Impact of a History of Obesity on Diabetic Microvascular Complications in Patients with Type 2 Diabetes

Round 1

Reviewer 1 Report

Comments and Suggestions for Authors

The manuscript written by Akifumi Kushiyama et colleagues evaluates the association between lifetime maximal BMI and the development of microvascular complications in Japanese patients with type 2 diabetes mellitus. The topic is relevant and the use of the historical maximum BMI as an indicator of cumulative exposure to obesity represents an innovative approach. The results suggest that a history of severe obesity is independently associated with the progression of proteinuria, while associations with decline in eGFR and progression of retinopathy lose significance after multivariate adjustment.

ABSTRACT

• The authors state that historical maximum BMI represents a critical predictor of diabetic microvascular complications. However, after multivariate adjustment, the association remains significant solely for the progression of proteinuria. The associations with eGFR decline and retinopathy are no longer significant. It is therefore recommended that the conclusions be reformulated in a more cautious manner.
• The use of the expression "microvascular complications" could be misleading since the most robust result concerns only the progression of proteinuria.

INTRODUCTION

• The concept of "metabolic memory" is extended to obesity, although it is generally developed to describe the long-term effects of glycemic exposure. The authors should clarify that the application to obesity represents a biological hypothesis and not an established paradigm.
• It would be useful to better clarify which gap in the literature the present study intends to fill compared to the previous Japanese study cited [Tanaka, S.; Tanaka, S.; Iimuro, S.; Ishibashi, S.; Yamashita, H.; Moriya, T.; Katayama, S.; Akanuma, Y.; Ohashi, Y.; Yamada, 455 N., et al. Maximum BMI and microvascular complications in a cohort of Japanese patients with type 2 diabetes: the Japan 456 Diabetes Complications Study. J Diabetes Complications 2016, 30, 790-797, doi:10.1016/j.jdiacomp.2016.02.020.].
• It is suggested to investigate the clinical rationale of the classification into five categories of BMI and the potential added value compared to using a simple cutoff for obesity.

RESULTS

• The group with a maximum BMI <18.5 kg/m² includes only one patient and does not appear statistically interpretable. It is suggested that it be excluded from the main descriptive analyses.
• The analysis of BMI trajectories is particularly interesting but difficult to interpret. Weight loss prior to diagnosis may reflect very different phenomena, including voluntary lifestyle improvements or weight loss secondary to uncontrolled diabetes.
• Figures 2-6 should report the number of subjects at risk below the Kaplan-Meier curves.

DISCUSSION

• The claim that antihypertensive therapy "masked" the effect of obesity on eGFR decline is not fully supported by the data. An alternative explanation could be confounding by indication.
• The therapeutic implications regarding the early use of SGLT2-inhibitors and GLP-1 receptor agonists should be presented with greater caution, as such interventions have not been directly evaluated.

Author Response

Response to Reviewer 1

General Comment:

The manuscript written by Akifumi Kushiyama et colleagues evaluates the association between lifetime maximal BMI and the development of microvascular complications in Japanese patients with type 2 diabetes mellitus. The topic is relevant and the use of the historical maximum BMI as an indicator of cumulative exposure to obesity represents an innovative approach. The results suggest that a history of severe obesity is independently associated with the progression of proteinuria, while associations with decline in eGFR and progression of retinopathy lose significance after multivariate adjustment.

Response:

We sincerely thank the reviewer for the positive evaluation of our study and for recognizing the innovativeness of using historical maximum BMI as an indicator of cumulative obesity exposure. We have carefully addressed all of your insightful suggestions to improve the rigor and clarity of our manuscript.

ABSTRACT

Comment 1: The authors state that historical maximum BMI represents a critical predictor of diabetic microvascular complications. However, after multivariate adjustment, the association remains significant solely for the progression of proteinuria. The associations with eGFR decline and retinopathy are no longer significant. It is therefore recommended that the conclusions be reformulated in a more cautious manner.

Comment 2: The use of the expression "microvascular complications" could be misleading since the most robust result concerns only the progression of proteinuria.

Response 1&2:

We completely agree with the reviewer's point. Since the independent association after multivariate adjustment was robustly observed only for the progression of proteinuria, generalizing the conclusion to "diabetic microvascular complications" as a whole could be misleading.

We have carefully revised the Abstract (Background, Results, and Conclusions sections) to tone down our conclusions and specifically emphasize "proteinuria progression" rather than broad "microvascular complications."

