Journal of Pediatric Surgery 49 (2014) 853–858
Contents lists available at ScienceDirect
Journal of Pediatric Surgery journal homepage: www.elsevier.com/locate/jpedsurg
AAP Papers
Fetal MRI lung volumes are predictive of perinatal outcomes in fetuses with congenital lung masses Irving J. Zamora a, b, Fariha Sheikh a, b, Christopher I. Cassady a, c, Oluyinka O. Olutoye a, b, e, Amy R. Mehollin-Ray a, c, Rodrigo Ruano a, d, Timothy C. Lee a, b, Stephen E. Welty a, d, Michael A. Belfort a, e, Cecilia G. Ethun a, b, Michael E. Kim a, b, Darrell L. Cass a, b, e,⁎ a
Texas Children's Fetal Center, Texas Children's Hospital, Houston, TX The Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, TX c Department of Radiology, Baylor College of Medicine, Houston, TX d Department of Pediatrics, Division of Neonatology, Baylor College of Medicine, Houston, TX e Department of Obstetrics and Gynecology, Baylor College of Medicine, Houston, TX b
a r t i c l e
i n f o
Article history: Received 10 January 2014 Accepted 27 January 2014 Key words: Congenital lung malformation Risk stratification CCAM CPAM Fetal MRI Fetal lung volumes Perinatal outcomes
a b s t r a c t Purpose: The purpose of this study was to evaluate fetal magnetic resonance imaging (MRI) as a modality for predicting perinatal outcomes and lung-related morbidity in fetuses with congenital lung masses (CLM). Methods: The records of all patients treated for CLM from 2002 to 2012 were reviewed retrospectively. Fetal MRI-derived lung mass volume ratio (LMVR), observed/expected normal fetal lung volume (O/E-NFLV), and lesion-to-lung volume ratio (LLV) were calculated. Multivariate regression and receiver operating characteristic analyses were applied to determine the predictive accuracy of prenatal imaging. Results: Of 128 fetuses with CLM, 93% (n = 118) survived. MRI data were available for 113 fetuses. In early gestation (b 26 weeks), MRI measurements of LMVR and LLV correlated with risk of fetal hydrops, mortality, and/or need for fetal intervention. In later gestation (N26 weeks), LMVR, LLV, and O/E-NFLV correlated with neonatal respiratory distress, intubation, NICU admission and need for neonatal surgery. On multivariate regression, LMVR was the strongest predictor for development of fetal hydrops (OR: 6.97, 1.58–30.84; p = 0.01) and neonatal respiratory distress (OR: 12.38, 3.52–43.61; p ≤ 0.001). An LMVR N 2.0 predicted worse perinatal outcome with 83% sensitivity and 99% specificity (AUC = 0.94; p b 0.001). Conclusion: Fetal MRI volumetric measurements of lung masses and residual normal lung are predictive of perinatal outcomes in fetuses with CLM. These data may assist in perinatal risk stratification, counseling, and resource utilization. © 2014 Elsevier Inc. All rights reserved.
Congenital lung masses are rare and encompass a wide spectrum of malformations of the developing fetal lung. Large fetal lung masses compress surrounding thoracic structures. Compression from the mass on normal lung architecture increases the likelihood for pulmonary hypoplasia and respiratory failure following birth [1]. Compression from these lesions on the mediastinum and heart may decrease cardiac venous return and pose a risk for fetal hydrops and fetal demise [2,3]. To date, ultrasound measurement of the CCAMvolume ratio (CVR) has been the primary predictor of prenatal and postnatal outcomes in fetuses with CLM [4,5]. Fetal MRI has been established as an adjunct to ultrasound in the imaging of fetuses with congenital lung malformations (CLM) as early as 18 weeks gestation [6,7]. Compared to ultrasonography, fetal MRI may be superior at characterizing the boundaries of the malformed lung and how it relates to the normal lung lobar anatomy and surrounding thoracic
structures [8–10]. Recently, it has been shown that fetal MRI may be superior in differentiating between the various types of CLMs, including bronchopulmonary sequestration, congenital lobar emphysema, congenital pulmonary airway malformations (CPAM), and bronchogenic cysts [11]. The utility of fetal MRI in assessing volume measurements of CLM both in terms of the mass itself, as well as the uninvolved, normal lung, has not been previously examined. Therefore, the purpose of this study was to evaluate fetal MRI as a modality for predicting perinatal outcomes, including the development of fetal hydrops, mortality, neonatal respiratory distress and lung-related morbidities in fetuses with CLM. 1. Materials and methods 1.1. Study cohort
⁎ Corresponding author at: Texas Children's Hospital, 6701 Fannin St. Suite 1210, Houston, TX 77030. Tel.: + 1 832 822 3135; fax: + 1 832 825 3141. E-mail address: [email protected] (D.L. Cass). http://dx.doi.org/10.1016/j.jpedsurg.2014.01.012 0022-3468/© 2014 Elsevier Inc. All rights reserved.
Permission to conduct this study was obtained from the institutional review board (H-29695) of Baylor College of Medicine,
854
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858
Houston, TX. The records of all patients referred to the Texas Children’s Fetal Center with a congenital lung malformation over a 11-year period from January 2002 to December 2012 were reviewed retrospectively. Fetuses that died owing to non–lung-related comorbidities and those without at least one available fetal MRI were excluded. Data collected from the charts included patient demographics, gestational age at diagnosis and at birth, mortality, fetal course including development of fetal hydrops (defined as presence of serous fluid in two or more body cavities) and need for fetal intervention, and postnatal course including the presence of neonatal respiratory distress (defined as any supplemental oxygen requirement beyond the initial resuscitation period), need for intubation, duration of intubation, NICU admission, need for urgent operation and hospital length of stay. All fetal MRI images were reviewed and lung mass volume ratio (LMVR), observed/expected normal fetal lung volumes (O/E-NFLV), and lesion-to-lung volume ratio (LLV) were calculated.
normal total fetal lung volume (ONTFLV), the volume of normal lung was measured, excluding the volume of the lung mass. This ONTFLV was used to calculate the O/E-NFLV by dividing the measured value by the mean expected TFLV for the respective gestational age of each fetus [13]. The lesion/normal total lung volume ratio was then calculated by dividing the measured lung lesion volume by the ONTFLV, and labeled the “lesion-to-lung volume ratio.” 1.3. Indications for fetal intervention As previously described [5], the primary indication for fetal intervention at our center was the presence of fetal hydrops, defined as serous fluid in two or more body cavities. Fetuses with hydrops and evidence of heart failure on fetal echocardiography were considered for open maternal-fetal surgery. Fetuses that had large lung masses causing persistent mediastinal compression (PMC) late in gestation were offered an ex-utero intrapartum treatment (EXIT) to lung resection approach [14,15].
1.2. Total fetal lung volume and lung mass volume ratio measurements 1.4. Statistical analysis A fetal MRI scan was performed at the time of the initial consultation for most patients referred for a CLM. For most patients this scan was performed before 26 weeks gestation. A subset of patients had a second fetal MRI performed later in gestation (N26 weeks). All lung volumes were calculated on contiguous axial single shot fast spin echo (SSFSE) T2-weighted MR images free from motion by using a freehand region of interest (ROI) tool. The CLM was defined by its hyperintense signal on T2-weighted images and the lesion area was outlined on contiguous sections (Fig. 1). The sum of all measurements was multiplied by the section thickness to obtain a volume in cubic centimeters, which was then divided by the head circumference in centimeters to derive the lung mass volume ratio (LMVR) corrected for gestational age [12]. The total fetal lung volume (TFLV) was measured using the same method. To obtain the observed
Outcomes between groups for categorical variables were compared using Chi-square analysis and Fisher's Exact test. Continuous variables were compared using 2-tailed Student’s t-tests for normally distributed data and Mann-Whitney U test for nonnormally distributed data. An alpha of b0.05 was considered statistically significant. The outcomes examined in this study were stratified according to the gestational age at the time of fetal MRI. For fetal MRIs at b 26 weeks, the primary outcome was the development of fetal hydrops, and secondary outcomes were need for fetal intervention and mortality. For fetal MRIs obtained N26 weeks, the primary outcome was the development of neonatal respiratory distress, and secondary outcomes were need for intubation, NICU admission, and urgent surgery in the neonatal period. Mortality was defined as survival status at the time of chart review. A stepwise multivariate logistic regression was performed for the prenatal categorical outcome of ‘development of hydrops.’ The covariates examined in the regression model were gestational age at diagnosis, gender, location of the CLM and prenatal imaging parameters obtained at b26 weeks gestational age (GA) including LMVR and O/E-NFLV. A separate stepwise regression model was performed for the postnatal categorical outcome of ‘neonatal respiratory distress.’ The covariates examined in the regression model were gestational age at diagnosis, gestational age at birth, gender, location of the CLM and prenatal imaging parameters including LMVR and O/E-NFLV. Interactions were evaluated between all the covariates in the regression models. Receiver operating characteristic curves (ROC) were utilized to determine the predictive accuracy of prenatal imaging for perinatal outcomes. The O/E-NFLV and LMVR measurements obtained at b 26 weeks were evaluated with ROC analysis against the outcome of ‘development of hydrops’ and the O/E-NFLV and LMVR obtained at N 26 weeks were evaluated against the outcome of ‘neonatal respiratory distress.’ Optimal cutoff points of greatest accuracy were derived by graphing the results of the ROC analyses and selecting the point of maximum sensitivity and specificity of each analysis. 2. Results
Fig. 1. A representative image of an axial single shot fast spin echo (SSFSE) T2-weighted MR image of a fetus with CLM. Using a freehand region of interest (ROI) tool the lung lesion and normal fetal lung areas were outlined. The lung lesion was identified as the region of most hyperintense signal in the chest.
