NeoAPACHE Study Protocol Ii. Relationship Between Radiographic Pulmonary Area and Pulmonary Hypertension, Mortality, and Hernia Recurrence in Newborns With CDH.


 To investigate the relationship between radiographic lung area and systolic pulmonary artery pressure (sPAP) on the first day of life, mortality, and hernia recurrence during the first year of life in infants with a congenital diaphragmatic hernia (CDH). Retrospective cohort study on 77 CDH newborns. Lung area was calculated by tracing the lung’s perimeter, excluding mediastinal structures and herniated organs, on the preoperative chest X-ray performed within 24 hours after birth. Echocardiographic sPAP value, deaths, and recurrence cases were recorded. Logistic and linear regression analyses were performed. Deceased infants showed lower areas and higher sPAP values. One cm2 of rising in the total, ipsilateral, and contralateral area was associated with a 22%, 43%, and 24% reduction in mortality risk. sPAP values showed a decreasing trend at birth, with a maximum of 1.84 mmHg reduction per unitary increment in the ipsilateral area. Recurrence patients showed lower areas, with recurrence risk decreasing by 14% and 29% per unit increment of the total and ipsilateral area.In CDH patients, lung area at birth reflects impaired lung development and defect size, being associated with increased sPAP values, mortality, and recurrence risk.Clinical Trial Registration: The trial was registered at ClinicalTrials.gov with identifier NCT04396028.


Introduction
Congenital Diaphragmatic Hernia (CDH) is a severe congenital malformation with a wide outcome variability 1 . Pulmonary hypoplasia and persistent pulmonary hypertension (PH) represent the two main determinants of patients' outcome, with still high mortality and morbidity [2][3][4][5][6][7][8][9][10] . The radiographic assessment of the lung area has been proposed as an alternative method to evaluate pulmonary hypoplasia soon after birth [11][12][13] . In newborns with CDH, lung area is correlated to the functional residual capacity measured through the diluted helium technique, and its increase is associated with tidal volume improvement in the rst year of life 14,15 . The chest radiographic thoracic area (CRTA) was also lower in patients with poor prognosis and was found to predict survival to discharge from the Neonatal Intensive Care Unit (NICU) better than lung to head ratio (LHR) 16 . However, a possible association between lung area and pulmonary hypertension in de ning patients' mortality has never been investigated.
Hernia recurrence represents one of the most common complications, and a sizeable diaphragmatic defect is one of the main independent risk factors [17][18][19][20][21][22][23] . The recurrence could occur weeks, months, or even years after the primary surgery, and patients often remain asymptomatic for a long time until complications arise. Therefore, the overall risk of recurrence during the life span remains unknown 8 . To our knowledge, an association between lung area and hernia recurrence has never been reported so far.
Since lung hypoplasia and vascular development are strictly related, our hypothesis was that lower lung areas at birth could determine higher mortality and higher pulmonary pressure [23][24][25] . Moreover, provided that poor lung development comes with low lung area at birth, we supposed that lung area could indirectly re ect diaphragmatic defect size and be therefore associated with hernia recurrence 14,16 .

Results
The radiographic pulmonary area was assessed on 77 patients, 49 of whom survived to discharge and were alive at the age of one year (36.4% mortality rate) (Fig. 1). The majority of CDH were left-sided, with a high prevalence of severe forms and liver herniation. Fetal endoscopic tracheal occlusion (FETO) was performed in one-third of the cases, while extracorporeal membrane oxygenation (ECMO) was required in three patients. In more than half of cases, a diaphragmatic patch was needed for surgical repair, and in one patient, an abdominal patch was also used ( Table 1).   (Fig. 2, Panel A). They also showed lower mean total, ipsilateral and contralateral pulmonary areas at birth (total pulmonary area: 15.1 ± 6.9 cm 2 vs 8.1 ± 4.6 cm 2 , p < 0.001; ipsilateral pulmonary area: 5.1 ± 3.4 cm 2 vs 1.8 ± 2.6 cm 2 , p < 0.001; contralateral pulmonary area, 10.0 ± 4.3 cm 2 vs 6.2 ± 3.0 cm 2 , p < 0.001) (Fig. 2, Panel B).
At birth, pulmonary area and sPAP were signi cantly associated: as the three areas increased, sPAP at T0 signi cantly decreased, as shown by the linear regression model (Table 3, Panel A). Following logistic regression analysis with death as the outcome variable, the increase in all radiographic parameters was also signi cantly related to improved survival in the rst year of life (Table 3, Panel B).
Finally, with increasing sPAP at T0, the risk of death signi cantly increased as well ( Table 3, Panel C).
The Receiver Operating Characteristic (ROC) curve analysis showed that the total pulmonary area had an area under the curve (AUC) of 0.808, and a cut off of 10.87 cm 2 predicted survival with 77.6% sensitivity and 75% speci city (Fig. 3, Panel A1). The ipsilateral pulmonary area had an AUC of 0.772, and a cut off of 2.08 cm 2 predicted survival with 81.6% sensitivity and 68% speci city (Fig. 3, Panel A2). The contralateral pulmonary area had an AUC of 0.775, and a cut off of 7.3 cm 2 predicted survival with 75% sensitivity and 68% speci city (Fig. 3, Panel A3).

