Pulmonary Vein Stenosis in Infants: a Systematic Review, Meta-Analysis, and Meta-Regression

ORIGINAL www.jpeds.com • THE JOURNAL OF PEDIATRICS ARTICLES Pulmonary Vein Stenosis in Infants: A Systematic Review, Meta-Analysis, and Meta-Regression

Carl H. Backes,MD1,2,3,4, Erin Nealon,MD1, Aimee K. Armstrong,MD1,2, Clifford L. Cua,MD1,2, Courtney Mitchell,BS4, Usha Krishnan,MD5, Rachel D. Vanderlaan, MD, PhD6, Mi Kyoung Song,MD7, Nicola Viola,MD8,9, Charles V. Smith, PhD10, Patrick I. McConnell,MD11,12,13, Brian K. Rivera,MS4, and Jeffrey Bridge, PhD14

Objective To quantify outcomes of infants (<1 year of age) diagnosed with pulmonary vein stenosis (PVS). Study design MEDLINE (PubMed), Scopus, and Web of Science were searched through February 1, 2017, with no language restrictions. Publications including infants diagnosed with primary PVS, deﬁned as the absence of preceding intervention(s), were considered. The study was performed according to Meta-analysis of Observa- tional Studies in Epidemiology guidelines, the Systematic Reviews, and Meta-Analysis checklist, and registered prospectively. The quality of selected reports was critically examined. Data extraction was independently performed by multiple observers with outcomes agreed upon a priori. Data were pooled using an inverse variance heterogeneity model with incidence of mortality the primary outcome of interest. Results Forty-eight studies of 185 infants were included. Studies were highly diverse with regards to the participants, interventions, and outcomes reported. The median (range) age at diagnosis was 5.0 (0.1-11.6) months. Pooled mortality was 58.5% (95% CI 49.8%-67.0%, I2 = 21.4%). We observed greater mortality incidence among infants with 3 or 4 vein stenoses than in those with 1 or 2 vein stenoses (83.3% vs 36.1%; P < .01). We observed greater mortality among infants with bilateral than unilateral disease (78.7% vs 26.0%; P < .01). Conclusions Studies of primary PVS during infancy are highly variable in their methodological quality and estimates of clinical outcomes; therefore, estimates of prognosis remain uncertain. Multicenter, interdisciplinary collaborations, including alignment of key outcome measurements, are needed to answer questions beyond the scope of available data. (J Pediatr 2018;198:36-45).

ulmonary vein stenosis (PVS) is a rare cardiac disease that, despite decades of research and development of new surgical and interventional techniques, continues to be associated with dismal prognoses.1-4 PVS obstructs blood ﬂow back P to the left atrium, thereby raising pulmonary venous pressure,2,5 resulting in pulmonary edema, pulmonary hypertension, cardiac failure, and death.6,7 Although acquired PVS is a well-described postoperative complication following surgical correction of total anomalous pulmonary venous connection and other heart lesions,8,9 less is known about the development of stenoses in the absence of preceding intervention, termed primary PVS.5 Primary PVS is an emerging problem during infancy,1,5 but few centers have extensive experience with the disease; therefore, the short- and longer-term prognoses are limited mostly to small studies from single centers and remains poorly 1 4,6,7,10-13 From the Department of Pediatrics, Nationwide characterized, with estimates of mortality ranging from 30% to 80%. Ac- Children’s Hospital and The Ohio State University; 2The Heart Center, Nationwide Children’s Hospital; curate estimates of prognosis of primary PVS during infancy, including predic- 3Department of Obstetrics and Gynecology, The Ohio tors of survival and associated outcomes, are needed to guide family-centered State University; 4Center for Perinatal Research, The Research Institute at Nationwide Children’s Hospital, counseling and to inform the feasibility and conduct of future clinical trials. To Columbus, OH; 5Department of Pediatric Cardiology, Columbia University, New York, NY; 6Division of Cardiac our knowledge, no systematic reviews on the outcomes of infants with primary Surgery, University of Toronto, Hospital for Sick Children, PVS have been published. The objectives of the present study were to use meta- Toronto, Ontario, Canada; 7Department of Pediatrics, Seoul National University Children’s Hospital, Seoul, analytic techniques to quantify prognoses among infants with primary PVS and Republic of Korea; 8Department of Thoracic and Cardiovascular Surgery, University Hospital of identify reasons for different estimates of outcomes in the literature. Southampton; 9Department of Cardiothoracic Surgery, Children’s Hospital of Southampton, Southampton, United Kingdom; 10Center for Developmental Therapeutics, Seattle Children’s Research Institute, University of Methods Washington School of Medicine, Seattle, WA; 11Department of Surgery, The Ohio State University, College of Medicine; 12Center for Pediatric Transplant Research, Nationwide Children’s Hospital; 13Department This study was performed according to the Meta-analysis of Observational Studies of Cardiothoracic Surgery, Nationwide Children’s Hospital; in Epidemiology (MOOSE) criteria14 and registered with the PROSPERO data- and 14Center for Innovation in Pediatric Practice, The 15 Research Institute at Nationwide Children’s Hospital, base (#CRD42016051613). MEDLINE (PubMed), Scopus, and Web of Science Columbus, OH were searched through February 1, 2017, with no language restrictions. A Supported by The American Heart Association (10CRP3730033 [to C.B.]).The authors declare no conﬂicts of interest.