Revision in Abstract (Conclusions):

"Conclusions: A history of maximum lifetime obesity is a critical predictor of the development of proteinuria progression in Japanese patients with type 2 diabetes. A history of obesity, a simple clinical metric, is a valuable tool for identifying high-risk patients..."

 

INTRODUCTION

Comment 3: The concept of "metabolic memory" is extended to obesity, although it is generally developed to describe the long-term effects of glycemic exposure. The authors should clarify that the application to obesity represents a biological hypothesis and not an established paradigm.

Response 3:

Thank you for this crucial conceptual clarification. We agree that "metabolic memory" is historically and fundamentally an established paradigm for glycemic exposure. Its application to obesity remains a biological hypothesis. We have revised the Introduction to clearly state that applying this concept to historical adiposity is a hypothesis.

Revision in Introduction (Paragraph 2):

"...Consequently, when used alone, a present BMI assessment may underestimate the historical exposure to adiposity and the hypothesized cumulative metabolic memory associated with past obesity. As an example of this proposed legacy effect, high adiposity over adulthood is associated with..."

Comment 4: It would be useful to better clarify which gap in the literature the present study intends to fill compared to the previous Japanese study cited [Tanaka, S. et al. J Diabetes Complications 2016].

Response 4:

We appreciate this suggestion, which aligns with a similar point raised by Reviewer 2. We have expanded the Introduction to explicitly clarify that while Tanaka et al which was already cited as Ref [21] evaluated past obesity as a binary variable (), our study fills the literature gap by evaluating the severity-dependent (dose-response) impact of lifetime maximum BMI stratified into five detailed categories (including severe obesity ). This allows for clinical risk stratification based on peak adiposity burden.

Revision in Introduction (Paragraph 3):

A previous Japanese cohort study suggested that a lifetime maximum BMI of ≥ 25 kg/m2 is a significant risk factor for microvascular complications [21]. However, that study dichotomized past obesity at a single cutoff of 25 kg/m2, leaving the incremental impact of obesity severity unexamined. While a simple binary cutoff of 25 kg/m² is clinically convenient, it groups patients with mild obesity together with those with severe obesity. According to the Japan Society for the Study of Obesity (JASSO) guidelines [22], obesity is structured into distinct classes because advanced obesity severity (Classes II and III) is in-crementally associated with an exponentially heightened burden of metabolic and renal disease [23]. Therefore, evaluating past obesity using a detailed stratified approach is necessary to determine the clinical rationale for severity-dependent risk.

 

Comment 5: It is suggested to investigate the clinical rationale of the classification into five categories of BMI and the potential added value compared to using a simple cutoff for obesity.

Response 5:

Thank you for this insightful comment. The clinical rationale for using the five categories defined by the Japan Society for the Study of Obesity (JASSO) is that severe obesity (Classes II and III, BMI  and ) carries an exponentially higher risk for metabolic and target-organ damage compared to Class I obesity (BMI 25 to ). Utilizing a simple cutoff (e.g., ) overlooks the high-risk subpopulation of severely obese individuals, who may require more aggressive preventative strategies. We have added this rationale to the Introduction and expanded on it in the Discussion.

Introduction: the changes are indicated at Response 5.

 

RESULTS

Comment 6: The group with a maximum BMI  includes only one patient and does not appear statistically interpretable. It is suggested that it be excluded from the main descriptive analyses.

Response 6:

We agree with the reviewer. Because Group I (underweight, ) provides no statistical power and cannot be meaningfully interpreted, we have removed Group I from the main descriptive analysis in Table 1 to improve clarity. The baseline characteristic comparisons are now conducted across Groups II through V ().

Revision in Results & Table 1:

Group I has been omitted from Table 1, and the footnote has been updated to reflect that the descriptive statistics and statistical tests are applied exclusively to Groups II through V.

Comment 7: The analysis of BMI trajectories is particularly interesting but difficult to interpret. Weight loss prior to diagnosis may reflect very different phenomena, including voluntary lifestyle improvements or weight loss secondary to uncontrolled diabetes.

Response 7:

Thank you for highlighting this clinically important nuance. We completely agree that a pre-visit weight "decrease" is a double-edged sword: it could represent a favorable voluntary lifestyle modification or, conversely, metabolic wasting (hyperglycemic catabolism/glucosuria) from long-standing, poorly controlled, or undiagnosed diabetes.