During the study period a total of 128 fetuses (65 male) were evaluated for CLM, 93% (118) survived, 63% (81) were left sided, and the average gestational age at diagnosis was 22.1 ± 4.2 weeks. Two fetuses died from non–lung mass related causes and were excluded from further analysis – one from complications owing to Trisomy 18 and one from complications owing to congenital heart disease. MRI data were available for a total of 113 fetuses. Of these fetuses, 57
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858
(50.4%) were male, 66 (58.4%) had left-sided lesions, 46 (40.7%) had right-sided lesions and 1 (0.9%) fetus had a bilateral lesion (Table 1). The mean GA at first MRI was 25.8 ± 4.3 weeks. Seventy-five fetuses had an MRI performed b 26 weeks GA (mean 23.2 ± 1.8 weeks), and 70 had an MRI performed N26 week GA (mean 32.7 ± 3.3 weeks). Thirty-eight fetuses (with high-risk lesions) had two MRIs. Overall, fetuses had relatively large normal lung volumes with a mean O/ENFLV of 87.35%; 43.4% of patients had an O/E-NFLV less than 80%, and only 15% of patients had an O/E-NFLV less than 50%. The mean LMVR was 23.1 cc and the mean LLV ratio was 1.3 (Table 1). As a group fetuses with CLM had a generally benign disease course. There was a low incidence of mortality (5.3%) and postnatal pulmonary morbidity. Pulmonary status at 30 DOL was available for 99 patients: the majority (95%) of these patients were on room air, one was on nasal cannula, one was on CPAP and three remained on mechanical ventilation. A small percentage (3.5%) of patients developed pulmonary hypertension: two of these required treatment with nitric oxide and only one needed sildenafil therapy. A minority (38.1%) of patients required NICU admission. Sixty-five percent of patients underwent surgical resection of the lung mass; most of these patients were readmitted for elective operation at a median age of 98 days. A total of 14 patients (with fetal MRIs) underwent fetal surgery for their CLM: three had open maternal-fetal surgery and 11 underwent EXIT-to-resection (Table 2). 2.1. Early gestation lung volumes and prenatal outcomes 2.1.1. O/E-NFLV as a predictor of prenatal outcomes In early gestation (b 26 weeks GA), MRI measurements of O/ENFLV, LMVR, and LLV were evaluated against prenatal outcomes. The O/E-NFLV was inversely associated with mortality. The five fetuses that died had significantly smaller lung volumes as compared to survivors (O/E-NFLV, 56.8 vs. 88.5%; p = 0.03). However, the O/ENFLV measurements were not markedly different between those patients who developed hydrops or who required fetal intervention as compared to those with an uneventful prenatal course (Table 3). To further examine the association between O/E-NFLV and the development of prenatal symptoms, we performed ROC curve analyses, which demonstrated a poor association between O/E-NFLV and predicting the development of fetal hydrops (AUC = 0.601; p = 0.412) and need for fetal intervention (AUC = 0.54; p = 0.69). 2.1.2. LMVR and LLV as predictors of prenatal outcomes The LMVR and LLV correlated strongly with the risk of developing hydrops, fetal demise and need for fetal intervention. The five fetuses that died had significantly larger lung lesions as measured by LMVR
Table 1 Demographics and prenatal imaging characteristics of all patients with CLM. Total patients Patients with prenatal MRI imaging Males, n (%) Left side, n (%) Right side, n (%) Bilateral, n (%) GA at birth (wk), mean ± SD GA at 1st MRI (wk), mean ± SD Mean CVR, mean ± SD Mean Max CVR, mean ± SD Mean O/E-NFLV (%), mean ± SD # with O/E-NFLV b80%, n (%) # with O/E-NFLV b50%, n (%) Mean LMVR, mean ± SD Mean LLV, mean ± SD
128 113 57 (50.4) 66 (58.4) 46 (40.7) 1 (0.9) 37.7 ± 3.6 25.8 ± 4.3 1.14 ± 1.23 1.25 ± 1.3 87.35 ± 31.8 49 (43.4) 17 (15) 1.10 ± 1.1 1.3 ± 2.9
CLM – congenital lung malformation; GA – gestational age; LMVR – lung mass volume ratio; LLV – lesion-to-lung volume ratio; O/E-NFLV – observed-to-expected normal fetal lung volume.
855
Table 2 Postnatal outcomes of all patients with CLM. Mortality, n (%) LOS, days median (range) Intubated, n (%) Duration of intubation, days median (range) Oxygen requirement at 30 DOL, n = 99 Room air, n (%) Nasal cannula, n (%) CPAP, n (%) Vent, n (%) PAH, n (%) Nitric Oxide Tx., n (%) Sildenafil Tx., n (%) NICU admission, n (%) NICU LOS, median days (range) Surgery, n (%) Readmitted for surgery, n (%) Age at surgery, median days (range) Fetal surgery, n = 13 OMFS EXIT Duration of postop stay, median days (range)
6 (5.3) 5 (0–215) 21 (18.6) 4 (0–42) 94 (95) 1 (0.9) 1 (0.9) 3 (2.7) 4 (3.5) 2 (1.8) 1 (0.9) 43 (38.1) 6.5 (1–215) 73 (64.6) 53 (46.9) 98 (0–2118) 3 (2.7) 11 (9.7) 4 (0–272)
CLM – congenital lung malformation; LOS – length of stay; DOL – day of life; CPAP – continuous positive airway pressure; PAH – pulmonary arterial hypertension; NICU – neonatal intensive care unit; OMFS – open maternal fetal surgery; EXIT – ex-utero intrapartum treatment.
(3.75 vs. 0.73; p b 0.001) and LLV (10.5 vs. 0.8; p b 0.001) than survivors. Fetuses that developed hydrops and who required fetal intervention also had significantly larger lung lesions as measured by both LMVR and LLV (Table 3). To determine which variables were independent predictors of prenatal outcomes we performed a multivariate logistic regression evaluating the covariates of O/E-NFLV, LMVR, GA at first MRI, rightsided lesion and gender. This analysis demonstrated that LMVR was the strongest predictor for the development of fetal hydrops (OR = 6.97, 95%CI: 1.58–30.84; p = 0.01), such that for every 1 unit increase in LMVR there was a 7-fold increase in the likelihood of developing hydrops (Table 4). To further examine the strength of the association between LMVR and the development of fetal hydrops, we performed a ROC curve analysis, which also demonstrated a strong correlation (AUC = 0.94; p = 0.001). Based on the analysis of the sensitivity and specificity we determined the optimal cutoff point was an LMVR N 2.0, which predicted worse perinatal outcome with 83% sensitivity and 99% specificity (Fig. 2). Overall, 107 (94.6%) out of 113 fetuses survived. Of the six that died, four were in the group with initial LMVR N 2.0. There were 19 fetuses (16.8%) with an LMVR N 2.0. These fetuses were born approximately 2 weeks earlier and showed a male and right-side preponderance compared to fetuses with smaller lung lesions (Table 6). Fetuses with LMVR N2.0 had a higher incidence of polyhydramnios (32%), hydrops (42%) and heart failure (32%), and 12 (63%) fetuses underwent fetal intervention (Table 7). Three of these patients underwent open fetal surgery: one had successful resection of a large left extralobar sequestration; one had a right pneumonectomy for mainstem bronchial atresia but died shortly after preterm delivery 4 weeks later at 24 weeks; and a third died intraoperatively during open fetal resection of the right lower lobe for a microcystic CCAM. Seven patients in this group underwent successful EXIT-to-resection procedures and one had a fetal thoracentesis that resulted in early cesarean delivery followed by immediate resection of a large left extralobar sequestration. All of these seven patients did well postoperatively. There were 94 fetuses (83.2%) with an LMVR b2.0. The majority of these lesions (n = 59, 63.4%) were located on the left side. In this group prenatal features of fetal compromise were rare – polyhydramnios (4.3%), hydrops (2.1%), heart failure (2.1%) – however, four patients required fetal intervention for persistent mediastinal
856
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858
Table 3 Fetal MRI features and perinatal outcomes for fetuses with CLM. Variable
O/E-NFLV (mean ± SD)
Outcomes based on F-MRI at b26 weeks Mortality Yes(n = 5) 56.8 ± 45.6 No(n = 70) 88.5 ± 30.8 Hydrops Yes(n = 6) 71.9 ± 41.9 No(n = 69) 87.7 ± 31.7 Fetal intervention Yes(n = 9) 79.9 ± 39.2 No(n = 65) 87.1 ± 32.1 Outcomes based on F-MRI at N26 weeks Neonatal respiratory distress Yes(n = 26) 66.9 ± 35.2 No(n = 41) 98 ± 35.7 Intubation Yes(n = 20) 66.8 ± 29.6 No(n = 48) 92.6 ± 37 NICU admission Yes(n = 32) 69.5 ± 34.8 No(n = 21) 92.9 ± 39.2 Surgery in neonatal period Yes(n = 23) 68.7 ± 36 No(n = 32) 91.3 ± 39.9
p-value
LMVR (mean ± SD)
p-value
LLV (mean ± SD)
p-value
0.03
3.75 ± 2.25 0.73 ± 0.57
b0.001
10.5 ± 9.9 0.8 ± 0.8
b0.001
0.26
3.19 ± 2.01 0.73 ± 0.69
b0.001
7.6 ± 9.6 0.9 ± 1.5
b0.001
0.55
1.88 ± 1.12 0.81 ± 1.02
0.004
4.2 ± 7.8 1.1 ± 2.2
0.012
0.001
2.11 ± 1.03 0.49 ± 0.49
b0.001
1.86 ± 2.3 0.27 ± 0.3
b0.001
0.007
2.14 ± 0.84 0.69 ± 0.85
b0.001
1.64 ± 1.3 0.56 ± 1.6
0.011
0.027
1.75 ± 1.13 0.61 ± 0.73
b0.001
1.54 ± 2.2 0.31 ± 0.45
0.013
0.03
2.16 ± 1.08 0.58 ± 0.61
b0.001
1.96 ± 2.4 0.32 ± 0.44
b0.001
F-MRI; fetal magnetic resonance imaging; CLM – congenital lung malformation; SD – standard deviation; LMVR – lung mass volume ratio; LLV – lesion-to-lung volume ratio; O/E-NFLV – observed-to-expected normal fetal lung volume; NICU – neonatal intensive care unit.