36 Volume 198 • July 2018 research librarian was involved in the search, and combina- dysfunction. Consistent with previous studies,6 patients pre- tions of the relevant medical subject heading terms, key words, senting with 3- or 4-vein involvement were distinguished from and word variants were adapted to the different electronic da- those with 1- or 2-vein involvement, when possible. Assess- tabases (Table I; available at www.jpeds.com). The reference ing the severity of stenosis across each pulmonary vein ac- lists of relevant articles and reviews were searched by hand for cording to published criteria3,13,21 was not possible due to additional reports. Nonpeer-reviewed or nonindexed studies inconsistent reporting. Instead, the primary authors’ descrip- (“gray literature”) were not included. tion of the stenosis (eg, focal, diffuse) were captured, where Two authors reviewed all abstracts independently. Agree- possible. If no intervention was performed, reasons were ab- ment regarding potential relevance was reached by consen- stracted, when available. Because disease progression follow- sus. Full-text copies of potentially eligible articles were obtained. ing initial treatment is well-recognized, the need for No articles were excluded based on study design. The same 2 reintervention was also assessed. Reasons for death were also reviewers independently extracted relevant data on study char- abstracted, when available. acteristics and the outcomes of interest. Inconsistencies were discussed with a third author and consensus reached. Data were Data Synthesis and Statistical Analyses collected in standardized format, as recommended by the Co- The inverse variance heterogeneity meta-analysis model pro- chrane Nonrandomized Studies Methods Group.16 For non- posed by Doi et al was selected a priori, based on the assump- English language studies included in the full-text review, articles tion that effects were heterogeneous and on expectations for (N = 22) were translated to English using computer software marked differences across studies.22 Denominators were ad- previously shown to be effective for systematic reviews.17 When justed, where appropriate, to include the numbers of re- data were unclear or missing, the corresponding author (N = 47) ported cases or outcomes of interest. For each primary outcome, was contacted via electronic mail at least twice, in an effort to the prevalence (%) and 95% CIs were calculated based upon obtain additional data to make a final determination of in- the data collected. Subgroup analyses (c2 or Fisher exact test) clusion eligibility. were undertaken to explore potential differences and sources Because PVS among older patients may represent a sepa- of heterogeneity in outcomes. Survival curves were analyzed rate disease process with unique etiologies and outcomes,1,5 we by the log-rank test and included cases with known out- chose to describe outcomes among infants <1 year of age at comes (death) or follow-up for at least 6 months. Forest plots the time of diagnosis of PVS. In rare cases in which the time were used to illustrate individual study findings and the meta- of diagnosis was unknown, infants were included if interven- analysis results. Because traditional meta-analysis based on tions were performed at <1 year of age. Individual cases from inverse variance places undue weight on studies with small or published studies that enrolled mixed populations (infants and large prevalence proportions, MetaXL data analysis software children or adults) were included if individual outcomes for (Epigear; Queensland, Australia) was used with double arcsine infants <1 year could be ascertained. Exclusion criteria were transformation to account for expected variability.23 For pre- as follows: (1) history of cardiovascular surgery or catheter- sentation, pooled transformations and CIs were back- based intervention (other than patent ductus arteriosus closure) transformed to proportions. because the etiology, prognosis, and treatment of those con- Variabilities across studies attributable to heterogeneity ditions may be different than those with primary PVS5,18; (2) beyond chance were estimated with the I2 statistic. Values of studies in which PVS was diagnosed post mortem; (3) studies ≤25%, 25%-75%, and ≥75% represented low, moderate, and that did not provide data on patient age at the time of PVS high heterogeneity, respectively.24 Publication bias was as- diagnosis (or intervention). Conference abstracts and nonpeer- sessed using LFK indices (no bias, index within ±1; minor reviewed literature were excluded. bias, index exceeds ±1 but within ±2; major bias, index Two reviewers independently assessed the methodological exceeds ±2).25 Data not provided for subgroup analysis were quality of studies using the Newcastle-Ottawa Scale for omitted. nonrandomized studies, which uses a star system to assess Mixed effects meta-regression analyses26 were conducted studies on the basis of selection of study groups, comparabil- using Comprehensive Meta-analyses v 2.2 (Biostat; Engle- ity of groups, and ascertainment of exposure/outcome.19,20 No wood, New Jersey) to examine whether continuous modera- studies were excluded on the basis of quality. tors (eg, proportion of infants with severe disease) across studies The primary outcome of interest was death from any cause. influenced mortality event rates. For each meta-regression, Study characteristics included study type (case series vs case studies were included in analysis only if the variable of inter- reports) and location (academic vs private setting). Case series est was known for all included infants. Only studies contrib- were defined as at least 2 infants meeting study criteria, whereas uting more than 1 patient were included in the meta-regression case reports were defined as 1 infant meeting criteria. To analysis. Prognostic factors considered at the study level were compare changes over time, the cohort was divided, based on 3 or 4 vein involvement, bilateral disease, age at diagnosis <6 year of publication, into “epoch 1” (before 2008) and “epoch months, premature infants, chronic lung disease, genetic 2” (2008-2016). Infant, diagnostic, and treatment character- anomalies, or primary receipt of surgical intervention. The slope istics collected are shown in Table II (available at (b) of the regression analysis, Q1 (whether effect size is related www.jpeds.com). Cardiopulmonary failure was defined as pul- to the covariate analyzed), Q2 (variance within the effect size), monary edema, dyspnea, cyanosis, tachypnea, or ventricular and t2 (variance not explained by the model) were reported.26