Radiographic pulmonary area and hernia recurrence
Survivors at the end of the rst year of life were divided into two groups based on hernia recurrence: recurrence (n = 10) and non-recurrence (n = 39) ( Table 4).  The recurrence group mainly included severe-moderate forms (80% vs. 28.2%), while most non-recurrence patients were mild (20% vs. 71.8%). Although the mean initial O/E-LHR% was not signi cantly different, the mean nal O/E-LHR% was lower in the recurrence group (44.4% vs. 56.9%, p = 0.029). Even though diaphragmatic patching was higher in the recurrence group, this difference was not signi cant. Recurrence patients required longer intensive care ( Table 4).
The ROC curve analysis showed that the total pulmonary area had an AUC of 0.759, and a cut off of 13.07 cm 2 predicted a one-year follow-up free of hernia recurrence with 71.8% sensitivity and 80% speci city (Fig. 3, Panel B1). The ipsilateral pulmonary area had an AUC of 0.790, and a cut off of 3.75 cm 2 had 74.4% sensitivity and 60% speci city (Fig. 3, Panel B2).

Comparison between FETO and non-FETO patients
We performed a comparison between patients undergoing prenatal treatment (FETO, n = 28) and those expectantly managed (non-FETO, n = 49) ( Table 5). FETO group was more severely affected, as showed by lower mean O/E-LHR% at diagnosis and a higher liver herniation rate. However, the mean O/E-LHR% before birth was not signi cantly different. Therefore, the previous differences largely disappeared. In addition, FETO had signi cantly lower initial but higher nal O/E-LHR% as compared to non-FETO. Finally, the mean total and contralateral pulmonary area were quite similar as well, while the ipsilateral area was signi cantly lower in the treated group (Table 5).