compression (Table 7). Four of these patients with lesions that remained large late in gestation underwent successful EXIT-toresection, and all did well postoperatively. One patient treated with fetal thoracoamnionic shunt placement and serial thoracentesis had fetal demise at 25 weeks and one with a borderline LMVR of 1.9 died of a cardiac arrest on DOL 15, twelve days after resection of the lung mass. These were the only 2 deaths in the group with an LMVR b 2.0.
2.2. Late gestation lung volumes and postnatal outcomes 2.2.1. O/E-NFLV as a predictor of postnatal outcomes In late gestation (N26 weeks GA), MRI measurements of O/E-NFLV, LMVR and LLV were evaluated against postnatal outcomes. The O/ENFLV was highly associated with all postnatal outcomes evaluated including development of neonatal respiratory distress, need for intubation, NICU admission and need for urgent surgery in the neonatal period (Table 3). To further examine the strength of the association between O/E-NFLV and the development of postnatal symptoms, we performed ROC curve analysis, which demonstrated moderate correlation between O/E-NFLV and the development of neonatal respiratory distress (AUC = 0.724; p = 0.005). Based on the analysis of the sensitivity and specificity we determined the optimal cutoff point was an O/E-NFLV b75%, which predicted a worse postnatal course with 57% sensitivity and 80% specificity.
2.2.2. LMVR and LLV as predictors of postnatal outcomes The LMVR and LLV both correlated strongly with the development of neonatal respiratory distress, need for intubation, NICU admission and urgent surgery in the neonatal period. Fetuses that had a compromised neonatal course had significantly larger lung lesions as compared to those that were asymptomatic in the neonatal period (Table 3). To determine which variables were independent predictors of postnatal outcomes, we performed a multivariate logistic regression evaluating the covariates of O/E-NFLV, LMVR, GA at birth, right-sided lesion and gender. This analysis demonstrated that LMVR remained the strongest predictor for the development of neonatal respiratory
Table 4 Multivariable logistic regression for development of fetal hydrops. Covariate
OR
95% CI
p
LMVR O/E-NFLV Right-sided lesion Gender (male) GA at 1st MRI
6.97 5.11 0.27 0.24 0.81
1.58-30.84 0.14-192.02 0.01-9.69 0.01-4.65 0.48-2.59
0.01 0.38 0.47 0.34 0.81
LMVR – lung mass volume ratio; GA – gestational age; LLV – lesion-to-lung volume ratio; O/E-NFLV – observed-to-expected normal fetal lung volume; OR – odds ratio; CI – confidence interval; MRI – magnetic resonance imaging.
Fig. 2. Receiver operating characteristic (ROC) curve of MRI derived lung mass volume ratio (LMVR) at b26 weeks for the outcome of development of fetal hydrops. The point of greatest accuracy corresponded to an LMVR N2.0.
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858 Table 5 Multivariable logistic regression for neonatal respiratory distress.
857
Table 7 Perinatal outcomes of fetuses stratified by LMVR.
Covariate
OR
95% CI
p
Variable
LMVR O/E-NFLV Right-sided lesion Gender (male) GA at Birth
12.38 1.06 3.03 0.75 0.81
3.52-43.61 0.98-1.03 0.51-18.01 0.15-3.82 0.57-1.15
b0.001 0.62 0.22 0.73 0.81
Polyhydramnios, n (%) Hydrops, n (%) Heart failure, n (%) Fetal intervention, n (%)
LMVR – lung mass volume ratio; GA – gestational age; LLV – lesion-to-lung volume ratio; O/E-TFLV – observed-to-expected total fetal lung volume; OR – odds ratio; CI – confidence interval; MRI – magnetic resonance imaging.
distress (OR = 12.38, 95%CI: 3.52–43.61; p b 0.001), such that for every 1 unit increase in LMVR there was an 12-fold increase in the likelihood of being symptomatic at birth (Table 5). To further examine the strength of the association between LMVR and the development of postnatal symptoms, we performed an ROC curve analysis, which also demonstrated a strong correlation (AUC = 0.91; p b 0.001). Based on the analysis of the sensitivity and specificity we determined the optimal cutoff point was an LMVR N1.3, which predicted a worse postnatal course with 86% sensitivity and 98% specificity (Fig. 3). In the 27 fetuses with an LMVR N1.3 during later gestation, indicators of a more complicated postnatal course were more prevalent – neonatal respiratory distress (88.5%), requiring intubation (69%) and NICU admission (92%). Patients in this group also had significantly longer initial hospital lengths of stay and required operative resection of the CLM at a younger age, most in the immediate postnatal period in their first day of life, and also had a longer postoperative length of stay (Table 7). This is in contrast to the 43 patients with an LMVR b 1.3, who had relatively mild postnatal courses – neonatal respiratory distress (7.3%), requiring intubation (4.8%), and NICU admission (34.5%). The majority (90%) of these patients were discharged home and returned later for elective resection of the lung lesion, which was typically followed by a short and uncomplicated postoperative course (Table 7). 3. Discussion
LMVR b2 n = 94 4 2 2 4
(4.3) (2.1) (2.1) (4.3)
LMVR b1.3 n = 43
LMVR N2 n = 19 6 (31.6) 8 (42.1) 6 (31.6) 12 (63.2)
p 0.002 b0.001 b0.001 b0.001
LMVR N1.3 n = 27
Neonatal respiratory distress, n (%) 3 (7.3) 23 (88.5) Required intubation, n (%) 2 (4.8) 18 (69.2) NICU admission, n (%) 10 (34.5) 22 (91.7) Length of stay, median days (range) 3 (0–39) 14 (1–69) Duration of intubation, median days (range) 2 (1–3) 4 (1–30) Duration of NICU stay, median days (range) 4 (1–32) 15 (7–63) Readmitted for elective surgery, n (%) 27 (90) 5 (20) Age at surgery, median days (range) 117 (32–576) 0 (0–539) Postoperative length of stay, median days (range) 3 (1–7) 10.5 (3–69)
b0.001 b0.001 0.019 0.005 0.168 0.107 b0.001 b0.001 b0.001
LMVR – lung mass volume ratio; CLMV – congenital lung malformation volume; NICU – neonatal intensive care unit.
was found to be most significant. Fetuses with larger lung volumes, O/E-NFLV of 75% or greater, predictably did well in general. However, patients with O/E-NFLV less than 75% had a higher incidence of neonatal respiratory distress, intubation, NICU admission, and urgent surgical resection in the immediate postnatal period. These patients had a greater degree of pulmonary hypoplasia with approximately 30% smaller lung volumes than asymptomatic patients with normal O/E-NFLV. In general it is rare for CLM patients to have severe pulmonary hypoplasia, unlike in patients with other conditions such as CDH, which typically have O/E-TFLV measurements ranging from b25% to 45% [16–19] and giant omphaloceles that have approximately 50% O/E-TFLV [20]. When we compared the three fetal MRI parameters, O/E-NFLV, LMVR and LLV, we found that the strongest predictor of outcomes throughout gestation was the LMVR, which correlated highly with prenatal outcomes when measured b26 weeks gestation and with postnatal outcomes when measured N26 weeks gestation. The accuracy of LMVR in predicting a favorable perinatal course was
Fetal MRI derived volumetric measurements of O/E-NFLV, LMVR and LLV accurately predict perinatal outcomes in fetuses with congenital lung malformations. When obtained before 26 weeks’ gestation, fetal MRI-based measurement of O/E-NFLV generally showed poor utility in predicting which fetuses may progress to hydrops and require fetal intervention, but was associated with mortality. However, when obtained later in gestation (N26 weeks) the predictive ability of O/E-NFLV for postnatal outcomes improved significantly. Overall, patients with congenital lung malformation have good prenatal and postnatal outcomes. In particular these patients have a very low risk of experiencing pulmonary symptoms in the neonatal period, likely owing to having relatively high mean O/E-NFLV (87%) as a group. In CLM patients with smaller residual lung volumes, lower measurements of O/E-NFLV correlated strongly with a more complicated postnatal course. A threshold value of 75% for O/E-NFLV Table 6 Demographic characteristics of patients stratified by LMVR. Variable
LMVR b2 n = 94
LMVR N2 n = 19
p
GA at diagnosis, mean ± SD GA at birth, mean ± SD Male, n (%) Right-sided lesion, n (%) Mortality, n (%)
25.4 ± 4.1 38.1 ± 3.3 44 (47.3) 34 (36.6) 2 (2.1)
27.5 ± 5.1 35.8 ± 4.5 13 (72.2) 12 (63.2) 4 (21.1)
0.061 0.013 0.045 0.032 0.007
LMVR – lung mass volume ratio; GA – gestational age; SD – standard deviation.