37 THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 198

All P values were 2 sided; the threshold of signiﬁcance em- (31.1%), atresia (2.2%), or a combination of characteristics ployed was a = 0.05. (24.4%) were described.

Results Treatment Characteristics Among infants with 1 or 2 vein stenoses (N = 72), vascular The flow diagram (Figure 1; available at www.jpeds.com) sum- surgery (40.3%) and nonintervention/medical therapy (31.9%) marizes the identified, screened, eligible, and included studies. were the most common treatments. Catheter-based therapy as The most common reason for exclusion in full-text review was initial intervention was observed in 8.3%; none of these un- PVS following surgical or catheter-based intervention (ac- derwent stent placement. There were no differences in mor- quired PVS). Interrater agreement on the inclusion/exclusion tality between those undergoing primary surgical intervention of articles was good (k = 0.76). (36.7%) or nonintervention/medical therapy, (26.1%, P = .41), The 48 studies published between May 1974 and Septem- or among those undergoing sutureless marsupialization vs patch ber 2016 included 185 infants (Table III). The majority of venoplasty (40.0% vs 33.3%, P > .99). studies (N = 42) provided outcomes beyond initial presenta- Among infants with 3 or 4 vein stenoses (N = 72), vascu- tion for a median (range) of 5 months (0.1-168 months); 3 lar surgery (47.2%) and nonintervention/medical therapy infants were lost to follow-up following diagnosis (30.6%) were the most common interventions. Catheter- (intervention).1,32 All studies were performed at academic in- based therapy was attempted in 19.4%, of which 7 under- stitutions, with most at single centers (N = 44). Interrater agree- went stent placement. We observed no difference in mortality ment on the methodological quality of included articles was among undergoing primary surgical intervention (91.2%) vs good (k = 0.89). Studies ranged from 3 to 6 stars on the nonintervention/medical therapy (90.9%, P = 1.0). Reasons for Newcastle-Ottawa scale (range, 0-9; with a lower score indi- no intervention/medical therapy (N = 22) included health- cating methodological weakness). Included studies were highly care providers noting medical futility (N = 5), familial with- diverse in terms of participants, interventions, and outcomes drawal of support (N = 5), or not specified (N = 12). measured (Table IV). No prospective observational studies were The pooled prevalence of death was 58.5% (95% CI 49.8%- identified. 67.0%) with low heterogeneity (I2 = 21.4; Figure 2); no pub- The median (range) age at diagnosis was 5.0 (0.1-11.8) lication bias was evident (LFK Index = 0.78). Case reports were months, and 52% of infants were female. Left-to-right shunt included, as limitation to case series did not change the esti- lesions and critical heart disease were observed in 106 (57.3%) mate of death (64.2% [95% CI 48.9%-78.2%]) with no het- and 29 (18.7%) infants, respectively. erogeneity (I2 = 0.0), but with evidence of publication bias (LFK