “Age”. While it is appreciated that Age was ultimately the only actual “predictor” of glycemic control, the readers should also be privy to mean longitudinal data sets of the other predictors as well, which could be included as Supplemental Information.  That is, without these data sets, the manuscript only shows the statistical outcomes (e.g., displayed in Table 2) without presenting the actual unadjusted data.
3. Because the statistical outcomes in Table 2 are fundamental to the claims of the study, it is important to provide a bit more description/explanation of the statistical values that support such conclusions/claims; e.g., in the caption (or in the Methods), describe for the scientist that is not a biostatistician what is meant by "estimates" and the meaning of positive vs. negative % values.  In addition, perhaps clarify what “predictor” and “HbA1c (%)” mean; e.g., “Predictor on HbA1c (%) over 24 months” is confusing to the reader at first reading.
4. There are several areas in which the authors did not bother to actually reference the data sets in Table 2 and Fig. 3 that are discussed in the Results section; e.g., Sections 3.2.2. and 3.2.3 should reference Fig. 3. Section “3. Additional unadjusted outcomes” appears to reference Table 2?

Author Response

First of all, we express our thanks to this reviewer for the careful evaluation of our manuscript. We are confident that the  amendments made approved the quality of the manuscript. 

See attached the point-to-point answers to the comment.

 

The authors performed a retrospective study on data sets of anthropometric and metabolic parameters of T1D patients collected over a ~20-year period and concluded that age at diagnosis was the only predictive factor of dysregulated glycemic control during the first 24 months after T1D onset.  Specifically, patients diagnosed with T1D before age 18 consistently showed higher HbA1c levels over 24 months, indicating a significant risk for poor long-term glycemic control.

The manuscript is well written, well organized, and is founded on sound statistical analyses of retrospective data sets that were well described and performed.

The following are a few minor concerns that if addressed should improve the manuscript.

Comment: We thank the reviewer for the constructive and encouraging feedback. We have addressed the minor concerns raised.

Comment 1: In Fig. 1, the legend should state that the data are "baseline values” or from the “Index visit” or at “diagnosis ".

Response 1: We clarified this matter in the caption of figure 1 by adding “at baseline”.

 

Comment 2: With respect to section “3.2. Predictors of HbA1C progression over 24 months with linear mixed model”, the manuscript should include most other longitudinal HbA1c data sets of other grouping (i.e., factors or predictors), similar to that in Fig .2 showing longitudinal data with respect to “Age”. While it is appreciated that Age was ultimately the only actual “predictor” of glycemic control, the readers should also be privy to mean longitudinal data sets of the other predictors as well, which could be included as Supplemental Information.  That is, without these data sets, the manuscript only shows the statistical outcomes (e.g., displayed in Table 2) without presenting the actual unadjusted data.

Response 2: We thank the reviewer for this helpful suggestion. To allow readers to visualize the longitudinal HbA1c patterns across the investigated predictors, we added graphical representations of HbA1c trajectories for the main predictor variables in Figure 3 (panels A–F). These figures display the estimated HbA1c trajectories derived from the statistical models over the 24-month follow-up period. The p-value represented the overall HbA1c trend in the total cohort without considering the age group. It was removed as it seems to be confusing for the reader. While Figure 2 shows the raw longitudinal HbA1c course by age group, Figure 3 provides model-based trajectories for the other predictors, allowing readers to visualize the temporal patterns underlying the statistical results.

 

Comment 3: Because the statistical outcomes in Table 2 are fundamental to the claims of the study, it is important to provide a bit more description/explanation of the statistical values that support such conclusions/claims; e.g., in the caption (or in the Methods), describe for the scientist that is not a biostatistician what is meant by "estimates" and the meaning of positive vs. negative % values.  In addition, perhaps clarify what “predictor” and “HbA1c (%)” mean; e.g., “Predictor on HbA1c (%) over 24 months” is confusing to the reader at first reading.

Response 3: We thank the reviewer for this helpful comment and agree that additional clarification improves the readability of the statistical results. Accordingly, we have expanded the description of the statistical outcomes presented in Table 2. Specifically, we clarified the meaning of the reported estimates and explained the interpretation of positive versus negative percentage values. These explanations have been added both to the caption of Table 2 and to the Methods section to ensure that the statistical results and their interpretation are more easily understood. Furthermore, irrelevant and insignificant data of table 2 which is not presented in figure 3 was removed.

 

Comment 4: There are several areas in which the authors did not bother to actually reference the data sets in Table 2 and Fig. 3 that are discussed in the Results section; e.g., Sections 3.2.2. and 3.2.3 should reference Fig. 3. Section “3. Additional unadjusted outcomes” appears to reference Table 2?

Response 4: We thank the reviewer for pointing this out. We have carefully revised the results section and added explicit references to the corresponding tables and figures wherever these were previously missing.

 

Reviewer 3 Report

Comments and Suggestions for Authors

For the Journal:
The manuscript was provided without line numbering, which makes it difficult to review and reference specific sentences. Please consider providing a version with line numbers to facilitate the review process.