Fig. 3. Receiver operating characteristic (ROC) curve of MRI derived lung mass volume ratio (LMVR) at N26 weeks for the outcome of neonatal respiratory distress. The point of greatest accuracy corresponded to an LMVR N1.3.
858
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858
most significant at a threshold of b2.0. In our series 97.9% of fetuses with an LMVR b2.0 did not develop hydrops or heart failure. This is a particularly important outcome since it is well known that fetuses with isolated CLM have a 100% survival when hydrops is absent, but when hydrops is present, the risk of fetal demise increases significantly unless the fetus is treated in some way [21]. An LMVR of b2.0 was also highly predictive of survival since only two deaths occurred in this group, resulting in a 97.9% survival rate. In contrast, fetuses with an LMVR greater than 2.0 experienced poorer outcomes with only a 78.9% survival rate. In our series 80% of the cases that developed hydrops were in this group with larger lung lesions. LMVR obtained after 26 weeks was also useful in predicting postnatal outcomes. Although the majority of the CLM patients had a relatively benign postnatal course, all four of the cases that developed pulmonary hypertension had LMVR N 1.3. Fetuses with an LMVR of 1.3 or greater also required more postnatal care as evident by a higher incidence of neonatal respiratory distress, need for intubation and NICU admission and had an overall longer postoperative and total hospital length of stay. The majority (80%) of fetuses with an LMVR N 1.3 required urgent operation for their lung mass in the neonatal period, which is in contrast to only 10% of fetuses with LMVR b 1.3 who had early surgical intervention. Although the intent of this study was to establish the utility and accuracy of fetal MRI in predicting perinatal outcomes in patients with CLM, we did perform cursory comparative analyses to the current method of ultrasound based CCAM-volume ratio (CVR). On ROC curve analyses we found that LMVR (AUC = 0.91) was slightly stronger at predicting neonatal respiratory distress than CVR (AUC = 0.89). We also noted that CVR tended to overestimate the lung mass volumes. A dedicated comparative study would be required to determine which of these two studies is more useful at predicting perinatal outcomes. Our group has previously helped refine the use of CVR in the management of fetuses with CLM. We recognize that ultrasoundbased CVR is already a highly accurate and less expensive means of risk-stratifying fetuses with CLM and we continue to advocate its use. It is reasonable to hypothesize that fetal MRI may be most useful for those fetuses with larger and higher-risk lesions in order to more precisely assess the lobar anatomy of the mass, relationship of the mass to the surrounding thoracic structures, and the degree of pulmonary hypoplasia which may enhance prenatal decision making and perinatal management. This study is limited by the inherent challenges of a retrospective review. Additionally, the results must be examined with an understanding that these are institution-specific findings and the utility of this study will depend on the reproducibility of these measurements across other institutions. To minimize the variability and to increase the level of consistency in the generation of fetal MRI derived measurements, we use a standardized protocol for obtaining the fetal MRI and a defined method for calculating the volumetric measurements. In this study we propose a novel utility for fetal MRI based measurements in fetuses with congenital lung malformations. We concluded that fetal MRI volumetric measurements of lung malformations utilizing the LMVR and residual normal lung corrected for gestational age (O/E-NFLV) are predictive of perinatal outcomes in fetuses with CLM. These imaging parameters could be useful for risk
stratification and to direct the perinatal management and counseling in cases of prenatally diagnosed lung masses. References [1] Cass DL, Olutoye OO, Ayres NA, et al. Defining hydrops and indications for open fetal surgery for fetuses with lung masses and vascular tumors. J Pediatr Surg 2012;47:40–5. [2] Adzick NS. Management of fetal lung lesions. Clin Perinatol 2009;36:363–76. [3] Morris LM, Lim FY, Livingston JC, et al. High-risk fetal congenital pulmonary airway malformations have a variable response to steroids. J Pediatr Surg 2009;44:60–5. [4] Crombleholme TM, Coleman B, Hedrick H, et al. Cystic adenomatoid malformation volume ratio predicts outcome in prenatally diagnosed cystic adenomatoid malformation of the lung. J Pediatr Surg 2002;37:331–8. [5] Cass DL, Olutoye OO, Cassady CI, et al. Prenatal diagnosis and outcome of fetal lung masses. J Pediatr Surg 2011;46:292–8. [6] Liu YP, Chen CP, Shih SL, et al. Fetal cystic lung lesions: evaluation with magnetic resonance imaging. Pediatr Pulmonol 2010;45:592–600. [7] Beydon N, Larroquet M, Coulomb A, et al. Comparison between US and MRI in the prenatal assessment of lung malformations. Pediatr Radiol 2013;43:685–96. [8] Chen CP, Liu YP, Lin SP, et al. Prenatal magnetic resonance imaging demonstration of the systemic feeding artery of a pulmonary sequestration associated with in utero regression. Prenat Diagn 2005;25:721–3. [9] Dhingsa R, Coakley FV, Albanese CT, et al. Prenatal sonography and MR imaging of pulmonary sequestration. AJR Am J Roentgenol 2003;180:433–7. [10] Levine D, Barnewolt CE, Mehta TS, et al. Fetal thoracic abnormalities: MR imaging. Radiology 2003;228:379–88. [11] Pacharn P, Kline-Fath B, Calvo-Garcia M, et al. Congenital lung lesions: prenatal MRI and postnatal findings. Pediatr Radiol 2013;43:1136–43. [12] Mehollin-Ray AR, Cassady CI, Cass DL, et al. Fetal MR imaging of congenital diaphragmatic hernia. Radiographics 2012;32:1067–84. [13] Rypens F, Metens T, Rocourt N, et al. Fetal lung volume: estimation at MR imaginginitial results. Radiology 2001;219:236–41. [14] Cass DL, Olutoye OO, Cassady CI, et al. EXIT-to-resection for fetuses with large lung masses and persistent mediastinal compression near birth. J Pediatr Surg 2013;48:138–44. [15] Hedrick HL, Flake AW, Crombleholme TM, et al. The ex utero intrapartum therapy procedure for high-risk fetal lung lesions. J Pediatr Surg 2005;40:1038–43. [16] Zamora IJ, Cass DL, Lee TC, et al. The presence of a hernia sac in congenital diaphragmatic hernia is associated with better fetal lung growth and outcomes. J Pediatr Surg 2013;48:1165–71. [17] Jani J, Cannie M, Sonigo P, et al. Value of prenatal magnetic resonance imaging in the prediction of postnatal outcome in fetuses with diaphragmatic hernia. Ultrasound Obstet Gynecol 2008;32:793–9. [18] Cannie M, Jani J, Meersschaert J, et al. Prenatal prediction of survival in isolated diaphragmatic hernia using observed to expected total fetal lung volume determined by magnetic resonance imaging based on either gestational age or fetal body volume. Ultrasound Obstet Gynecol 2008;32:633–9. [19] Spaggiari E, Stirnemann J, Bernard JP, et al. Prognostic value of a hernia sac in congenital diaphragmatic hernia. Ultrasound Obstet Gynecol 2013;41:286–90. [20] Danzer E, Victoria T, Bebbington MW, et al. Fetal MRI-calculated total lung volumes in the prediction of short-term outcome in giant omphalocele: preliminary findings. Fetal Diagn Ther 2012;31:248–53. [21] Adzick NS, Harrison MR, Crombleholme TM, et al. Fetal lung lesions: management and outcome. Am J Obstet Gynecol 1998;179:884–9.
Discussion Discussant: Dr. Chris Weldon, (Boston, MA): Could you define further neonatal respiratory distress? Was that simply tachypnea? Was that need for emergent intubation, extracorporeal resuscitation techniques? Response: Dr. Zamora: It's a great question. It really was anyone who required any type of supplemental oxygen, beyond the normal neonatal resuscitation period, so any sustained supplemental oxygen, intubation and mechanical ventilation.
