Correspondence Concerning ‘Aerobic Exercise Plus Hypocaloric Diet on Serum Leptin Concentrations in Adults With Primary Hypertension and Overweight/Obesity: Results From the EXERDIET‐HTA Trial’

1.

We thank the correspondent for a careful and constructive reading of our substudy and for raising concerns that provide an opportunity to clarify our methods, findings, and conclusions. We address each of the five points in turn.

2. Concern 1: Inference From Within‐Group p Values in the Context of a Non‐Significant Time × Group Interaction

The correspondent correctly identifies a tension in our presentation. The time × group interaction for serum leptin was not statistically significant (p > 0.05, Table S3), which is the appropriate test for between‐group inference in a two‐arm repeated‐measures design. We acknowledge that the within‐group p values (exercise, EX: p = 0.003; attention control, AC: p = 0.06) reflect changes from each group's baseline and, in isolation, do not establish that the EX group changed more than the AC group. The distinction is well established in the statistical literature. Gelman and Stern formally demonstrated that changes in statistical significance between two comparisons are not, in themselves, evidence of a difference between those comparisons, and that the correct test of group differences is the interaction term [1]. The test was non‐significant in our data. The claim in the Highlights section that supervised EX ‘reduced [leptin] more than only a hypocaloric diet’ was intended to convey the direction of the observed effect estimates, not to assert a formally tested superiority. We accept that this framing is imprecise and potentially misleading. A more accurate characterisation of the primary finding is that both groups showed reductions in leptin concentrations following the intervention, with a numerically larger reduction in the EX group (36% vs. 23%), but that the between‐group difference did not reach statistical significance.

3. Concern 2: Pooling of Three EX Sub‐Protocols

The three EX sub‐protocols differed in volume and intensity: high‐volume, moderate‐intensity continuous training (45 min/session); high‐volume, high‐intensity interval training (45 min total volume) and low‐volume, high‐intensity interval training (20 min total volume). We concur that these represent meaningfully different EX doses. The decision to pool all of them was based on the original design of the EXERDIET‐HTA trial, in which the EX arms shared the common intervention component of supervised aerobic EX combined with the same hypocaloric DASH diet, and on a preliminary repeated‐measures ANCOVA that found no significant time × group interaction across the three EX sub‐groups for leptin (p > 0.05, Table S1). The aim of the present substudy was not to compare the three EX modalities, but rather to evaluate the overall effect of adding supervised aerobic EX to a dietary intervention on serum leptin concentrations. We acknowledge the limitation that the correspondent identifies: sub‐group cells of approximately 17 participants each are underpowered to detect a dose–response effect on leptin, so a null interaction test does not establish equivalence. The low‐volume protocol, delivering approximately 40 min of supervised EX per week, likely falls below the thresholds cited in the literature for significant leptin reduction (180 min/week moderate‐intensity or 120 min/week high‐intensity) [2]. Its inclusion in the pooled EX group may therefore have attenuated the estimated EX effect. Objective monitoring of physical activity using accelerometry (recommended as a minimally biased method for detecting between‐group differences in activity change in RCTs) [3] was not employed in this study, which limits the precision with which EX dose can be characterised. We reported this as a limitation (Section 4.1). Moreover, the disaggregated data are available in Table S1. We agree that future work in this area should be designed with sufficient statistical power to examine dose–response relationships for leptin as a primary outcome, and we encourage readers to consult Table S1 when interpreting the pooled effect estimation.

4. Concern 3: Classification of the Comparator as an AC

The correspondent raises a legitimate point regarding the characterisation of the AC arm. Participants in that group received: a hypocaloric DASH diet identical to that prescribed to the EX group, biweekly weight monitoring with nutritional counselling, and advice to follow the WHO physical activity recommendations (i.e., at least 30 min of moderate‐intensity aerobic EX on 5–7 days per week, combined with dynamic resistance EX). This recommendation constitutes a substantive lifestyle prescription rather than an inert or purely social comparator. We used the term ‘attention control’ to mean that participants received equivalent professional contact and dietary support, without the structured, supervised EX sessions. In the literature, the distinction between ‘attention control’ and an ‘active comparator’ is methodologically consequential. Previous reports have distinguished between ACs—in which participants receive contact time and social interaction without the active ingredient—and active comparators, which deliver an independent, meaningful intervention [4, 5]. We recognise, however, that this terminology carries connotations of an inert comparator that do not fully reflect the AC's design. A more precise descriptor would be ‘active comparator receiving physical activity counselling plus hypocaloric diet’. We further acknowledge, as stated in Section 4.1, that we did not collect objective physical activity data from the AC group following counselling, so the actual EX dose achieved by AC participants is unknown. This limitation prevents us from attributing the observed leptin reductions solely to supervised EX.