Discussion
We demonstrated an association between radiographic lung area, sPAP values, and death. Among survivors, lung area was also associated with hernia recurrence. As previously reported, our ndings suggest a possible role of radiographic measurement as an easy, non-invasive, and reproducible tool in the early prediction of mortality and morbidity among CDH patients 15 .
Our results con rmed the invariable association between pulmonary hypoplasia, PH, and death risk, as expected on a pathophysiological basis 23,25,27 .
In our cohort, lower O/E-LHR% in the deceased group indicated a more fetal severe lung impairment, which was then re ected in smaller pulmonary areas at birth. Consequently, lung area and death were inversely related: one cm 2 of rising in the ipsilateral area was associated with a 43% reduction in mortality, while variations in the total and contralateral area determined a reduction of 22% and 24%, respectively.
Large-size defects have been previously associated with worse survival and pulmonary hypertension, suggesting that low lung volumes depict the link between these two elements 6 . Similarly, in our cohort, deceased infants were characterized by persistently higher sPAP values than survivors. In particular, sPAP values at birth showed a decreasing trend by 1.84 mmHg, with each 1 cm 2 increase in the ipsilateral area.
Radiographic assessment and rst sPAP measurement, both performed within 24 hours after birth, described the lung condition in a de ning moment. Subsequently, several factors could in uence pulmonary hypertension, such as pharmacological treatments, infections, patency of the ductus arteriosus, or surgery timing. The contribution of these factors could be neither captured nor predicted by the imaging performed at birth. For these reasons, it would be of interest to match lung area assessments and sPAP values at T1, T2, and T3, to clarify if this association is still con rmed over time and de ne possible trends.
Our ndings were consistent with previous literature. A signi cantly lower CRTA was reported in newborns with CDH who died compared to survivors, and a CRTA greater than 12.99 cm 2 was found to predict survival to discharge from NICU better than O/E-LHR% at diagnosis, with 85% sensitivity and 73% speci city 16 . In our cohort, the best cut off and sensitivity were slightly lower. Even if speci city was higher, it still carried a wide margin of error in identifying patients at risk of death. Probably, the decision to consider only the aerated portions of the lungs led to an underestimation of lung surface, leading to low cut off values and discriminatory capacity. month had a higher incidence of postoperative complications and worse survival, and persistently severe PH at 1 month was associated with increased mortality 6 . Similarly, looking at our results, we could assume that the most critical factor might not be the absolute value of sPAP or the presence of PH in the rst hours after birth, rather its persistence over time 6 .
Small lung areas at birth result from poor lung development, but the interference of mechanical compression exerted by the herniated organs plays a considerable role. Therefore, after mechanical compression has been removed, the effective lung area's evaluation could greatly de ne lung hypoplasia and associated outcome. For example, Dimitriou et al. calculated the difference between the pre-and postoperative radiographic measurements, showing that postoperative improvement was higher in patients with a good outcome. They concluded that poor prognosis was correlated to low postoperative rather than low preoperative values, which was probably more related to mechanical compression than lung hypoplasia 14 . Therefore, the radiographic assessment of postoperative lung areas and the relative increase from preoperative values should be included in further analysis.
We demonstrated a signi cant association between preoperative radiographic measurements and hernia recurrence among survivors during the rst year of life.
The overall recurrence rate of 20.4% in our cohort was in line with the literature reports 17,22,28 . In particular, the recurrence rate was higher in those patients with lower nal O/E LHR%, prolonged invasive respiratory support, and need for intensive care. Similarly, Al-Iede et al. found a longer duration of mechanical ventilation and hospitalization in children with recurrence 19 . Notably, these patients showed a signi cantly lower mean total pulmonary area at birth than non-recurrence, mainly due to a signi cantly lower ipsilateral pulmonary area.
As a consequence, we respectively observed a 14% and 29% reduction in recurrence risk in our cohort per unit increment of the total and ipsilateral area. Although it showed a lower trend in the recurrence group, the contralateral area did not seem to in uence. The total radiographic area showed the best speci city in discriminating those patients at risk of recurrence, while the ipsilateral area showed better sensitivity.
Taken together, the lower ipsilateral area and O/E-LHR% re ected the presence of a large diaphragmatic defect as the cause of poor lung development, indirectly con rming defect size as the leading risk factor for hernia recurrence [17][18][19][20]22 . In other words, we speculate that recurrence patients were somehow "predisposed" to this complication since birth and could have been identi ed early in the postnatal course.
The recurrence group's high patching rate suggested the presence of a wide defect, although this difference did not reach statistical signi cance. We cannot deduce any speci c contribution of the patch in determining the recurrence risk due to the low sample size.
We observed that tracheal occlusion improved lung development and outcome through the descriptive comparison between FETO and non-FETO patients' characteristics.
Since prenatal treatment is reserved for severe cases of CDH, the FETO group included only patients at one end of the spectrum of disease severity 26 . Nevertheless, nal O/E-LHR% before birth was even better in the FETO than in the untreated group, and the total and ipsilateral lung areas at birth did not signi cantly differ. Likewise, Dassios et al. observed that patients previously submitted to FETO procedure had a CRTA comparable to untreated patients with a similar mortality rate, re ecting a lung catch-up growth favored by the prenatal procedure 16 .
However, there was an inconsistency among higher mortality rates, similar sPAP values, and prolonged need for pulmonary vasodilators in the FETO group compared to the other. At rst impression, we could argue that only a quantitative but not qualitative improvement in lung development was achieved with prenatal intervention since similar lung areas at birth seemed not to guarantee the same survival chances, apparently due to more severe pulmonary hypertension in FETO patients. However, after excluding mild cases, the two groups showed a similar outcome, suggesting that the initial ndings were due to many mild cases in the untreated group. The nal O/E LHR% dramatically improved after the procedure, but the ipsilateral area was even signi cantly better in FETO than non-FETO patients.
In our cohort, the non-FETO group, which was primarily constituted by moderate cases, showed a 41.2% in which Gore-Tex ® patch was inserted 32 . Although we cannot conclude on the patch's contribution in determining the risk of re-herniation, we can observe that lower nal O/E LHR%, lower radiographic area at birth, and higher patching all suggest a common denominator in uencing the risk of this complication 17,19,20,22 . Finally, we can speculate that lung catch-up growth in FETO patients conferred the same risk of recurrence as the untreated counterparts.
To the best of our knowledge, our study seems to be the rst to evaluate the association between radiographic lung area and two important outcomes affecting newborns with CDH: pulmonary hypertension and hernia recurrence.
The radiographic measurement is easy, rapid, and can be performed soon after birth on the chest X-ray routinely performed at NICU admission. It would contribute to the early identi cation of infants at greater risk of developing higher sPAP values in the immediate postnatal period and at a higher likelihood of long-term hernia recurrence, as well as higher mortality, with a potential impact on patients' management.
For example, the combined serial evaluation of lung area and sPAP over time could help to de ne trajectories related to the risk of persistently elevated sPAP and to identify patients evolving towards chronic pulmonary hypertension. Similarly, the preoperative radiographic assessment could help identify a subgroup of patients at higher risk of recurrence, directing them towards a more strict surgical followup.
The ipsilateral and contralateral areas were considered separately, evaluating the impact of hernia on each lung. We demonstrated that the ipsilateral area, which is more seriously affected by visceral herniation, has the most signi cant in uence on patient outcomes.
Finally, focusing on FETO patients, we con rm the positive effects of the fetal procedure on lung catch-up growth and patient outcome.
Patients from our cohort showed a broad spectrum of disease severity, including infants requiring fetal surgery and ECMO support, and the standardization of treatment in accordance with international guidelines guarantee uniformity of care.
A certain technical di culty in tracing the lungs' perimeter in severe forms must be underlined rst among our study's limitations. We arbitrarily decided to consider only those parts of the radiograms where a lung plot was present, corresponding to those regions effectively recruited and ventilated. This methodological decision could constitute a bias leading to underestimating the lung dimensions since atelectasis areas had been excluded from the measurement.
Another signi cant limitation is related to the retrospective design of the study. The sample size was limited, and some missing data regarding sPAP estimation could not be integrated with further assessments, nor the number of patients could be incremented. We recognize that several factors could in uence pulmonary vascular resistance and mortality throughout the hospital stay. The sample size should then be increased to allow a more uni ed subgroups analysis.
As research perspectives, serial assessments of the radiographic lung area during the hospital stay, remarkably the comparison between pre-and postoperative condition, could clarify the association between area, sPAP, and mortality, especially in most severe forms. We also aim to investigate the radiographic lung area's possible role in predicting the need for ECMO support through a prospective multicenter cohort study.
In conclusion, the radiographic pulmonary area on the rst day of life re ects impaired lung development during fetal life and the extent of the diaphragmatic defect in CDH patients. Lower lung areas are associated with higher sPAP values at birth, death, and hernia recurrence. Further studies are needed to consolidate these results and de ne the radiographic lung area's possible role for CDH early risk assessment, monitoring, and outcome prediction.