Contents lists available at ScienceDirect
Journal of Pediatric Surgery journal homepage: www.elsevier.com/locate/jpedsurg
AAP Papers
Fetal MRI lung volumes are predictive of perinatal outcomes in fetuses with congenital lung masses Irving J. Zamora a, b, Fariha Sheikh a, b, Christopher I. Cassady a, c, Oluyinka O. Olutoye a, b, e, Amy R. Mehollin-Ray a, c, Rodrigo Ruano a, d, Timothy C. Lee a, b, Stephen E. Welty a, d, Michael A. Belfort a, e, Cecilia G. Ethun a, b, Michael E. Kim a, b, Darrell L. Cass a, b, e,⁎ a
Texas Children's Fetal Center, Texas Children's Hospital, Houston, TX The Michael E. DeBakey Department of Surgery, Baylor College of Medicine, Houston, TX c Department of Radiology, Baylor College of Medicine, Houston, TX d Department of Pediatrics, Division of Neonatology, Baylor College of Medicine, Houston, TX e Department of Obstetrics and Gynecology, Baylor College of Medicine, Houston, TX b
a r t i c l e
i n f o
Article history: Received 10 January 2014 Accepted 27 January 2014 Key words: Congenital lung malformation Risk stratification CCAM CPAM Fetal MRI Fetal lung volumes Perinatal outcomes
a b s t r a c t Purpose: The purpose of this study was to evaluate fetal magnetic resonance imaging (MRI) as a modality for predicting perinatal outcomes and lung-related morbidity in fetuses with congenital lung masses (CLM). Methods: The records of all patients treated for CLM from 2002 to 2012 were reviewed retrospectively. Fetal MRI-derived lung mass volume ratio (LMVR), observed/expected normal fetal lung volume (O/E-NFLV), and lesion-to-lung volume ratio (LLV) were calculated. Multivariate regression and receiver operating characteristic analyses were applied to determine the predictive accuracy of prenatal imaging. Results: Of 128 fetuses with CLM, 93% (n = 118) survived. MRI data were available for 113 fetuses. In early gestation (b 26 weeks), MRI measurements of LMVR and LLV correlated with risk of fetal hydrops, mortality, and/or need for fetal intervention. In later gestation (N26 weeks), LMVR, LLV, and O/E-NFLV correlated with neonatal respiratory distress, intubation, NICU admission and need for neonatal surgery. On multivariate regression, LMVR was the strongest predictor for development of fetal hydrops (OR: 6.97, 1.58–30.84; p = 0.01) and neonatal respiratory distress (OR: 12.38, 3.52–43.61; p ≤ 0.001). An LMVR N 2.0 predicted worse perinatal outcome with 83% sensitivity and 99% specificity (AUC = 0.94; p b 0.001). Conclusion: Fetal MRI volumetric measurements of lung masses and residual normal lung are predictive of perinatal outcomes in fetuses with CLM. These data may assist in perinatal risk stratification, counseling, and resource utilization. © 2014 Elsevier Inc. All rights reserved.
Congenital lung masses are rare and encompass a wide spectrum of malformations of the developing fetal lung. Large fetal lung masses compress surrounding thoracic structures. Compression from the mass on normal lung architecture increases the likelihood for pulmonary hypoplasia and respiratory failure following birth [1]. Compression from these lesions on the mediastinum and heart may decrease cardiac venous return and pose a risk for fetal hydrops and fetal demise [2,3]. To date, ultrasound measurement of the CCAMvolume ratio (CVR) has been the primary predictor of prenatal and postnatal outcomes in fetuses with CLM [4,5]. Fetal MRI has been established as an adjunct to ultrasound in the imaging of fetuses with congenital lung malformations (CLM) as early as 18 weeks gestation [6,7]. Compared to ultrasonography, fetal MRI may be superior at characterizing the boundaries of the malformed lung and how it relates to the normal lung lobar anatomy and surrounding thoracic
structures [8–10]. Recently, it has been shown that fetal MRI may be superior in differentiating between the various types of CLMs, including bronchopulmonary sequestration, congenital lobar emphysema, congenital pulmonary airway malformations (CPAM), and bronchogenic cysts [11]. The utility of fetal MRI in assessing volume measurements of CLM both in terms of the mass itself, as well as the uninvolved, normal lung, has not been previously examined. Therefore, the purpose of this study was to evaluate fetal MRI as a modality for predicting perinatal outcomes, including the development of fetal hydrops, mortality, neonatal respiratory distress and lung-related morbidities in fetuses with CLM. 1. Materials and methods 1.1. Study cohort
⁎ Corresponding author at: Texas Children's Hospital, 6701 Fannin St. Suite 1210, Houston, TX 77030. Tel.: + 1 832 822 3135; fax: + 1 832 825 3141. E-mail address: [email protected] (D.L. Cass). http://dx.doi.org/10.1016/j.jpedsurg.2014.01.012 0022-3468/© 2014 Elsevier Inc. All rights reserved.
Permission to conduct this study was obtained from the institutional review board (H-29695) of Baylor College of Medicine,
854
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858
Houston, TX. The records of all patients referred to the Texas Children’s Fetal Center with a congenital lung malformation over a 11-year period from January 2002 to December 2012 were reviewed retrospectively. Fetuses that died owing to non–lung-related comorbidities and those without at least one available fetal MRI were excluded. Data collected from the charts included patient demographics, gestational age at diagnosis and at birth, mortality, fetal course including development of fetal hydrops (defined as presence of serous fluid in two or more body cavities) and need for fetal intervention, and postnatal course including the presence of neonatal respiratory distress (defined as any supplemental oxygen requirement beyond the initial resuscitation period), need for intubation, duration of intubation, NICU admission, need for urgent operation and hospital length of stay. All fetal MRI images were reviewed and lung mass volume ratio (LMVR), observed/expected normal fetal lung volumes (O/E-NFLV), and lesion-to-lung volume ratio (LLV) were calculated.
normal total fetal lung volume (ONTFLV), the volume of normal lung was measured, excluding the volume of the lung mass. This ONTFLV was used to calculate the O/E-NFLV by dividing the measured value by the mean expected TFLV for the respective gestational age of each fetus [13]. The lesion/normal total lung volume ratio was then calculated by dividing the measured lung lesion volume by the ONTFLV, and labeled the “lesion-to-lung volume ratio.” 1.3. Indications for fetal intervention As previously described [5], the primary indication for fetal intervention at our center was the presence of fetal hydrops, defined as serous fluid in two or more body cavities. Fetuses with hydrops and evidence of heart failure on fetal echocardiography were considered for open maternal-fetal surgery. Fetuses that had large lung masses causing persistent mediastinal compression (PMC) late in gestation were offered an ex-utero intrapartum treatment (EXIT) to lung resection approach [14,15].
1.2. Total fetal lung volume and lung mass volume ratio measurements 1.4. Statistical analysis A fetal MRI scan was performed at the time of the initial consultation for most patients referred for a CLM. For most patients this scan was performed before 26 weeks gestation. A subset of patients had a second fetal MRI performed later in gestation (N26 weeks). All lung volumes were calculated on contiguous axial single shot fast spin echo (SSFSE) T2-weighted MR images free from motion by using a freehand region of interest (ROI) tool. The CLM was defined by its hyperintense signal on T2-weighted images and the lesion area was outlined on contiguous sections (Fig. 1). The sum of all measurements was multiplied by the section thickness to obtain a volume in cubic centimeters, which was then divided by the head circumference in centimeters to derive the lung mass volume ratio (LMVR) corrected for gestational age [12]. The total fetal lung volume (TFLV) was measured using the same method. To obtain the observed
Outcomes between groups for categorical variables were compared using Chi-square analysis and Fisher's Exact test. Continuous variables were compared using 2-tailed Student’s t-tests for normally distributed data and Mann-Whitney U test for nonnormally distributed data. An alpha of b0.05 was considered statistically significant. The outcomes examined in this study were stratified according to the gestational age at the time of fetal MRI. For fetal MRIs at b 26 weeks, the primary outcome was the development of fetal hydrops, and secondary outcomes were need for fetal intervention and mortality. For fetal MRIs obtained N26 weeks, the primary outcome was the development of neonatal respiratory distress, and secondary outcomes were need for intubation, NICU admission, and urgent surgery in the neonatal period. Mortality was defined as survival status at the time of chart review. A stepwise multivariate logistic regression was performed for the prenatal categorical outcome of ‘development of hydrops.’ The covariates examined in the regression model were gestational age at diagnosis, gender, location of the CLM and prenatal imaging parameters obtained at b26 weeks gestational age (GA) including LMVR and O/E-NFLV. A separate stepwise regression model was performed for the postnatal categorical outcome of ‘neonatal respiratory distress.’ The covariates examined in the regression model were gestational age at diagnosis, gestational age at birth, gender, location of the CLM and prenatal imaging parameters including LMVR and O/E-NFLV. Interactions were evaluated between all the covariates in the regression models. Receiver operating characteristic curves (ROC) were utilized to determine the predictive accuracy of prenatal imaging for perinatal outcomes. The O/E-NFLV and LMVR measurements obtained at b 26 weeks were evaluated with ROC analysis against the outcome of ‘development of hydrops’ and the O/E-NFLV and LMVR obtained at N 26 weeks were evaluated against the outcome of ‘neonatal respiratory distress.’ Optimal cutoff points of greatest accuracy were derived by graphing the results of the ROC analyses and selecting the point of maximum sensitivity and specificity of each analysis. 2. Results
Fig. 1. A representative image of an axial single shot fast spin echo (SSFSE) T2-weighted MR image of a fetus with CLM. Using a freehand region of interest (ROI) tool the lung lesion and normal fetal lung areas were outlined. The lung lesion was identified as the region of most hyperintense signal in the chest.