5. Concern 4: Differential Attrition and Complete‐Case Analysis

The correspondent accurately notes the unequal number of participants included in the final analyses: 17 of 18 participants in the AC group and 30 of 51 participants in the EX group. We agree that this imbalance is methodologically relevant and should be taken into account when interpreting the findings. As stated in Section 4.1, the present substudy was analysed using a completers‐only approach because some participants did not attend the post‐intervention assessment, resulting in missing follow‐up data. Consequently, an intention‐to‐treat analysis was not performed. We acknowledged in the manuscript that this approach may introduce attrition bias, as participants who completed the assessments may differ from those with missing follow‐up data, and that the findings should therefore be interpreted with caution. However, the present analysis was conducted as a substudy of the EXERDIET‐HTA trial, focused specifically on participants with available leptin measurements at the required assessment points. Leptin was not the primary endpoint of the original trial, and the substudy was not originally powered or designed to perform a primary intention‐to‐treat analysis for this biochemical outcome. For this reason, the complete‐case analysis was used to describe the observed leptin response among participants with valid outcome data. We agree that pre‐specified sensitivity analyses, such as mixed‐model approaches or multiple imputation under transparent assumptions, would have strengthened the assessment of the robustness of the findings. Nevertheless, we would not interpret the complete‐case findings as definitive evidence of a treatment effect in the full randomised population. Rather, they should be interpreted as complete‐case estimates from an exploratory substudy. In this context, the EX group's within‐group leptin reduction should be viewed cautiously, as the complete‐case approach may overestimate the effect in the full enrolled population. Future studies examining leptin as a primary outcome should pre‐specify strategies for handling missing outcome data and should prioritise intention‐to‐treat or sensitivity analyses to assess the robustness of intervention effects.

6. Concern 5: Reversion of Leptin Concentrations at 6‐Month Follow‐Up

The 6‐month follow‐up data show that leptin concentrations in both groups returned to values statistically indistinguishable from baseline (Figure 2; PRE‐to‐6 months comparisons: p > 0.05 for leptin in both groups, Table S3). Specifically, leptin in women returned from 51.6 ng/mL at baseline to 50.5 ng/mL at 6 months, and in men from 20.7 to 18.6 ng/mL. We reported this transparently and discussed its mechanistic implications in Section 4, noting the concomitant increases in fat mass and the leptin/FM ratio during the unsupervised follow‐up period. The pattern of leptin rebound following cessation of supervised intervention is consistent with published evidence [6]. These findings support the view that supervised EX and dietary intervention produce transient rather than sustained changes in leptin homeostasis in the absence of behavioural support. We agree with the correspondent that complete reversion of the primary outcome at the only follow‐up time point is a substantive finding that constrains the clinical interpretation of the intervention. A transient reduction in leptin, fully reversed within 6 months of supervision ending, has limited implications for long‐term cardiometabolic management unless accompanied by a maintenance strategy. We intended the conclusion that ‘sustained lifestyle modifications are essential’ to acknowledge this. However, we accept that the framing of the overall conclusion, describing the intervention as ‘an effective strategy for regulating leptin’, could be read as overstating the durability of the effect. A more complete summary of the findings would emphasise that the reduction was significant during the supervised intervention period but was not maintained independently thereafter.

