This was a nationwide observational follow-up study of all individuals born in Norway at gestational age (GA) <28 weeks or birth weight (BW) <1,000 gram during 1999–2000 (15). The inclusion process, data collection, and outcome at discharge from the neonatal intensive care unit (NICU) have been described in previous reports (16). PDA surgery was performed at four different hospitals. The indication for surgery was determined at the discretion of the neonatologists responsible for neonatal care and was based on clinical signs and echocardiographic evaluation.

The present study was conducted during 2018–2020, enrolling three groups (Figure 1):

Vyntus^® PNEUMO spirometer (Vyaire Medical GmbH, Leibnizstrasse, Hoechberg, Germany) was used to perform spirometry according to guidelines (17). Forced vital capacity (FVC), forced expiratory volume in 1 second (FEV₁) and FEV₁/FVC were recorded. Raw data were transformed to z-scores using the reference equations of the Global Lung Function Initiative (18).

Peak exercise capacity was determined using a computerized incremental treadmill (Woodway PPS 55 Med, Weil am Rhein, Germany) exercise test according to a modified Bruce protocol (19) using a Vyntus CPX unit powered by SentrySuite software (Vyaire Medical GmbH, Hoechberg, Germany). Speed and elevation were increased every 90 s from an initial slow-walking phase. The test was stopped when the subject indicated severe exhaustion, preferably supported by a respiratory exchange ratio (RER) exceeding 1.05 or heart rate exceeding 95% of predicted maximal heart rate (20).

Variables of gas exchange and airflow were measured breath by breath and averaged over 10 s. The highest values for oxygen uptake determined during the last 60 s were recorded as peak values (VO₂peak). VO₂peak was reported as ml/kg/min and as the percentage of predicted using reference equations from a large sample of Norwegian subjects of relevant age (21). Exercise performance was described by the completed distance (meters) on the treadmill. The percentage inspiratory time to total time in a respiratory cycle (T_i/T_tot%) was used to describe the breathing pattern. Breathing reserve was the difference between maximal voluntary ventilation (FEV₁ x 35) and peak minute ventilation reported as the percentage of maximal voluntary ventilation.

CPET in the PDA-surgery group was performed with concomitant continuous transnasal flexible video-laryngoscopy (ENF TYPE V2, video processor CV-170, OLYMPUS, Tokyo, Japan) as described previously (22). LVCP was identified and later verified by laryngeal stroboscopy. The video recordings of the laryngeal inlet during treadmill running were assessed and rated for laryngeal obstruction according to a modified version of the classification described by Maat et al. (23). Because of laryngeal asymmetry in subjects with LVCP, a modified CLE-score (0–24 points) was developed, assessing the right and left glottic and supraglottic areas separately. The visually assessed medial rotation of the aryepiglottic folds and medialization of the vocal folds were scored ranging from normal (0 points) to maximal (3 points) at moderate (fast walking) and at maximal effort. The left and right sides were scored separately. The total modified CLE-score was the sum of the sub-scores at moderate and maximal exercise.

All participants were asked to complete an online questionnaire mapping several health issues. The PDA-surgery group filled in an additional paper-based questionnaire adapted from the Voice Handicap Index with questions regarding voice symptoms (24). The question on physical activity was adapted from the European Community Respiratory Health Survey II questionnaire (25). The questions “Do you have breathing problems beyond normal during physical exertion?” Do you make scraping sounds or other abnormal sounds from the throat during physical exertion? Is your voice hoarser than in others of the same age? and “Does your voice affect participation in singing?” were custom-made for the project.

Data were analyzed using the statistical software SPSS version 26 (IBM SPPS Statistics, NY, USA) and MedCalc version 19.5.3 (MedCalc Software Ltd, Osted, Belgium). Group comparisons were performed using the independent samples t-tests (equal variance not assumed) with 95% confidence intervals (95% CI), Mann-Whitney U-tests, or Fisher's exact tests, as appropriate. Analysis of covariance was used when the outcome for completed distance and VO₂peak was adjusted for gender and self-reported physical activity (hours of exercise per week) and to adjust for bronchopulmonary dysplasia (BPD) when comparing lung function variables between the PDA-surgery group and the EP-born control group. To examine whether the difference in VO₂peak between all EP-born and term-born controls differed by gender, an interaction term for gender and group affiliation was included. Linear regression with the modified CLE-score and gender as predictors was used to investigate whether VO₂peak was associated with the CLE-score after adjusting for gender. P-values ≤ 0.05 was characterized as statistically significant.

The Regional Committee for Medical and Health Research Ethics in Western Norway approved the study. Informed written consents were obtained from all participants, or their parents if subjects were not competent to give consent.