During the study period a total of 128 fetuses (65 male) were evaluated for CLM, 93% (118) survived, 63% (81) were left sided, and the average gestational age at diagnosis was 22.1 ± 4.2 weeks. Two fetuses died from non–lung mass related causes and were excluded from further analysis – one from complications owing to Trisomy 18 and one from complications owing to congenital heart disease. MRI data were available for a total of 113 fetuses. Of these fetuses, 57
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858
(50.4%) were male, 66 (58.4%) had left-sided lesions, 46 (40.7%) had right-sided lesions and 1 (0.9%) fetus had a bilateral lesion (Table 1). The mean GA at first MRI was 25.8 ± 4.3 weeks. Seventy-five fetuses had an MRI performed b 26 weeks GA (mean 23.2 ± 1.8 weeks), and 70 had an MRI performed N26 week GA (mean 32.7 ± 3.3 weeks). Thirty-eight fetuses (with high-risk lesions) had two MRIs. Overall, fetuses had relatively large normal lung volumes with a mean O/ENFLV of 87.35%; 43.4% of patients had an O/E-NFLV less than 80%, and only 15% of patients had an O/E-NFLV less than 50%. The mean LMVR was 23.1 cc and the mean LLV ratio was 1.3 (Table 1). As a group fetuses with CLM had a generally benign disease course. There was a low incidence of mortality (5.3%) and postnatal pulmonary morbidity. Pulmonary status at 30 DOL was available for 99 patients: the majority (95%) of these patients were on room air, one was on nasal cannula, one was on CPAP and three remained on mechanical ventilation. A small percentage (3.5%) of patients developed pulmonary hypertension: two of these required treatment with nitric oxide and only one needed sildenafil therapy. A minority (38.1%) of patients required NICU admission. Sixty-five percent of patients underwent surgical resection of the lung mass; most of these patients were readmitted for elective operation at a median age of 98 days. A total of 14 patients (with fetal MRIs) underwent fetal surgery for their CLM: three had open maternal-fetal surgery and 11 underwent EXIT-to-resection (Table 2). 2.1. Early gestation lung volumes and prenatal outcomes 2.1.1. O/E-NFLV as a predictor of prenatal outcomes In early gestation (b 26 weeks GA), MRI measurements of O/ENFLV, LMVR, and LLV were evaluated against prenatal outcomes. The O/E-NFLV was inversely associated with mortality. The five fetuses that died had significantly smaller lung volumes as compared to survivors (O/E-NFLV, 56.8 vs. 88.5%; p = 0.03). However, the O/ENFLV measurements were not markedly different between those patients who developed hydrops or who required fetal intervention as compared to those with an uneventful prenatal course (Table 3). To further examine the association between O/E-NFLV and the development of prenatal symptoms, we performed ROC curve analyses, which demonstrated a poor association between O/E-NFLV and predicting the development of fetal hydrops (AUC = 0.601; p = 0.412) and need for fetal intervention (AUC = 0.54; p = 0.69). 2.1.2. LMVR and LLV as predictors of prenatal outcomes The LMVR and LLV correlated strongly with the risk of developing hydrops, fetal demise and need for fetal intervention. The five fetuses that died had significantly larger lung lesions as measured by LMVR
Table 1 Demographics and prenatal imaging characteristics of all patients with CLM. Total patients Patients with prenatal MRI imaging Males, n (%) Left side, n (%) Right side, n (%) Bilateral, n (%) GA at birth (wk), mean ± SD GA at 1st MRI (wk), mean ± SD Mean CVR, mean ± SD Mean Max CVR, mean ± SD Mean O/E-NFLV (%), mean ± SD # with O/E-NFLV b80%, n (%) # with O/E-NFLV b50%, n (%) Mean LMVR, mean ± SD Mean LLV, mean ± SD
128 113 57 (50.4) 66 (58.4) 46 (40.7) 1 (0.9) 37.7 ± 3.6 25.8 ± 4.3 1.14 ± 1.23 1.25 ± 1.3 87.35 ± 31.8 49 (43.4) 17 (15) 1.10 ± 1.1 1.3 ± 2.9
CLM – congenital lung malformation; GA – gestational age; LMVR – lung mass volume ratio; LLV – lesion-to-lung volume ratio; O/E-NFLV – observed-to-expected normal fetal lung volume.
855
Table 2 Postnatal outcomes of all patients with CLM. Mortality, n (%) LOS, days median (range) Intubated, n (%) Duration of intubation, days median (range) Oxygen requirement at 30 DOL, n = 99 Room air, n (%) Nasal cannula, n (%) CPAP, n (%) Vent, n (%) PAH, n (%) Nitric Oxide Tx., n (%) Sildenafil Tx., n (%) NICU admission, n (%) NICU LOS, median days (range) Surgery, n (%) Readmitted for surgery, n (%) Age at surgery, median days (range) Fetal surgery, n = 13 OMFS EXIT Duration of postop stay, median days (range)
6 (5.3) 5 (0–215) 21 (18.6) 4 (0–42) 94 (95) 1 (0.9) 1 (0.9) 3 (2.7) 4 (3.5) 2 (1.8) 1 (0.9) 43 (38.1) 6.5 (1–215) 73 (64.6) 53 (46.9) 98 (0–2118) 3 (2.7) 11 (9.7) 4 (0–272)
CLM – congenital lung malformation; LOS – length of stay; DOL – day of life; CPAP – continuous positive airway pressure; PAH – pulmonary arterial hypertension; NICU – neonatal intensive care unit; OMFS – open maternal fetal surgery; EXIT – ex-utero intrapartum treatment.
(3.75 vs. 0.73; p b 0.001) and LLV (10.5 vs. 0.8; p b 0.001) than survivors. Fetuses that developed hydrops and who required fetal intervention also had significantly larger lung lesions as measured by both LMVR and LLV (Table 3). To determine which variables were independent predictors of prenatal outcomes we performed a multivariate logistic regression evaluating the covariates of O/E-NFLV, LMVR, GA at first MRI, rightsided lesion and gender. This analysis demonstrated that LMVR was the strongest predictor for the development of fetal hydrops (OR = 6.97, 95%CI: 1.58–30.84; p = 0.01), such that for every 1 unit increase in LMVR there was a 7-fold increase in the likelihood of developing hydrops (Table 4). To further examine the strength of the association between LMVR and the development of fetal hydrops, we performed a ROC curve analysis, which also demonstrated a strong correlation (AUC = 0.94; p = 0.001). Based on the analysis of the sensitivity and specificity we determined the optimal cutoff point was an LMVR N 2.0, which predicted worse perinatal outcome with 83% sensitivity and 99% specificity (Fig. 2). Overall, 107 (94.6%) out of 113 fetuses survived. Of the six that died, four were in the group with initial LMVR N 2.0. There were 19 fetuses (16.8%) with an LMVR N 2.0. These fetuses were born approximately 2 weeks earlier and showed a male and right-side preponderance compared to fetuses with smaller lung lesions (Table 6). Fetuses with LMVR N2.0 had a higher incidence of polyhydramnios (32%), hydrops (42%) and heart failure (32%), and 12 (63%) fetuses underwent fetal intervention (Table 7). Three of these patients underwent open fetal surgery: one had successful resection of a large left extralobar sequestration; one had a right pneumonectomy for mainstem bronchial atresia but died shortly after preterm delivery 4 weeks later at 24 weeks; and a third died intraoperatively during open fetal resection of the right lower lobe for a microcystic CCAM. Seven patients in this group underwent successful EXIT-to-resection procedures and one had a fetal thoracentesis that resulted in early cesarean delivery followed by immediate resection of a large left extralobar sequestration. All of these seven patients did well postoperatively. There were 94 fetuses (83.2%) with an LMVR b2.0. The majority of these lesions (n = 59, 63.4%) were located on the left side. In this group prenatal features of fetal compromise were rare – polyhydramnios (4.3%), hydrops (2.1%), heart failure (2.1%) – however, four patients required fetal intervention for persistent mediastinal
856
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858
Table 3 Fetal MRI features and perinatal outcomes for fetuses with CLM. Variable
O/E-NFLV (mean ± SD)
Outcomes based on F-MRI at b26 weeks Mortality Yes(n = 5) 56.8 ± 45.6 No(n = 70) 88.5 ± 30.8 Hydrops Yes(n = 6) 71.9 ± 41.9 No(n = 69) 87.7 ± 31.7 Fetal intervention Yes(n = 9) 79.9 ± 39.2 No(n = 65) 87.1 ± 32.1 Outcomes based on F-MRI at N26 weeks Neonatal respiratory distress Yes(n = 26) 66.9 ± 35.2 No(n = 41) 98 ± 35.7 Intubation Yes(n = 20) 66.8 ± 29.6 No(n = 48) 92.6 ± 37 NICU admission Yes(n = 32) 69.5 ± 34.8 No(n = 21) 92.9 ± 39.2 Surgery in neonatal period Yes(n = 23) 68.7 ± 36 No(n = 32) 91.3 ± 39.9
p-value
LMVR (mean ± SD)
p-value
LLV (mean ± SD)
p-value
0.03
3.75 ± 2.25 0.73 ± 0.57
b0.001
10.5 ± 9.9 0.8 ± 0.8
b0.001
0.26
3.19 ± 2.01 0.73 ± 0.69
b0.001
7.6 ± 9.6 0.9 ± 1.5
b0.001
0.55
1.88 ± 1.12 0.81 ± 1.02
0.004
4.2 ± 7.8 1.1 ± 2.2
0.012
0.001
2.11 ± 1.03 0.49 ± 0.49
b0.001
1.86 ± 2.3 0.27 ± 0.3
b0.001
0.007
2.14 ± 0.84 0.69 ± 0.85
b0.001
1.64 ± 1.3 0.56 ± 1.6
0.011
0.027
1.75 ± 1.13 0.61 ± 0.73
b0.001
1.54 ± 2.2 0.31 ± 0.45
0.013
0.03
2.16 ± 1.08 0.58 ± 0.61
b0.001
1.96 ± 2.4 0.32 ± 0.44
b0.001
F-MRI; fetal magnetic resonance imaging; CLM – congenital lung malformation; SD – standard deviation; LMVR – lung mass volume ratio; LLV – lesion-to-lung volume ratio; O/E-NFLV – observed-to-expected normal fetal lung volume; NICU – neonatal intensive care unit.
