Aerobic Exercise Training Response in Preterm-Born Young Adults with Elevated Blood Pressure and Stage 1 Hypertension: A Randomized Clinical Trial

Rationale Premature birth is an independent predictor of long-term cardiovascular risk. Individuals affected are reported to have a lower rate of V˙o2 at peak exercise intensity (V˙o2PEAK) and at the ventilatory anaerobic threshold (V˙o2VAT), but little is known about their response to exercise training. Objectives The primary objective was to determine whether the V˙o2PEAK response to exercise training differed between preterm-born and term-born individuals; the secondary objective was to quantify group differences in V˙o2VAT response. Methods Fifty-two preterm-born and 151 term-born participants were randomly assigned (1:1) to 16 weeks of aerobic exercise training (n = 102) or a control group (n = 101). Cardiopulmonary exercise tests were conducted before and after the intervention to measure V˙o2PEAK and the V˙o2VAT. A prespecified subgroup analysis was conducted by fitting an interaction term for preterm and term birth histories and exercise group allocation. Measurements and Main Results For term-born participants, V˙o2PEAK increased by 3.1 ml/kg/min (95% confidence interval [CI], 1.7 to 4.4), and the V˙o2VAT increased by 2.3 ml/kg/min (95% CI, 0.7 to 3.8) in the intervention group versus controls. For preterm-born participants, V˙o2PEAK increased by 1.8 ml/kg/min (95% CI, −0.4 to 3.9), and the V˙o2VAT increased by 4.6 ml/kg/min (95% CI, 2.1 to 7.0) in the intervention group versus controls. No significant interaction was observed with birth history for V˙o2PEAK (P = 0.32) or the V˙o2VAT (P = 0.12). Conclusions The training intervention led to significant improvements in V˙o2PEAK and V˙o2VAT, with no evidence of a statistically different response based on birth history. Clinical trial registered with www.clinicaltrials.gov (NCT02723552).

Preterm birth affects 10% of live births worldwide (1), with survival rates into adulthood continuing to improve (2). Being born preterm has now been identified as an independent risk factor for cardiovascular (3)(4)(5)(6) and pulmonary vascular diseases (7,8). In line with this, young adults who were born preterm have been shown to have early signs of cardiopulmonary remodeling (9), including limitations under physiological stress conditions (8).
Whole-body _ VO 2 measured at the ventilatory anaerobic threshold ( _ VO 2VAT ) and at peak exercise intensity ( _ VO 2PEAK ) are markers of exercise capacity and surrogate measures of cardiovascularrelated morbidity and mortality (10,11). Preterm-born individuals have a lower _ VO 2PEAK (12), whereas the results for _ VO 2VAT remain inconclusive (13)(14)(15)(16)(17)(18)(19). Given that _ VO 2 is the product of heart rate, stroke volume, and the arterio-mixed venous oxygen difference (20), respiratory and cardiovascular impairments are likely to affect multiple, interrelated components of the oxygen uptake, transport, and utilization chain (21). Indeed, evidence of an impaired cardiac response in preterm-born individuals during exercise is accumulating (8,15,22,23), and mechanical ventilatory constraints have been shown to partly explain reductions in exercise endurance in this population group (24). For instance, Yang and colleagues recently reported that pretermborn individuals have a reduced _ VO 2PEAK , likely resulting from the combined effects of impaired lung function and altered heart structure (25) .
Cardiovascular fitness in early life has been shown to be important for long-term health. Shah and colleagues found that each 1-minute reduction in exercise test duration from baseline to year 7 of follow-up was associated with a 21% increased risk in allcause mortality and a 20% increased risk of cardiovascular disease in long-term followup (26). To the best of our knowledge, only one exercise intervention study has been conducted with people who were born preterm, which was conducted in early childhood and did not use the gold standard measure of cardiopulmonary exercise testing to assess exercise capacity (27). Therefore, we performed a prespecified exploratory subgroup analysis of preterm-born young adults with elevated blood pressure in the Trial of Exercise to Prevent HypeRtension in young Adults (TEPHRA) (28,29). The primary aim of TEPHRA was to reduce blood pressure in young adults with a 16-week aerobic exercise training intervention. However, no effect on systolic or diastolic blood pressure was found, although the _ VO 2PEAK was significantly increased (29). The purpose of this subgroup analysis was to explore whether the effect of the exercise intervention on _ VO 2PEAK and _ VO 2VAT differed between young adults born at term and those born preterm. On the basis of previous evidence of cardiopulmonary limitations during exercise (22,23), we hypothesized that those who were born preterm would have impaired adaptability in _ VO 2PEAK and _ VO 2VAT in response to the exercise training intervention compared with term-born young adults.