Our finding that the "decrease" group had a higher incidence of complications strongly supports the latter explanation—that pre-visit weight loss in this cohort primarily reflects a longer period of untreated glucotoxicity. We have added this balanced interpretation to the Discussion section.

Revision in Discussion (Paragraph 1):

"Our sub-analysis stratified by historical BMI trajectory further corroborates this phenomenon. It is important to note that weight loss prior to the first visit may reflect distinct phenomena: voluntary lifestyle clinical improvements or weight loss secondary to uncontrolled diabetes (hyperglycemic catabolism). Given that the 'decrease' group in our study exhibited a higher incidence of complications, this trajectory most likely represents historical exposure to severe, untreated hyperglycemia and metabolic dysregulation prior to clinical presentation, rather than protective weight loss."

Comment 8: Figures 2-6 should report the number of subjects at risk below the Kaplan-Meier curves.

Response 8:

We agree that including the number of subjects at risk is standard practice and enhances the transparency of the Kaplan-Meier curves. We have updated Figures 2, 3, 4, 5, and 6 to include the "Number at risk" tables directly beneath the horizontal axes.

 

DISCUSSION

Comment 9: The claim that antihypertensive therapy "masked" the effect of obesity on eGFR decline is not fully supported by the data. An alternative explanation could be confounding by indication.

Response 9:

We deeply appreciate the reviewer's insightful methodological and clinical critique regarding the interpretation of Figure 4. We agree that our initial interpretation was incomplete, and the alternative explanation of "confounding by indication" is highly plausible.

In the cohort receiving antihypertensive therapy, the baseline risk of eGFR decline was uniformly elevated even among the normal weight (Group II) and Class I obesity (Group III) categories compared to the non-medicated cohort. This phenomenon strongly suggests confounding by indication: patients with greater baseline cardiovascular or renal strain—independent of obesity—were selectively prescribed these medications, thereby compressing the observable risk differentials across BMI groups.

However, we also note that in this treated cohort, the risk of eGFR decline did not worsen further in the severe obesity groups (Groups IV and V). This leaves two compelling possibilities: (1) a "ceiling effect," where the overarching risk induced by underlying hypertension/renal strain reaches a maximum plateau, masking the incremental impact of peak adiposity, or (2) a genuine therapeutic effect of renin-angiotensin system (RAS) inhibitors, which are frequently used in these patients and may have effectively mitigated the specific pathophysiology of obesity-induced glomerular hyperfiltration.

We have extensively revised the Discussion section to reflect this comprehensive, multi-faceted interpretation, moving away from a simplistic "masking" claim.

Revision in Discussion (Paragraph 3):

Interestingly, the association between lifetime maximum obesity and eGFR decline varied depending on the use of antihypertensive medication. In patients not receiving antihypertensive therapy, the deleterious effect of past severe obesity on GFR was clearly pronounced. In contrast, among patients taking antihypertensives, the event risk tended uniformly elevated across all BMI categories, including normal weight and Class I obesity. This suggests the presence of confounding by indication; patients with advanced underlying hypertensive stress or baseline renal vulnerability, regardless of their weight history, were appropriately channeled into receiving antihypertensive therapy, thereby elevating the risk of the entire treated sub-cohort. Interestingly, the severe obesity groups did not exhibit further renal deterioration. This lack of incremental risk in the severe obesity under treatment could point to a ceiling effect. Alternatively, the potential for a genuine renoprotective therapeutic effect remains; since renin-angiotensin system (RAS) blockers were highly prescribed in this cohort, these agents may have actively counteracted the intrarenal RAS pathway that drives obesity-induced glomerular hyperfiltration and subsequent renal injury [29,30].

Comment 10: The therapeutic implications regarding the early use of SGLT2-inhibitors and GLP-1 receptor agonists should be presented with greater caution, as such interventions have not been directly evaluated.

Response 10:

We agree with the reviewer's caution. Since our study is an observational retrospective cohort and did not directly evaluate the therapeutic efficacy of SGLT2 inhibitors or GLP-1 receptor agonists, we must avoid overstating these implications. We have rephrased this section to frame these treatments more cautiously as general guideline-supported options for high-risk patients, rather than direct conclusions from our data.