In the PDA-surgery group, sixteen (53%) subjects were diagnosed with LVCP. Two subjects (7%) had laryngeal stenosis in addition to LVCP, and one subject (3%) presented right-sided arytenoid prolapse with overlying left-sided arytenoid fold making vocal cord assessment during phonation difficult, and LVCP could therefore not be determined (these three subjects are referred to as other pathology and they were excluded from further analysis). Thirteen subjects (43%) had a normal laryngeal exam (no LVCP or major anatomic pathology). One subject with LVCP and all three subjects with other pathology were aware of their laryngeal pathology before entering this study, the remaining 12 were not. Within the PDA-surgery group, those with LVCP had more often received postnatal steroids compared to those with a normal larynx, whereas other neonatal characteristics were similar (Supplementary File).

Only 14% of those with LVCP compared to 69% of those without LVCP reported no voice-related symptoms (p = 0.006) (Table 3). Around 50% reported abnormal sounds from the throat and breathing problems during physical exertion, with no differences between the groups with and without LVCP. All three subjects with other pathology reported voice symptoms and breathing problems during physical exertion.

The three participants with other pathology were excluded from the analyses of lung function and exercise capacity. Within the PDA-surgery group, we did not find statistically significant differences in spirometry values between subjects with or without LVCP. However, clinically relevant differences could not be excluded given the wide the confidence intervals (Table 4).

The PDA surgery group had reduced z-FVC, z-FEV₁, and z-FEV₁/FVC, compared to the EP-born controls and the term-born controls (Table 4; Figure 2; Supplementary File). Neonatal BPD was present in 80% of the PDA-surgery group and in 37% of the EP-born controls, and BPD was associated with reduced z-FVC and z-FEV₁. Adjusting for BPD, z-FEV₁ was still significantly lower in the PDA-surgery group compared to the EP-born group with a mean (95%CI) difference of 0.89 (1.17, 1.61), p = 0.02.

All participants ran to perceived maximal exhaustion and all achieved RER above 1.05 or heart rate above 95% predicted. Within the PDA-surgery group, we did not find statistically significant differences in completed distance, VO₂peak (ml/kg/min as well as the percentage of predicted), or self-reported physical activity for subjects with and without LVCP (Table 5; Supplementary File). However, clinically relevant differences could not be excluded given the wide confidence intervals. T_i/T_tot% was higher in the participants with LVCP compared to those without LVCP, and also higher than in the EP-born controls and the term-born controls. Mean (95% CI) difference between those with LVCP vs. all the other groups combined was 2.8% (1.6, 4.1) p < 0.001.

The PDA-surgery group had similar exercise capacity and self-reported physical activity as the EP-born control group. All EP-born participants combined (PDA-surgery and EP-born controls), ran a shorter distance, had lower VO₂peak (ml/kg/min), and reported less physical activity compared to term-born controls (Table 5; Figures 2, 3; Supplementary File). Adjusted for gender, mean (95% CI) difference in completed distance and VO₂peak between all the EP-born participants combined vs. the term-born controls was 218 (114, 322) meters, p < 0.001, and 4.9 (1.8, 8.0) ml/kg/min, p = 0.002, respectively. There was no significant interaction effect between gender and group affiliation (all EP-born and term-born) on VO₂peak (p = 0.16). After additional controlling for physical activity, the completed distance on the treadmill was still shorter for EP-born participants compared to the term-born control group [mean (95% CI) difference 150 (39, 260) meters, p = 0.009]. Moreover, VO₂peak difference was slightly reduced [3.2 (−0.2, 6.7) ml/kg/min] and no longer statistically significant (p = 0.07).

In the PDA-surgery group, 27/30 participants performed a CLE test. Among these, the modified CLE-score at moderate and maximal effort could not be determined in three subjects, and the total score was derived from the sub-scores at rest and maximal effort, or at rest and moderate effort.

In the group with LVCP, all but one had a modified CLE-score >4, indicating laryngeal obstruction during exercise. In those with no LVCP, only three subjects had a modified CLE-score >4 (Figure 4), which suggests they had the specific diagnosis of EILO as all had normal larynx at rest (27). Figure 5 demonstrates the laryngeal inlet in three participants, one with a normal larynx and two with LVCP.

The group of subjects reporting breathing difficulties during exercise and those reporting abnormal sounds from the throat during exercise did not have a higher modified CLE-score than those without these symptoms (Supplementary File). Furthermore, the modified CLE-score was not associated with VO₂peak after adjusting for gender (p = 0.40).