Study Design and Participants
TEPHRA is a single-center, open-label, parallel-arm, randomized clinical trial including young adults. Fifty-two pretermand 151 term-born participants were randomly assigned with minimization according to sex, age (,24 yr, 24-29 yr, and 30-35 yr), and gestational age at birth (,32 wk, 32-37 wk, and .37 wk), in a 1:1 ratio to a control group or an aerobic training and physical activity intervention. The recruitment strategy was aimed to recruit a higher percentage of preterm-born participants than in the general population. The exercise intervention comprised a 16-week aerobic training program, selfmonitoring with a wrist-worn physical activity device, and motivational coaching. The control group was offered lifestyle educational materials. A flow diagram of participants' enrollment and allocation is presented in Figure 1.
The main objective of TEPHRA was to compare the effect of a 16-week supervised aerobic exercise training intervention with usual care/minimal intervention on ambulatory blood pressure levels in young adults with high-normal or elevated blood pressure (28,29). The primary outcome variable was a 24-hour awake ambulatory blood pressure change (systolic or diastolic) from baseline to 16 weeks (28). Secondary objectives consisted of the investigation of associations between baseline cardiovascular phenotypes, including the preterm-born phenotype, and response to exercise intervention across outcomes (28). Related measures were oxygen uptake kinetics across submaximal and peak exercise at baseline and 16-week follow-up, as described in Table 1 of the published study protocol (28). Full details of the study's (secondary) objectives, outcome measures, and procedures can be found in the study protocol. No changes were made to the study outcomes after publication of the study protocol.
Participants were recruited through open recruitment; General Practitioner records; invitations from hospital birth registers; online advertising on Facebook, Instagram, and Twitter; and invitations after

At a Glance Commentary
Scientific Knowledge on the Subject: Preterm birth is associated with long-term cardiopulmonary risk. Young adults born preterm are reported to have a lower rate of _ VO 2 at peak exercise intensity and at the ventilatory anaerobic threshold compared with their term-born peers, but little is known about their response to exercise training.
What This Study Adds to the Field: In this trial subgroup analysis of young adults with elevated blood pressure and stage 1 hypertension, a 16-week aerobic exercise training intervention led to significant improvements in peak exercise _ VO 2 and ventilatory anaerobic threshold, with no evidence of a statistically different response based on preterm birth history. participation in previous studies. Eligibility criteria comprised the following: participants were 18 to 35 years old; had a 24-hour awake ambulatory systolic and/or diastolic blood pressure higher than 115/75 mm Hg and lower than 159/99 mm Hg; had a body mass index less than 35 kg/m 2 ; were not on and had not previously been prescribed hypertension medications; had verifiable birth history of preterm birth (,37 wk) or full-term birth (>37 wk); and had the ability to access and use a computer and the Internet. Gestational age and birth weight were verified by medical records or personal child health records. Exclusion criteria comprised the following: being pregnant; having participated in structured exercise more than once per week or maintained high cardiovascular fitness; any contraindication to exercise; the inability to walk briskly on the flat for 15 minutes; and any evidence of cardiomyopathy, inherited cardiac abnormalities, or other significant cardiovascular disease. The study was conducted in the University of Oxford's Division of Cardiovascular Medicine sites at the John Radcliffe Hospital in Oxford, UK. Enrollment occurred between June 30, 2016 and October 26, 2018; the final follow-up was completed on January 9, 2020.
The trial protocol and any subsequent amendments were approved by the University of Oxford as host institution and study sponsor and by the South Central Research Ethics Committee for the National Health Service Health Research Authority (Reference no. 16-SC-0016). A trial steering committee and independent data and safety monitoring board monitored the study, and all participants provided written informed consent. The investigators ensured that the study was conducted in accordance with the principles of the Declaration of Helsinki as well as in accordance with relevant regulations and good clinical practice.