compression (Table 7). Four of these patients with lesions that remained large late in gestation underwent successful EXIT-toresection, and all did well postoperatively. One patient treated with fetal thoracoamnionic shunt placement and serial thoracentesis had fetal demise at 25 weeks and one with a borderline LMVR of 1.9 died of a cardiac arrest on DOL 15, twelve days after resection of the lung mass. These were the only 2 deaths in the group with an LMVR b 2.0.
2.2. Late gestation lung volumes and postnatal outcomes 2.2.1. O/E-NFLV as a predictor of postnatal outcomes In late gestation (N26 weeks GA), MRI measurements of O/E-NFLV, LMVR and LLV were evaluated against postnatal outcomes. The O/ENFLV was highly associated with all postnatal outcomes evaluated including development of neonatal respiratory distress, need for intubation, NICU admission and need for urgent surgery in the neonatal period (Table 3). To further examine the strength of the association between O/E-NFLV and the development of postnatal symptoms, we performed ROC curve analysis, which demonstrated moderate correlation between O/E-NFLV and the development of neonatal respiratory distress (AUC = 0.724; p = 0.005). Based on the analysis of the sensitivity and specificity we determined the optimal cutoff point was an O/E-NFLV b75%, which predicted a worse postnatal course with 57% sensitivity and 80% specificity.
2.2.2. LMVR and LLV as predictors of postnatal outcomes The LMVR and LLV both correlated strongly with the development of neonatal respiratory distress, need for intubation, NICU admission and urgent surgery in the neonatal period. Fetuses that had a compromised neonatal course had significantly larger lung lesions as compared to those that were asymptomatic in the neonatal period (Table 3). To determine which variables were independent predictors of postnatal outcomes, we performed a multivariate logistic regression evaluating the covariates of O/E-NFLV, LMVR, GA at birth, right-sided lesion and gender. This analysis demonstrated that LMVR remained the strongest predictor for the development of neonatal respiratory
Table 4 Multivariable logistic regression for development of fetal hydrops. Covariate
OR
95% CI
p
LMVR O/E-NFLV Right-sided lesion Gender (male) GA at 1st MRI
6.97 5.11 0.27 0.24 0.81
1.58-30.84 0.14-192.02 0.01-9.69 0.01-4.65 0.48-2.59
0.01 0.38 0.47 0.34 0.81
LMVR – lung mass volume ratio; GA – gestational age; LLV – lesion-to-lung volume ratio; O/E-NFLV – observed-to-expected normal fetal lung volume; OR – odds ratio; CI – confidence interval; MRI – magnetic resonance imaging.
Fig. 2. Receiver operating characteristic (ROC) curve of MRI derived lung mass volume ratio (LMVR) at b26 weeks for the outcome of development of fetal hydrops. The point of greatest accuracy corresponded to an LMVR N2.0.
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858 Table 5 Multivariable logistic regression for neonatal respiratory distress.
857
Table 7 Perinatal outcomes of fetuses stratified by LMVR.
Covariate
OR
95% CI
p
Variable
LMVR O/E-NFLV Right-sided lesion Gender (male) GA at Birth
12.38 1.06 3.03 0.75 0.81
3.52-43.61 0.98-1.03 0.51-18.01 0.15-3.82 0.57-1.15
b0.001 0.62 0.22 0.73 0.81
Polyhydramnios, n (%) Hydrops, n (%) Heart failure, n (%) Fetal intervention, n (%)
LMVR – lung mass volume ratio; GA – gestational age; LLV – lesion-to-lung volume ratio; O/E-TFLV – observed-to-expected total fetal lung volume; OR – odds ratio; CI – confidence interval; MRI – magnetic resonance imaging.
distress (OR = 12.38, 95%CI: 3.52–43.61; p b 0.001), such that for every 1 unit increase in LMVR there was an 12-fold increase in the likelihood of being symptomatic at birth (Table 5). To further examine the strength of the association between LMVR and the development of postnatal symptoms, we performed an ROC curve analysis, which also demonstrated a strong correlation (AUC = 0.91; p b 0.001). Based on the analysis of the sensitivity and specificity we determined the optimal cutoff point was an LMVR N1.3, which predicted a worse postnatal course with 86% sensitivity and 98% specificity (Fig. 3). In the 27 fetuses with an LMVR N1.3 during later gestation, indicators of a more complicated postnatal course were more prevalent – neonatal respiratory distress (88.5%), requiring intubation (69%) and NICU admission (92%). Patients in this group also had significantly longer initial hospital lengths of stay and required operative resection of the CLM at a younger age, most in the immediate postnatal period in their first day of life, and also had a longer postoperative length of stay (Table 7). This is in contrast to the 43 patients with an LMVR b 1.3, who had relatively mild postnatal courses – neonatal respiratory distress (7.3%), requiring intubation (4.8%), and NICU admission (34.5%). The majority (90%) of these patients were discharged home and returned later for elective resection of the lung lesion, which was typically followed by a short and uncomplicated postoperative course (Table 7). 3. Discussion
LMVR b2 n = 94 4 2 2 4
(4.3) (2.1) (2.1) (4.3)
LMVR b1.3 n = 43
LMVR N2 n = 19 6 (31.6) 8 (42.1) 6 (31.6) 12 (63.2)
p 0.002 b0.001 b0.001 b0.001
LMVR N1.3 n = 27
Neonatal respiratory distress, n (%) 3 (7.3) 23 (88.5) Required intubation, n (%) 2 (4.8) 18 (69.2) NICU admission, n (%) 10 (34.5) 22 (91.7) Length of stay, median days (range) 3 (0–39) 14 (1–69) Duration of intubation, median days (range) 2 (1–3) 4 (1–30) Duration of NICU stay, median days (range) 4 (1–32) 15 (7–63) Readmitted for elective surgery, n (%) 27 (90) 5 (20) Age at surgery, median days (range) 117 (32–576) 0 (0–539) Postoperative length of stay, median days (range) 3 (1–7) 10.5 (3–69)
b0.001 b0.001 0.019 0.005 0.168 0.107 b0.001 b0.001 b0.001
LMVR – lung mass volume ratio; CLMV – congenital lung malformation volume; NICU – neonatal intensive care unit.
was found to be most significant. Fetuses with larger lung volumes, O/E-NFLV of 75% or greater, predictably did well in general. However, patients with O/E-NFLV less than 75% had a higher incidence of neonatal respiratory distress, intubation, NICU admission, and urgent surgical resection in the immediate postnatal period. These patients had a greater degree of pulmonary hypoplasia with approximately 30% smaller lung volumes than asymptomatic patients with normal O/E-NFLV. In general it is rare for CLM patients to have severe pulmonary hypoplasia, unlike in patients with other conditions such as CDH, which typically have O/E-TFLV measurements ranging from b25% to 45% [16–19] and giant omphaloceles that have approximately 50% O/E-TFLV [20]. When we compared the three fetal MRI parameters, O/E-NFLV, LMVR and LLV, we found that the strongest predictor of outcomes throughout gestation was the LMVR, which correlated highly with prenatal outcomes when measured b26 weeks gestation and with postnatal outcomes when measured N26 weeks gestation. The accuracy of LMVR in predicting a favorable perinatal course was
Fetal MRI derived volumetric measurements of O/E-NFLV, LMVR and LLV accurately predict perinatal outcomes in fetuses with congenital lung malformations. When obtained before 26 weeks’ gestation, fetal MRI-based measurement of O/E-NFLV generally showed poor utility in predicting which fetuses may progress to hydrops and require fetal intervention, but was associated with mortality. However, when obtained later in gestation (N26 weeks) the predictive ability of O/E-NFLV for postnatal outcomes improved significantly. Overall, patients with congenital lung malformation have good prenatal and postnatal outcomes. In particular these patients have a very low risk of experiencing pulmonary symptoms in the neonatal period, likely owing to having relatively high mean O/E-NFLV (87%) as a group. In CLM patients with smaller residual lung volumes, lower measurements of O/E-NFLV correlated strongly with a more complicated postnatal course. A threshold value of 75% for O/E-NFLV Table 6 Demographic characteristics of patients stratified by LMVR. Variable
LMVR b2 n = 94
LMVR N2 n = 19
p
GA at diagnosis, mean ± SD GA at birth, mean ± SD Male, n (%) Right-sided lesion, n (%) Mortality, n (%)
25.4 ± 4.1 38.1 ± 3.3 44 (47.3) 34 (36.6) 2 (2.1)
27.5 ± 5.1 35.8 ± 4.5 13 (72.2) 12 (63.2) 4 (21.1)
0.061 0.013 0.045 0.032 0.007
LMVR – lung mass volume ratio; GA – gestational age; SD – standard deviation.