In the PDA-surgery group in this present study, more had BPD and the lung function was poorer in adulthood, compared to the EP-born control group. There is convincing evidence that preterm-born survivors with or without BPD have an increased risk for poor adult lung function (35). The association between PDA surgery and BPD has been reported earlier (8, 36), and may be explained by more severe neonatal respiratory illness, as PDA surgery tends to be performed as “rescue therapy” in infants with already advanced lung disease and/or failed pharmacological treatment of PDA (5, 37). However, population-based observational studies have suggested that early surgical ligation is an independent risk factor for BPD (38, 39). A re-examination of the only randomized controlled trial investigating the effects of prophylactic PDA ligation vs. delayed ligation revealed a significant increase in BPD incidence in those who were ligated prophylactically (40). Animal studies support a link between PDA ligation and the development of chronic lung disease by increased expression of genes involved in pulmonary inflammation and decreased alveolar fluid clearance (41). However, these issues are incompletely understood.

In individuals with LVCP, the para-median position of a paralyzed left vocal cord would be expected to interfere with the normal exercise-induced dilation of the glottis, and thus potentially compromise airflow capacity and exercise capacity. We found that subjects with LVCP had prolonged inspiration and a tendency for a lower peak respiratory rate at peak exercise. By laryngoscopy, we observed severe laryngeal obstruction during exercise in several individuals affected with LVCP (Figure 5). However, LVCP and the modified CLE-score were not associated with VO₂peak. This finding is in line with our previous study, where no association between VO₂peak and LVCP was found in EP-born adults (12). The results from these two studies suggest that it is possible to obtain average exercise capacity despite a relatively severe laryngeal obstruction.

A number of long-term sequelae of EP birth may affect subsequent exercise capacity, such as cardiopulmonary and neuromuscular impairment, reduced skeletal muscle mass, and behavioral issues such as less participation in physical activity (2). A review of 22 studies on exercise capacity concluded that children and adults born preterm have 13% lower VO₂peak (ml/min/kg) than term-born, in line with the ~11% (-4.9 ml/kg/min) lower VO₂peak observed for all our EP-born participants combined (14). Similar to previous reports, we found that a lower amount of physical activity may be an explanatory factor for the relatively modest deficit in VO₂peak (42). It is still not determined if an increased level of physical activity will lead to improved exercise capacity in EP-born adults. Morales Mestre et al. conducted a randomized intervention study on EP-born children diagnosed with BPD and found that a structured exercise program improved exercise capacity (43). We encourage more research to be invested in this area to expand the knowledge on participation in physical activity and trainability in the EP-born population.

The strengths of this study were a population-based design with several centers responsible for the PDA surgery, and a high rate of participation. It was a limitation that only the PDA-surgery group was examined with laryngoscopy. Undiscovered LVCP or other laryngeal pathology might have been present in the EP-born control group, due to e.g., pressure from a large PDA or a large pulmonary trunk (44). Furthermore, laryngoscopy was not performed in the neonatal period and preoperative pathology or spontaneous postoperative improvement of LVCP could not be assessed. Cardiopulmonary exercise data for the PDA-surgery group were obtained from CLE-tests, which we have shown can be used interchangeably with data obtained from a regular CPET (22). Information on physical activity was self-reported and not determined by a more objective method like accelerometry or diary. Furthermore, the question on physical activity did not include aspects of mode and intensity, factors that may have affected the correlation between VO₂peak and physical activity.

The number of eligible subjects was determined by the number of EP-born infants who underwent PDA surgery in Norway during 1999–2000. The sample size was relatively small with large variation within the groups, resulting in a reduced power to detect differences in the subgroup analyses. About one-third of the eligible EP-born adults who had undergone PDA surgery were lost to follow-up (Figure 1). Recruiting young adults with a busy schedule is challenging and individuals with voice or breathing symptoms might have been more motivated to participate than individuals without such symptoms. The estimated prevalence of LVCP in this cohort lies within the range of 16/48 (33%) to 34/48 (71%) if no one or all non-participating subjects were diagnosed with LVCP. Furthermore, the study protocol requested treadmill running which might have motivated those able to and familiar with running to participate. More subjects in the participating group had a normal neonatal cerebral ultrasound compared to the non-participating group, implying a selection of subjects with less neurological sequela (Table 1).

Management of PDA in EP-born individuals is still under debate (45). Reports suggesting associations with negative post-operative outcomes have contributed to a decline in the rate of PDA surgery in the last decade (46). However, selection by indication represents a challenge and may not have been fully accounted for when reporting on outcomes (5, 47). Choice of surgical procedure may also affect outcomes. Surgical ligation has been associated with higher rates of LVCP than surgical clipping (48). Unfortunately, we did not have complete information on surgical methods in our data set. New catheter-based procedures add options for PDA closure also for infants <1,000 g (49). Irrespective of future guidelines for PDA management, a population of EP-born subjects with a history of neonatal PDA surgery already exists. Therefore, clinicians caring for EP-born children and adults should be aware of symptoms and long-term outcomes associated with PDA-surgery and LVCP to ensure proper follow-up.