Anthropometry
Height and weight were measured to the nearest centimeter and 0.1 kg, respectively, with participants' footwear removed and light clothing worn. With a measuring tape, waist circumference was measured 2 cm above the iliac crest, and hip circumference was measured at the point of widest overall girth near the level of the greater trochanter of the femur or midbuttock.  Definition of abbreviations: BF VAT = breathing frequency at the ventilatory anaerobic threshold; BMI = body mass index; bpm = beats per minute; CI PEAK = cardiac index at peak exercise intensity; CI VAT = cardiac index at the ventilatory anaerobic threshold; CVD = cardiovascular disease; DBP = diastolic blood pressure; HDL = high-density lipoprotein; HOMA = Homeostatic Model Assessment; HR = heart rate; IQR = interquartile range; RER* = respiratory exchange ratio; SBP = systolic blood pressure; SVI PEAK = stroke volume index at peak exercise intensity; SVI VAT.est = stroke volume index at the ventilatory anaerobic threshold; VE VAT = minute ventilation at the ventilatory anaerobic threshold; _ VO 2VAT = _ V O 2 measured at the ventilatory anaerobic threshold; VT VAT = tidal volume at the ventilatory anaerobic threshold; z = z-score. Data are presented as mean (SD). Frequencies are presented as n (%). RERs* are presented as mean (SD). Bold P values are statistically significant (P , 0.05).

Spirometry and Cardiopulmonary Exercise Test
All participants underwent spirometry testing according to standard guidelines by using the Cortex Metalyzer 3B (Cortex Biophysik GmbH) (30). Before each test, the spirometer underwent calibration checks of the flow and volume with a 3.0-L calibration syringe (Futuremed). Each participant was seated and guided through the test while wearing a nose clip, with their lips tightly sealed around the mouthpiece. At least three acceptable forced expiratory maneuvers were conducted with every participant. Visual checks of the waveforms were performed to ensure consistency and data quality before the highest values were selected and recorded on a clinical research form. Parameters recorded included FEV 1 and FVC (23). FEV 1 and FVC z-scores and percent predicted values were calculated using the Global Lung Initiative online calculator (http://gli-calculator.ersnet.org/). Afterward, participants completed a peak cardiopulmonary exercise test on a seated stationary cycle ergometer (Ergoline GmbH) using a validated incremental protocol. Respiratory gases were measured breath by breath using the same Metalyzer as that used for spirometry. Continuous electrocardiogram monitoring was used to record heart rate, and blood pressure readings were taken every 4 minutes with a manual mercury sphygmomanometer (ACCOSON Freestyle). Participants maintained 60 revolutions per minute during the test, which started with one quiescent minute of resting measurements followed by 2 minutes of warm-up at 20 watts. After this, workload increased to 35 watts. To normalize test duration to 8-12 minutes, participants who reported higher activity or fitness levels had their workload increased to 75 watts after the warm-up. The workload was then increased in increments of 15 watts each minute, while participants cycled continuously until exhaustion prevented them from maintaining 50 revolutions per minute or safety termination criteria were met. The test ended with a 2-minute cool-down period at 50 watts and revolutions of the participant's own preference.