index =−3.38). Mortality was associated with numbers of veins Diagnostic Assessment involved (log-rank test, P < .01; Figure 3, A) and laterality (log- PVS was defined and diagnosed in different ways in the primary rank test, P < .01; Figure 3, B). Pooled mortality prevalence is studies (Table I). Echocardiography was the primary method shown in Table V (available at www.jpeds.com). We ob- for diagnosis in 9 studies (18.8%), cardiac catheterization in served greater mortality among infants with 3- or 4-vein ste- 16 (33.3%), and a combination of both methods in 12 (25.0%) noses than those with 1- or 2-vein stenoses (83.3% vs 36.1%; studies. Additional diagnostic methods included computed to- P < .01) and among infants with bilateral compared with uni- mography in 3 studies and magnetic resonance angiography lateral disease (78.7% vs 26.0%; P < .01). in 1 study. Reported echocardiographic criteria included flow Mortality of preterm (56.9%) and term infants with PVS velocity (range 1.5- >1.6 m/second) and echocardiographic (65.6%; P = .27) did not differ. Mortality in infants diag- mean pressure gradient (>5mmHg,>4 mm Hg, and ≥3mm nosed in the first 6 months of life (64.1%) did not differ from Hg). those diagnosed 6-12 months of life (55.0%; P = .28). Twenty- The pre- and postintervention assessments of cardiac he- three reports (62 infants) detailed evidence of restenosis or modynamics were markedly disparate, with some simply noting disease progression following initial intervention. Outcomes pulmonary hypertension (yes/no), whereas other studies pro- among those undergoing reintervention (N = 11) are shown vided mean pulmonary artery pressures or combinations of in Table VI (available at www.jpeds.com). mean pulmonary artery pressures and pulmonary capillary Among 107 infants who died, the cause of death could not wedge pressures. In 24 studies (47 infants), initial be determined in 49 (46%). Among the 58 infants with a known echocardiograms showed no evidence of pulmonary venous cause of death, etiologies included restenosis/ disease progres- disease, but PVS became apparent with subsequent follow-up. sion (N = 33, 56.9%), pulmonary hypertensive crisis, cardio- respiratory failure, and/or pulmonary hemorrhage (N = 17, Pulmonary Vein Characteristics 29.3%), sepsis (N = 5, 8.6%), palliative care/withdrawal of care The most common location of stenosis reported at diagnosis (N = 2, 3.4%), and thrombosis at the surgical site (N = 1, 1.7%). was left upper pulmonary vein (53.5%). Bilateral vein steno- Autopsy findings were available in 21 cases. Common themes ses at presentation were seen in 51.9% of reported cases. Uni- among autopsies included medial hypertrophy of vessel, fibrous lateral left-sided disease (73.0%) was more common than right- intimal thickening, and intimal proliferation. sided disease (27.0%). Among cases (N = 45) providing a clear There were no differences in survival among infants in epoch description of the type of stenosis, focal (42.2%), diffuse 2 (42.7%) compared with epoch 1 (39.2%; P = .47) or in the