Methods
The present study was carried out in accordance with the principles of good clinical practice and the Helsinki Declaration, as well as the national legislative and administrative provisions in force. This study was approved by the local Ethics Committee (Milan Area 2, Italy) with approval number OSMAMI-04/05/2020-0015998-U, and due to the retrospective nature of the study the informed consent was waived by the Ethics Committee.
The study is part of a cohort study called NeoAPACHE (Assessment of the Pulmonary Area in Newborns with Congenital diaphragmatic HErnia), performed at NICU of Fondazione IRCCS Ca' Granda Ospedale Maggiore Policlinico, Milan, Italy, on CDH patients over a 6 years-period (January 2012 -December 2018).
A comprehensive description of the study design has been previously published 15 .
As a primary objective, we aimed to evaluate the relationship between radiographic pulmonary area assessed on the rst day of life and: 1. Pulmonary hypertension at birth, indirectly estimated by measuring the sPAP through tricuspid valve regurgitation gradient; The secondary objectives were: 2. Death during the rst year of life; 3. Recurrence of CDH among survivors at one year of life.
Moreover, the radiological features and outcomes of neonates submitted to FETO compared to those expectantly managed were analyzed and described.

Subjects
As previously described, we enrolled all newborns having a preoperative chest X-ray performed within 24 h after birth at our NICU. Death within 1 hour, rotated, and air leak radiographs were excluded 15 . Surgery was performed as soon as the patient was clinically stable through median laparotomy with either primary repair or Gore-Tex ® patch insertion 33 . After discharge, all patients were included in a multidisciplinary follow-up program 8 3. total pulmonary area (cm 2 ), obtained as the sum of the preceding two.

Data collection
Data regarding prenatal history, clinical, and surgical course were collected from each patient's medical records. Echocardiograms performed after birth (T0), pre-surgery (T1), post-surgery (T2), 7 days after surgery (T3) were reviewed, and reported sPAP values were recorded. CDH recurrence after surgical repair and the number of deaths in the rst year of life were considered. Data acquisition was anonymous.

Statistical analysis
Continuous variables were reported as mean (standard deviation) or median (interquartile range); categorical variables were presented as number and percentage. For the comparison between groups, Student's T-Test, Mann-Whitney U Test, or Fisher Exact Test were applied as appropriate.
The radiographic assessment's reproducibility has already been demonstrated in the primary analysis, using the Bland Altman plot and calculating the Pearson Correlation index 15 .
Logistic regression models were used to evaluate the relationship between the lung area and death or hernia recurrence risk. Linear regression models were used to assess the effects of lung area on sPAP values. The models were corrected for gestational age at birth, as this variable could independently in uence the patient's lung development and survival.
The ROC curve was also calculated to assess the radiographic measurement's discriminatory capacity, thus analyzing the test's sensitivity and speci city