Intervention
The intervention stipulated three 60-minute aerobic training sessions (completed on bicycle ergometers) on separate days per week for 16 weeks at an exercise intensity of 60-80% of peak heart rate measured at baseline. A wrist-worn heart rate and activity monitor (Fitbit Charge HR; Fitbit, Inc.) was gifted to the participants who were encouraged to wear it daily. To track physical activity in the intervention group, activity from the wrist-worn activity monitor was tracked using the Fitabase data management platform and records kept of training sessions attended. The compliance threshold for the intervention was set at 80%, equivalent to 39 or more independent training sessions, with no more than 2 weeks between sessions. A compliant session was defined, a priori, as a supervised one. This definition was refined by the trial committees during the trial to being an aerobic session defined as a supervised gym session 40 to 60 minutes long; a self-reported aerobic session 40 to 60 minutes long; a day with a step count of 8,000 steps or more measured by their Fitbit; or a total of 40 or more Fitbit active minutes that were defined as fairly and vigorously active. After the 16 weeks of intervention, participants attended a 60-minute motivational coaching session where they reflected on their physical activity behavior as well as the intervention to set long-term physical activity goals. The intervention team was trained in motivational coaching and consisted of physiologists, physiotherapists, clinical nurse specialists, and a physician.  week aerobic training intervention in the full-study group (black), followed by the subgroup effects for term-born (green) and pretermborn (blue) young adults on the primary outcome ( _ VO 2PEAK ) and secondary outcome ( _ VO 2VAT ). Subgroup effects did not differ significantly on the basis of interaction analyses (P = 0.32 and P = 0.12, respectively). 95% CI = 95% confidence interval; _ VO 2PEAK = _ VO 2 measured at peak exercise intensity; _ VO 2VAT = _ VO 2 measured at the ventilatory anaerobic threshold.

ORIGINAL ARTICLE Statistical Analysis
Baseline subgroup cohort characteristics were tested for normality by visual inspection of histograms and normal quantile-quantile plots. Continuous variables are presented as mean (standard deviation) when the data were normally distributed and median (interquartile range) when the data were non-normally distributed.
Frequencies are presented as ns with percentages. Wilcoxon rank-sum tests were applied on the continuous data and chi-square tests for the frequencies. All analyses were conducted in R, version 3.6.1 (July 5, 2019). The main effects of the trial were analyzed by fitting general linear models with 16-week follow-up _ VO 2PEAK and _ VO 2VAT values as outcome measures, adjusting for baseline values of the outcome measure and minimization factors (age as continuous variable; sex; and gestation >37, or ,37 weeks' gestation). For instance, the regression equation for adjusted _ VO 2PEAK was as follows: Subgroups (term-and preterm-born) were defined by splitting the gestational age variable from the minimization procedure (gestation: ,32, 32-37, or >37 weeks' gestation) into two levels: >37 and ,37 weeks' gestation. The subgroup analysis was then performed by fitting additional linear models, including an interaction term between treatment allocation (exercise, control) and gestational age category (>37 or ,37 weeks' gestation) to test whether the intervention effects would vary significantly across subgroups. Hence, for adjusted _ VO 2PEAK , the regression equation was as follows: Complete case analyses were performed throughout. Model assumptions were tested and confirmed by plotting residuals versus fitted values. The results for preterm and term groups are presented as adjusted mean differences with 95% confidence intervals (CI). An a level of 0.05 was considered statistically significant. No adjustments were done for multiple testing. The cardiac index and stroke volume index (SVI) at the ventilatory anaerobic threshold and peak exercise intensity (CI VAT , SVI VAT , CI PEAK , and SVI PEAK , respectively) were estimated according to Stringer, Hansen, and Wasserman on the basis of oxygen uptake (31) and then indexed to body surface area according to the Du Bois formula.