Comments to the Authors:

1. The sentence describing the etiology of type 1 diabetes should begin with genetic predisposition. A more appropriate formulation would be: “The etiology of T1D is believed to involve a combination of genetic predisposition, virus-induced immune responses, and environmental factors [1].”
2. In reference [11] (Sheleme et al., 2020), the authors studied patients with both type 1 and type 2 diabetes. Therefore, the statement “Several predictors of insufficient glycemic control have been described for both type 1 and type 2 diabetes in the literature, including DKA at diagnosis, overweight and obesity, comorbidities, first-born status, baseline HbA1c >9.5%, a higher titer of GAD autoantibodies, low socioeconomic status, family history of diabetes, non-White ethnicity, and poor adherence to treatment or diet [9–11]” should clarify that the listed predictors apply to both T1D and T2D, rather than exclusively to T1D, or alternatively exclude predictors that are specific to T2D.
3. The authors should clarify how the diagnosis of T1D was established in the study cohort. In particular, please specify how individuals without detectable autoantibodies were classified. Autoantibody positivity is a hallmark of T1D development and classification. Approximately 90–95% of individuals with T1D have at least one islet autoantibody at diagnosis (Type 1 Diabetes: A Review, PMID: 41697686).
4. Figures containing graphs should indicate the statistical tests that were applied. In addition, on Figure 2 it is unclear between which age groups the reported statistical difference (p < 0.001) was observed and at which time points. Please clarify this in the figure or figure legend.

Author Response

First of all, we thank this reviewer for the careful evaluation of our manuscript. 

Comment 1: The sentence describing the etiology of type 1 diabetes should begin with genetic predisposition. A more appropriate formulation would be: “The etiology of T1D is believed to involve a combination of genetic predisposition, virus-induced immune responses, and environmental factors [1].”

Response 1: The text has been updated to reflect the suggested formulation to improve accuracy and clarity.

 

Comment 2: In reference [11] (Sheleme et al., 2020), the authors studied patients with both type 1 and type 2 diabetes. Therefore, the statement “Several predictors of insufficient glycemic control have been described for both type 1 and type 2 diabetes in the literature, including DKA at diagnosis, overweight and obesity, comorbidities, first-born status, baseline HbA1c >9.5%, a higher titer of GAD autoantibodies, low socioeconomic status, family history of diabetes, non-White ethnicity, and poor adherence to treatment or diet [9–11]” should clarify that the listed predictors apply to both T1D and T2D, rather than exclusively to T1D, or alternatively exclude predictors that are specific to T2D.

Response 2: We thank the reviewer for the careful observation regarding the cited study. As correctly pointed out, the publication by Sheleme et al. (2020) included patients with both type 1 and type 2 diabetes and did not distinguish between the two diabetes types in the analysis of predictors of glycemic control. To avoid potential confusion, we have removed this reference from the manuscript. Instead, we have added an additional reference that specifically investigates predictors of glycemic control in individuals with type 1 diabetes (Aeppli et al 2019).

 

Comment 3: The authors should clarify how the diagnosis of T1D was established in the study cohort. In particular, please specify how individuals without detectable autoantibodies were classified. Autoantibody positivity is a hallmark of T1D development and classification. Approximately 90–95% of individuals with T1D have at least one islet autoantibody at diagnosis (Type 1 Diabetes: A Review, PMID: 41697686).

Response 3: We thank the reviewer for this important comment. The selection of participants in our study was based on the diagnosis generated by the registry search, which served as the primary source for cohort identification. In clinical practice in Austria, in cases where islet autoantibodies are not detectable, the diagnosis of type 1 diabetes is typically established based on absent or markedly reduced C-peptide levels, in combination with the clinical presentation and after exclusion of other forms of diabetes, such as monogenic diabetes.

In addition, we would like to emphasize the temporal context of our cohort. Our analysis includes individuals diagnosed with diabetes starting in 2001, a time when autoantibody panels were more limited than today because fewer islet autoantibodies were known and routinely tested. As a consequence, some individuals classified as antibody-negative at that time might have been positive if broader antibody panels (e.g., including ZnT8) had been available. This may partly explain the slightly higher proportion of individuals without detectable autoantibodies in our cohort.

This important limitation has been added to the discussion section.

 

Comment 4: Figures containing graphs should indicate the statistical tests that were applied. In addition, on Figure 2 it is unclear between which age groups the reported statistical difference (p < 0.001) was observed and at which time points. Please clarify this in the figure or figure legend.

Response 4: We thank the reviewer for this helpful comment. The previously stated p-value indicated the HbA1c trend over time in the total cohort and not the comparison of age groups. In the revised manuscript, the p-value previously shown in Figure 2 has been removed. Figure 2 is intended to provide a descriptive visualization of the overall HbA1c trajectory across age groups during the first 24 months after diagnosis. The statistical comparisons of predictors and their interaction with time were performed using linear mixed-effects models, and these results are presented in the corresponding analyses and illustrated in Figure 3. We clarified this in the revised figure legend to avoid potential misunderstanding.

Round 2

Reviewer 1 Report

Comments and Suggestions for Authors

Thank you for considering my comments. All the issues I recommened for revision have been addressed. Please review the corrected manuscript for typographical errors, as there are several typos throughout the corrected part.

I recommend the manuscript for publication after those minor revisions.

Author Response

Thank you very much for your careful re-evaluation of our revised manuscript and for your positive feedback.

Following your recommendation, we have carefully rechecked the entire manuscript once more and corrected few typographical and minor language errors throughout the text to further improve clarity and readability. The corrections made following your review are marked in green in the attached manuscript.