Revision in Discussion (Paragraph 5):

"Identifying patients with a history of severe obesity can serve as a vital trigger for closer clinical monitoring. While our study did not directly evaluate specific pharmacological interventions, modern guidelines, including the KDIGO and ADA Standards of Care, recommend considering next-generation cardiometabolic therapies such as SGLT2 inhibitors and GLP-1 receptor agonists for patients with T2D at high risk of renal and vascular complications [37,38]."

Reviewer 2 Report

Comments and Suggestions for Authors

L66: A previous Japanese cohort study suggested that a lifetime maximum BMI of ≥ 25 kg/m2 is a significant risk factor for microvascular complications [21]. However, that study dichotomized past obesity at 25 kg/m2, which is the obesity criterion applied exclusively in Japan and other Asian countries [22]. (Reference 22 included A cohort of individuals with type 2 diabetes from 59 institutes in Japan was followed for 8years). They concluded that Past obesity as well as current obesity were associated with increased risks of microvascular complications. Please clarify what additional knowledge this study provides beyond previous work

L131: From an initial cohort of 4,207 patients, 1,041 (832 men; mean age 54.9 ± 10.5 years) were included in the final analysis. Please clearly discus potential selection bias and whether excluded patients differed significantly from included patients.

Author Response

Response to Reviewer 2

Comment 1: L66: A previous Japanese cohort study suggested that a lifetime maximum BMI of  is a significant risk factor for microvascular complications [21]. However, that study dichotomized past obesity at ... Please clarify what additional knowledge this study provides beyond previous work.

Response:

We appreciate the opportunity to clarify the distinct incremental value of our work over the benchmark study by Tanaka et al. (the JDCS study) [21].

The primary additional knowledge our study provides is the severity-dependent relationship of historical obesity on specific complications. While Tanaka et al. treated past obesity as a simple binary variable ( vs. ), we stratified patients into five detailed groups based on Asian/Japanese criteria. This allowed us to discover that a history of severe obesity (BMI ) poses a uniquely high independent risk for proteinuria progression, a nuance completely masked by binary dichotomization. We have clarified this explicitly in the Introduction (and in our response to Reviewer 1, Comment 4) and Discussion.

Introduction

A previous Japanese cohort study suggested that a lifetime maximum BMI of ≥ 25 kg/m2 is a significant risk factor for microvascular complications [21]. However, that study dichotomized past obesity at a single cutoff of 25 kg/m2, leaving the incremental impact of obesity severity unexamined. While a simple binary cutoff of 25 kg/m² is clinically convenient, it groups patients with mild obesity together with those with severe obesity. According to the Japan Society for the Study of Obesity (JASSO) guidelines [22], obesity is structured into distinct classes because advanced obesity severity (Classes II and III) is incrementally associated with an exponentially heightened burden of metabolic and renal disease [23]. Therefore, evaluating past obesity using a detailed stratified approach is necessary to determine the clinical rationale for severity-dependent risk.

 

Discussion

Crucially, our utilization of the five-tiered JASSO classification [22] revealed severi-ty-dependent pathological dynamics, thereby demonstrating a clinical advancement over the previous benchmark Japanese cohort study [21].

 

Discussion (already discussed in previous version of manuscript)

The severely obese group was the most at risk for retinopathy, a finding potentially revealed for the first time in this analysis,

 

Comment 2: L131: From an initial cohort of 4,207 patients, 1,041 (832 men; mean age  years) were included in the final analysis. Please clearly discuss potential selection bias and whether excluded patients differed significantly from included patients.

Response 2:

Thank you for pointing out this potential limitation. In a retrospective cohort study, missing data or strict adherence to inclusion criteria can introduce selection bias.

A major reason for exclusion was the lack of maximum lifetime weight data or insufficient frequency of follow-up examinations. This could introduce selection bias, as patients included in the final analysis might represent a population with higher health literacy (capable of recalling past maximum weight) or better compliance with clinic visits.

We have performed a baseline characteristic comparison between the included and excluded cohorts. Generally, the excluded group did not show major clinical discrepancies in major baseline variables (e.g., age, HbA1c), but to maintain full transparency, we have explicitly discussed this potential selection bias in the Limitations section of the Discussion.

Revision in Discussion (Limitations):

Second, out of 4,207 initial patients, a significant portion was excluded due to missing historical weight data or insufficient follow-up intervals, which introduces a potential selection bias. For example, the final analyzed cohort may represent individuals with high health literacy or clinical adherence to remember the lifetime maximum body weight, which could limit the generalizability of our findings to the broader diabetic population.