Exercise Intervention: Main Effects
Of the participants who discontinued the study, nine were in the exercise intervention group and five were in the control group. There was no evidence of exercise intolerance being the reason for withdrawal. Cohort characteristics, including gestational age, anthropometrics, demographics, biochemistry, blood pressure, and lung function, as well as peak heart rate and peak respiratory exchange ratio, are shown in Table 1. The cohort was .90% White. Regarding the main effects, adjusted _ VO 2PEAK and _ VO 2VAT increased significantly after 16 weeks of exercise training by 2.7 ml/kg/min (95% CI, 1.6 to 3.8) and 2.9 ml/kg/min (95% CI, 1.6 to 4.3) in the intervention group compared with controls as calculated by Equation 1. The full regression models are itemized in Table 2.

Exercise Intervention: Pretermand Term-born Subgroup Interaction Effects
Adding an interaction term to the model, as per Equation 2, returned the following subgroup results. In term-born individuals in the intervention group, adjusted _ VO 2PEAK increased significantly by 3.1 ml/kg/min (95% CI, 1.7 to 4.4), compared with termborn controls, whereas there was no change in preterm-born participants (interaction effect P = 0.32; Figure 2). The adjusted _ VO 2VAT significantly increased in the intervention group compared with that in the control group for both the term-born participants (2.3 ml/kg/min; 95% CI, 0.7 to Definition of abbreviations: adj. R 2 = adjusted R 2 ; BF VAT = breathing frequency at the ventilatory anaerobic threshold; bpm = beats per minute; CI est.PEAK = cardiac index estimate at peak exercise intensity; CI est.VAT = cardiac index estimate at the ventilatory anaerobic threshold; HR VAT = heart rate at the ventilatory anaerobic threshold; VE VAT = minute ventilation at the ventilatory anaerobic threshold; VT VAT = tidal volume at the ventilatory anaerobic threshold; SVI est.PEAK = stroke volume index estimate at peak exercise intensity; SVI est.VAT = stroke volume index estimate at the ventilatory anaerobic threshold. 3.8) and the preterm-born participants (4.6 ml/kg/min; 95% CI, 2.1 to 7.0; interaction effect P = 0.12; Figure 2). There was a greater magnitude of increase in _ VO 2VAT compared with that of _ VO 2PEAK in preterm-born participants, which was not the case for the term-born participants ( Figure 2).

Post Hoc Subgroup Analysis
We conducted eight post hoc analyses to further explore cardiopulmonary adaptations. The first four were to test whether, at the ventilatory anaerobic threshold, heart rate (HR VAT ), minute ventilation (VE VAT ), breathing frequency (BF VAT ), and tidal volume (VT VAT ) would be exclusively increased in preterm-born participants. The remaining four analyses were conducted to test whether SVI VAT , SVI PEAK , CI VAT , and CI PEAK increased less in the preterm-born intervention group compared with preterm-born controls than in the term-born intervention group compared with term-born controls. The main effects for the combined study groups were analyzed as per Equation 1, with the results presented in Table 3. The subgroup effects were again analyzed by using Equation 2, with the results shown in Figure 3.
Adjusted HR VAT increased significantly by 13 beats per minute (bpm) (95% CI, 3.8 to 22.4) in the preterm-born intervention group compared with preterm-born controls, with no significant change in term-born participants (interaction effect P = 0.13; Figure 3). Adjusted VE VAT increased significantly only in the preterm-born intervention group compared with preterm-born controls by 8.9 L/min (95% CI, 3.2 to 14.6), with no significant change in term-born participants (interaction effect P = 0.06; Figure 3). Adjusted BF VAT did not change significantly in the preterm-born or term-born intervention group compared with their respective control groups. VT VAT increased significantly in term-born participants in the intervention group compared with term-born controls (2.0 dL; 95% CI, 0.4 to 0.3) whereas there was no significant change in the preterm-born participants (interaction effect P = 0.58; Figure 3).
Adjusted SVI VAT showed no significant change in the preterm-born participants, whereas it increased significantly in the term-born intervention group by 2.5 ml/m 2 (95% CI, 0.5 to 4.5) compared with term-born controls (interaction effect P = 0.56; Figure 3). Similarly, adjusted SVI PEAK showed no significant change in the preterm-born participants, whereas it increased significantly in the term-born intervention group by 2.6 ml/m 2 (95% CI, 0.7 to 4.6) compared with term-born controls (interaction effect P = 0.58; Figure 3). Adjusted CI VAT increased significantly by 8.2 dL/min/m 2 (95% CI, 2.6 to 13.9) in the preterm-born intervention group compared with preterm-born controls and increased significantly in the term-born intervention group by 6.5 dL/min/m 2 (95% CI, 3.0 to 10.1) compared with term-born controls (interaction effect P = 0.61; Figure 3). Adjusted CI PEAK showed no significant change in the preterm-born participants, whereas it increased significantly in the termborn intervention group by 6.0 dL/min/m 2 (95% CI, 2.9 to 9.2) compared with termborn controls (interaction effect P = 0.44; Figure 3).