Fig. 3. Receiver operating characteristic (ROC) curve of MRI derived lung mass volume ratio (LMVR) at N26 weeks for the outcome of neonatal respiratory distress. The point of greatest accuracy corresponded to an LMVR N1.3.
858
I.J. Zamora et al. / Journal of Pediatric Surgery 49 (2014) 853–858
most significant at a threshold of b2.0. In our series 97.9% of fetuses with an LMVR b2.0 did not develop hydrops or heart failure. This is a particularly important outcome since it is well known that fetuses with isolated CLM have a 100% survival when hydrops is absent, but when hydrops is present, the risk of fetal demise increases significantly unless the fetus is treated in some way [21]. An LMVR of b2.0 was also highly predictive of survival since only two deaths occurred in this group, resulting in a 97.9% survival rate. In contrast, fetuses with an LMVR greater than 2.0 experienced poorer outcomes with only a 78.9% survival rate. In our series 80% of the cases that developed hydrops were in this group with larger lung lesions. LMVR obtained after 26 weeks was also useful in predicting postnatal outcomes. Although the majority of the CLM patients had a relatively benign postnatal course, all four of the cases that developed pulmonary hypertension had LMVR N 1.3. Fetuses with an LMVR of 1.3 or greater also required more postnatal care as evident by a higher incidence of neonatal respiratory distress, need for intubation and NICU admission and had an overall longer postoperative and total hospital length of stay. The majority (80%) of fetuses with an LMVR N 1.3 required urgent operation for their lung mass in the neonatal period, which is in contrast to only 10% of fetuses with LMVR b 1.3 who had early surgical intervention. Although the intent of this study was to establish the utility and accuracy of fetal MRI in predicting perinatal outcomes in patients with CLM, we did perform cursory comparative analyses to the current method of ultrasound based CCAM-volume ratio (CVR). On ROC curve analyses we found that LMVR (AUC = 0.91) was slightly stronger at predicting neonatal respiratory distress than CVR (AUC = 0.89). We also noted that CVR tended to overestimate the lung mass volumes. A dedicated comparative study would be required to determine which of these two studies is more useful at predicting perinatal outcomes. Our group has previously helped refine the use of CVR in the management of fetuses with CLM. We recognize that ultrasoundbased CVR is already a highly accurate and less expensive means of risk-stratifying fetuses with CLM and we continue to advocate its use. It is reasonable to hypothesize that fetal MRI may be most useful for those fetuses with larger and higher-risk lesions in order to more precisely assess the lobar anatomy of the mass, relationship of the mass to the surrounding thoracic structures, and the degree of pulmonary hypoplasia which may enhance prenatal decision making and perinatal management. This study is limited by the inherent challenges of a retrospective review. Additionally, the results must be examined with an understanding that these are institution-specific findings and the utility of this study will depend on the reproducibility of these measurements across other institutions. To minimize the variability and to increase the level of consistency in the generation of fetal MRI derived measurements, we use a standardized protocol for obtaining the fetal MRI and a defined method for calculating the volumetric measurements. In this study we propose a novel utility for fetal MRI based measurements in fetuses with congenital lung malformations. We concluded that fetal MRI volumetric measurements of lung malformations utilizing the LMVR and residual normal lung corrected for gestational age (O/E-NFLV) are predictive of perinatal outcomes in fetuses with CLM. These imaging parameters could be useful for risk
stratification and to direct the perinatal management and counseling in cases of prenatally diagnosed lung masses. References [1] Cass DL, Olutoye OO, Ayres NA, et al. Defining hydrops and indications for open fetal surgery for fetuses with lung masses and vascular tumors. J Pediatr Surg 2012;47:40–5. [2] Adzick NS. Management of fetal lung lesions. Clin Perinatol 2009;36:363–76. [3] Morris LM, Lim FY, Livingston JC, et al. High-risk fetal congenital pulmonary airway malformations have a variable response to steroids. J Pediatr Surg 2009;44:60–5. [4] Crombleholme TM, Coleman B, Hedrick H, et al. Cystic adenomatoid malformation volume ratio predicts outcome in prenatally diagnosed cystic adenomatoid malformation of the lung. J Pediatr Surg 2002;37:331–8. [5] Cass DL, Olutoye OO, Cassady CI, et al. Prenatal diagnosis and outcome of fetal lung masses. J Pediatr Surg 2011;46:292–8. [6] Liu YP, Chen CP, Shih SL, et al. Fetal cystic lung lesions: evaluation with magnetic resonance imaging. Pediatr Pulmonol 2010;45:592–600. [7] Beydon N, Larroquet M, Coulomb A, et al. Comparison between US and MRI in the prenatal assessment of lung malformations. Pediatr Radiol 2013;43:685–96. [8] Chen CP, Liu YP, Lin SP, et al. Prenatal magnetic resonance imaging demonstration of the systemic feeding artery of a pulmonary sequestration associated with in utero regression. Prenat Diagn 2005;25:721–3. [9] Dhingsa R, Coakley FV, Albanese CT, et al. Prenatal sonography and MR imaging of pulmonary sequestration. AJR Am J Roentgenol 2003;180:433–7. [10] Levine D, Barnewolt CE, Mehta TS, et al. Fetal thoracic abnormalities: MR imaging. Radiology 2003;228:379–88. [11] Pacharn P, Kline-Fath B, Calvo-Garcia M, et al. Congenital lung lesions: prenatal MRI and postnatal findings. Pediatr Radiol 2013;43:1136–43. [12] Mehollin-Ray AR, Cassady CI, Cass DL, et al. Fetal MR imaging of congenital diaphragmatic hernia. Radiographics 2012;32:1067–84. [13] Rypens F, Metens T, Rocourt N, et al. Fetal lung volume: estimation at MR imaginginitial results. Radiology 2001;219:236–41. [14] Cass DL, Olutoye OO, Cassady CI, et al. EXIT-to-resection for fetuses with large lung masses and persistent mediastinal compression near birth. J Pediatr Surg 2013;48:138–44. [15] Hedrick HL, Flake AW, Crombleholme TM, et al. The ex utero intrapartum therapy procedure for high-risk fetal lung lesions. J Pediatr Surg 2005;40:1038–43. [16] Zamora IJ, Cass DL, Lee TC, et al. The presence of a hernia sac in congenital diaphragmatic hernia is associated with better fetal lung growth and outcomes. J Pediatr Surg 2013;48:1165–71. [17] Jani J, Cannie M, Sonigo P, et al. Value of prenatal magnetic resonance imaging in the prediction of postnatal outcome in fetuses with diaphragmatic hernia. Ultrasound Obstet Gynecol 2008;32:793–9. [18] Cannie M, Jani J, Meersschaert J, et al. Prenatal prediction of survival in isolated diaphragmatic hernia using observed to expected total fetal lung volume determined by magnetic resonance imaging based on either gestational age or fetal body volume. Ultrasound Obstet Gynecol 2008;32:633–9. [19] Spaggiari E, Stirnemann J, Bernard JP, et al. Prognostic value of a hernia sac in congenital diaphragmatic hernia. Ultrasound Obstet Gynecol 2013;41:286–90. [20] Danzer E, Victoria T, Bebbington MW, et al. Fetal MRI-calculated total lung volumes in the prediction of short-term outcome in giant omphalocele: preliminary findings. Fetal Diagn Ther 2012;31:248–53. [21] Adzick NS, Harrison MR, Crombleholme TM, et al. Fetal lung lesions: management and outcome. Am J Obstet Gynecol 1998;179:884–9.
Discussion Discussant: Dr. Chris Weldon, (Boston, MA): Could you define further neonatal respiratory distress? Was that simply tachypnea? Was that need for emergent intubation, extracorporeal resuscitation techniques? Response: Dr. Zamora: It's a great question. It really was anyone who required any type of supplemental oxygen, beyond the normal neonatal resuscitation period, so any sustained supplemental oxygen, intubation and mechanical ventilation.