7. Concern 6: Whether the EX‐Associated Leptin Reduction Persists After Adjustment for Weight Change

The correspondent raises a precise mechanistic question: given that BMI is the strongest independent correlate of serum leptin in our regression model (12% of variance overall; 24% in women), and given that the EX group lost slightly more body mass than the AC group, was the additional leptin reduction in the EX group entirely attributable to greater fat‐mass loss, or did the EX stimulus contribute independently? The correspondent is correct that a formal adjustment analysis—specifically, a between‐group comparison of Δleptin after covarying for Δfat mass—was not performed. We accept this as an analytical gap. We wish to clarify, however, that our Discussion did not claim EX reduces leptin independently of all adiposity‐related changes. Rather, we proposed two complementary mechanistic pathways, each grounded in the literature and extending beyond a purely adiposity‐mediated account. The first is improvement in body composition—specifically, a decrease in fat mass percentage accompanied by relative preservation of fat‐free mass (FFM). The EX group maintained significant reductions in FM% at 6 months follow‐up. In contrast, the AC group did not, suggesting that supervised EX produced a qualitatively different pattern of adipose tissue remodelling beyond the shared caloric deficit. The significant Spearman correlations between Δfat mass and Δleptin (r = 0.351, p = 0.006) are directly consistent with this pathway. Critically, FFM was independently and inversely associated with leptin in our sex‐stratified regression for women (β = −0.443, p < 0.05), consistent with myokine‐mediated downregulation of adipose tissue leptin production—specifically via irisin, IL‐6, and IL‐15 release from contracting skeletal muscle, which has been shown to suppress adipocyte leptin secretion through muscle–fat crosstalk [7, 8]. The second is enhanced leptin sensitivity mediated by favourable metabolic adaptations associated with regular aerobic EX. Aerobic training upregulates AMPK/PGC‐1α signalling in skeletal muscle, improving mitochondrial biogenesis, insulin sensitivity and systemic anti‐inflammatory capacity [9]. These adaptations have been shown to restore hypothalamic responsiveness to leptin signalling, including reactivation of leptin receptor‐positive neurons in the ventromedial hypothalamic nucleus, partly independently of fat mass [10]. The increase in the leptin/FM% ratio observed in both groups from POST to 6 months—which we interpreted as indicating greater leptin resistance following cessation of supervision—is consistent with reversal of this EX‐induced sensitisation, and aligns with the observation that VO_2peak, a surrogate of the mitochondrial and anti‐inflammatory adaptations underpinning leptin sensitisation, was maintained only in the EX group at 6 months. We fully concede that the decisive statistical test (i.e., a repeated‐measures ANCOVA with Δfat mass entered as a time‐varying covariate alongside the group × time interaction) was not conducted. However, our expectation, given the already non‐significant time × group interaction and the primacy of FM change as a leptin correlate in our data, is that the residual group effect after FM adjustment would remain non‐significant, which supports the conservative conclusion that the numerically larger leptin reduction in the EX group was primarily driven by greater fat‐mass loss. Nevertheless, the present design cannot fully resolve the proportion of that additional fat‐mass loss attributable to the EX stimulus versus the incidental augmentation of caloric deficit, nor whether the leptin sensitisation mechanism contributed an independent component during the supervised period. We flag both questions explicitly as priorities for future work.

8. Summary

We are grateful to the correspondent for a rigorous and fair critique. In summary, we accept that: (1) between‐group inference should rest on the interaction test rather than within‐group p values, and the interaction was not significant; (2) pooling of heterogeneous protocols based on an underpowered null test warrants caution; (3) the comparator arm is more precisely described as an ‘active comparator’ than an ‘attention control’; (4) the completers‐only analysis in the context of differential attrition likely overestimates the EX effect; (5) the conclusion should more clearly reflect the complete reversion of leptin at 6 months and (6) the data do not formally test whether EX reduces leptin beyond its weight‐loss effect, and the mechanistic pathways we proposed (i.e., body composition improvement and enhanced leptin sensitivity) while biologically plausible and internally consistent with our data, require a weight‐adjusted analysis or an isocaloric design to be definitively attributed to the exercise stimulus per se. These points collectively indicate that the most defensible summary of the data is: both a supervised exercise plus hypocaloric diet programme and physical activity counselling plus hypocaloric diet might induce clinically meaningful but transient within‐group reductions in serum leptin in adults with primary hypertension and overweight/obesity, with no statistically significant difference between groups, and with full reversion in both groups at 6 months. Whether the numerically larger reduction in the EX group reflects a specific exercise effect beyond fat‐mass loss remains to be determined. We hope this response clarifies the scope and limitations of our findings and contributes to a more precise understanding of leptin dynamics in this population.

We remain committed to transparent reporting and welcome further dialogue.

Conflicts of Interest

The authors declare no conflicts of interest.

Data Availability Statement

The data underlying this study are available from the corresponding author upon reasonable request.