Discussion
This was the first analysis exploring the potential benefits of an aerobic exercise training intervention in preterm-born young adults. There was no statistical evidence of a smaller increase in _ VO 2PEAK and _ VO 2VAT in preterm-born participants compared with term-born controls in response to the intervention. In addition, the increase in _ VO 2VAT in the preterm-born intervention group was twice as large as that in the termborn intervention group compared with their respective control groups. However, this difference was not statistically significant, indicating that there was no difference in cardiopulmonary adaptation between preterm-and term-born participants.
The observed greater increase in _ VO 2VAT compared with _ VO 2PEAK after the intervention in preterm-born participants meant that _ VO 2VAT occurred at a higher percentage of _ VO 2PEAK . Such higher percentages are paradoxically observed in healthy elite athletes as well as in diseased populations, including those with pulmonary hypertension and heart failure (10,32,33). The post hoc tests of HR VAT , VE VAT , BF VAT , and VT VAT were performed to test whether the _ VO 2VAT indeed occurred at a higher percentage of _ VO 2PEAK . None of them showed significant interactions, although HR VAT , VE VAT , and BF VAT were consistently higher in the preterm-born intervention group than in the preterm-born control group, which was not the case among the term-born participants and thus further suggests that _ VO 2VAT occurred at a higher percentage of _ VO 2PEAK . Changes in _ VO 2 can be due to both systemic and central physiological changes.
Evidence of systemic changes in pretermborn populations includes lower muscular fitness in preterm-born individuals (34). Furthermore, postmortem studies of infants born preterm showed lower activity and content of pyruvate dehydrogenase, respiratory chain complexes III and IV, and citrate synthase (35,36). Subsequently, in rats, Tetri and colleagues demonstrated sex-specific higher skeletal muscle fatigability, lower muscle mitochondrial oxidative capacity, more mitochondrial damage, and greater glycolytic enzyme expression (37). A recent study using the same animal model demonstrated signs of inflammation in skeletal muscles, which associated with muscle fiber atrophy, fiber type shifting from slow-to fast-twitch fibers, and impaired muscle function (38), which persisted into adulthood.
Other studies have shown central cardiovascular changes in preterm-born individuals. It has been observed that preterm-born individuals have smaller cardiac ventricular volumes (39) that are associated with functional cardiac impairments during exercise, which are predictive of their reduced exercise tolerance (8,15,22,23). In patients with heart failure with preserved ejection fraction, training-induced increases in _ VO 2PEAK are mainly due to increased oxygen extraction rather than cardiac adaptations (40). Indeed, Haykowski and colleagues showed that changes in cardiac output after 16 weeks of endurance training only explained 16% of the increase in _ VO 2PEAK , with no improvements in left ventricular systolic or diastolic function (41). In line with this, meta-analysis evidence in patients with heart failure has shown that the anaerobic threshold improves the most after exercise training (42). Given the known stroke volume impairments (22,43,44), it is plausible that _ VO 2PEAK in the preterm-born intervention group increased primarily by systemic factors. Indeed, this would explain the predominant increase in _ VO 2VAT as compared with a change in _ VO 2PEAK after the intervention, as the former depends on skeletal muscle factors and the latter depends primarily on the ability to increase cardiac output. However, the statistical nonsignificant differences in _ VO 2PEAK and _ VO 2VAT indicate that potential central and systemic differences may be small and not of clinical relevance for the majority of preterm-born individuals.
The nonsignificant 13-bpm increase in HR VAT after the intervention among preterm-born participants was surprising but comports with the notion that _ VO 2VAT occurred at a higher percentage of _ VO 2PEAK . Given that, normally, an increase in the ventilatory anaerobic threshold to a higher workload occurs reciprocally to the hallmark effect of exercise training to reduce heart rate at submaximal workloads, one would expect only minor increases in HR VAT , if at all (45,46). The increased HR VAT could mean that expected improvements in the left ventricular SVI may be smaller than for those born at term, such that preterm-born participants stayed reliant on heart rate to increase cardiac output. This observation may support the findings by Corrado and colleagues, who observed that the preterm heart appears to be heart rate dependent but is sensitive to changes in afterload (47). The authors used four-dimensional flow magnetic resonance imaging to investigate cardiac hemodynamics before and after the administration of sildenafil (afterload reduction) and metoprolol (heart rate reduction) in young adults who were born very to extremely prematurely. They showed that sildenafil resulted in an increased cardiac index, mediated by increases in heart rate and stroke volume, whereas metoprolol reduced the cardiac index by 0.37 L/min/m 2 and the heart rate by 5 bpm. However, the absence of significant interaction effects for SVI VAT and SVI PEAK does not conclusively support the idea that the preterm participants were substantially reliant on heart rate. Also, Haraldsdottir and colleagues did not observe a higher heart rate during submaximal levels of exercise in pretermborn participants compared with their term-born peers, despite an impaired stroke volume during exercise (15), but this was a relatively small sample size that may have resulted in a lack of statistical power (48). Nevertheless, the higher estimates we observed for CI VAT and HR VAT in pretermborn participants warrant further research to better understand the exercise response in people born preterm (22,23).

