In our study, pulmonary hypertension was detected in 23 patients (13.5%) according to our criteria. No differences were found regarding the presence of PH in relation to age, patient status at admission measured by BROSJOD or NT-proBNP values. Similarly, patients with PH were not more likely to have poor clinical outcomes based on the need of PPRS, PICU admission or longer LOS.
While measurable TVPJ was scant in our cohort, but 50/160 cases had an EI in systole ≥ 1.1, no patients had bowing of the septum into the LV cavity, meaning that PH, when present, was at most mild or moderate.
Contrary to the literature reports, in our cohort only 13.5% had PH. This may be due to our stricter criteria to define PH requiring to have at least 2 of the defined criteria such as TRVJ, RV AT/ETLV EI, among others [24, 29]. Most of the literature that concludes that patients with bronchiolitis have PH only measure one of these parameters. When we analyzed our cohort, 57% of patients had one criterion, similar to previously reported by others. However, these patients with just one criterion did not have a worse outcome.
The RV AT assessment has been suggested to be more sensitive than TVPJ velocity to detect early or latent pulmonary vasculopathy, but it is also known that high heart rates may reduce accuracy [17, 21, 24]. Mallery et al. described that for heart rates higher than 160 beats per minute, the AT would not reduce despite any grade of PH [25]. The ratio AT/ET takes into consideration heart rate and is more appropriate for pediatric studies [16]. In our cohort, 35% of cases had an AT/ET lower than 0.3, with no discrimination between severity and no correlation with worse echocardiographic parameters (TAPSE), higher NT proBNP values or longer LOS.
Therefore, we think that is important to use at least two criteria to better define which patients really have PH. Taking into account that only 13.5% of patients of a relatively large cohort of patients had at least two criteria for PH, it was impossible for us to create a score to give value to each of these parameters. Larger studies are needed to confirm our findings and to propose a score for PH evaluation.
Rodriguez et at. recently reported a prospective cohort study including 93 healthy infants admitted with RSV infection, with an incidence of PH of 22%. Infants with PH had a worse clinical course in terms of PICU admission, time of supplemental oxygen and longer LOS. In addition, they found that plasma NT-proBNP values were an accurate biomarker for PH (AUC 0.93) [10]. Our results were not consistent with these findings. While we agree that obtaining an echocardiogram early upon admission improves the predictive values of current clinical scores to better predict high-risk patients, our results suggest that PH might not be the reason for worse clinical outcomes or higher NT-proBNP values.
Previous studies focused on the assessment of cardiac function in patients with bronchiolitis have concluded that PH leads to RV dysfunction secondary to increased afterload and hypothesized that observed LV dysfunction is due to interventricular interactions [15, 19]. In our cohort, patients requiring PPRS had increased NT-proBNP values and lower TAPSE compared to those with low respiratory support. In agreement with Thorburn et al., we did not find any association between PH and parameters of worse cardiac function [11].
These findings reinforce the hypothesis that direct myocardial damage by RSV is the most likely cause for the cardiovascular manifestations observed, possibly explained by myocardial injury secondary to hypoxia and release of inflammatory mediators [1, 11, 33–36].
Limitations
This study was limited by its nature as a single center study. Although the study was prospective in the way it was conceived, the analysis of PH was off-line, and the measurements were already recorded. Furthermore, we could not assess TVRJ in most of the cohort, likely because the majority of patients were not sedated, and a long and comprehensive echocardiogram was poorly tolerated. Moreover, NT-proBNP values were only available in 26% of the cohort.