Limitations
Despite being prespecified, this study was an exploratory subgroup analysis and must therefore be regarded as hypothesis generating. As such, we were not able to explore all birth history factors related to preterm birth that may have impacted the response to the trial intervention. The cardiac index and SVI estimates were made on the basis of _ VO 2 and are consequently not independent from it. It allowed, however, an interpretation of _ VO 2PEAK and _ VO 2VAT with regard to cardiac function based on a strict method. Furthermore, subgroup analyses as part of the main trial naturally lack statistical power (49), raising the possibility of false negatives. For example, to detect the 13% difference in _ VO 2PEAK between preterm-and term-born groups reported from meta-analyses, a minimum of 47 participants in each group (instead of 25 vs. 23 as presented in this study) would be needed when using a twosided test at a 5% significance level and 80% power (12,48). Future, adequately powered trials should investigate the potential for negative responses to exercise training and whether the response is dependent on the degree of prematurity. The recruitment approach may also limit the generalizability of the results, given that participants on antihypertensive medication were excluded from the trial and there was a prespecified, purposeful overrepresentation of preterm-born young adults compared with the general population. Finally, although the exercise training program was chosen to replicate interventions identified by a systematic review of randomized controlled trials aiming to reduce blood pressure (28,50), different intensities or volumes of training may have had larger effects on oxygen uptake and ventilatory threshold.

Conclusions
In this exploratory hypothesis-generating subgroup analysis of young adults with elevated blood pressure and stage 1 hypertension, preterm-born individuals showed no significant differences compared with term-born individuals in their cardiopulmonary exercise adaptation after 16 weeks of aerobic exercise training. Future larger studies should investigate whether this is a true effect or perhaps due to small numbers.