38 Backes et al July 2018 ORIGINAL ARTICLES

Table III. Manuscripts included in the meta-analysis Newcastle-Ottawa Scale Age at Primary method(s) Eligible diagnosis, of diagnosis; Selection, Comparability, Outcome, Authors Year cases Characteristics of study mean (mo) definition of PVS of 3 stars of 2 stars of 3 stars More than 1 included PVS case Bini et al27† 1984 3 Single-center; all infants with 8.7 ± 2.1 Echocardiography and *** N/A *** FTT, severe dyspnea, catheterization; no definition subcostal retractions Breinholt et al6 1999 10 Single-center; excluded 4.7 ± 2.4 Echocardiography; Doppler non- *** N/A *** infants with anomalous PV phasic velocity >1.6 m/s return Chakrabarti et al28 2007 5 Two referral centers; 7.2 ± 2.4 Echocardiography; no definition *** N/A *** excluded infants with anomalous PV return Charlagorla et al29† 2016 13 Single-center; infants 6.5 ± 5.7 Unknown; no definition ** * *** undergoing surgical/ 7.2 ± 4.6 catheter-based intervention Devaney et al30 2006 14 Single-center; single vein – Unknown; no definition ** N/A *** disease not included Driscoll et al31 1982 4 Single-center, all presented 5.3 ± 1.5 Catheterization; angiographic *** N/A *** with heart failure and FTT discrete stenosis Drossner et al7 2008 10 Single-center, excluded 5.2 ± 3.7 Echocardiography; mean *** N/A *** scimitar syndrome, gradient >5mmHg anomalous PV return and hypoplastic left heart syndrome Heching et al32 2014 20 Single-center; infants <37 wk 4.9 ± 2.2 Echocardiography; mean pulse ** N/A ** of gestation; exclude wave gradient ≥3mmHg TAPVC and HLHS Holcomb et al33 1999 2 Single-center; full term 0.1 ± 0.1 Echocardiography, high velocity *** N/A *** infants with PPHN flow; catheterization; increased pressures Holt et al1 2007 18 Multicenter database; 3.8 ± 2.1 Unknown; no definition *** N/A *** excluded TAPVC, cor triatriatum Kanter et al34 2014 2 Single-center, infants – Unknown; no definition ** N/A *** undergoing surgical therapy Laux et al35 2016 8 Multicenter study; infants 4.7 ± 1.9 Catheterization; angiography ** N/A *** <37 wk of gestation referred for evaluation of PH Mendeloff et al36 1995 4 Single-center; presented with 3.7 ± 2.1 Echocardiography and *** N/A *** heart failure and cyanosis catheterization; no definition Presbitero et al37 1983 2 Single-center; all with 4.5 ± 3.5 Cardiac catheterization, ** N/A ** additional ventricular angiography septal defect Santoro et al38 1996 2 Single-center; patients 4.0 ± 1.4 Echocardiography, turbulent, *** N/A *** undergoing intraoperative high velocity flow stent implantation Song et al18 2013 14 Multicenter; excluded infants 5.5 ± 2.8 Echocardiography; monophasic *** N/A *** with TAPVC or left lower flow pattern, flow >1.5 m/s PVS by a large left-to-right shunt lesion and cardiomegaly Stewart et al39 1992 2 Single-center; Trisomy 21 5.6 ± 4.7 Echocardiography, turbulent ** N/A *** infants flow; catheterization; angiography Swier et al11 2016 9 Single-center; infants with 6.3 ± 2.7 Echocardiography; mean ** N/A *** severe BPD; excluded gradient >4mmHg other congenital heart lesions and genetic anomalies van de Laar et al40 2009 3 Single-center, genetic study 4.2 ± 5.1 Echocardiography, flow >1.6 m/ ** N/A *** of a single family s; catheterization; angiography Viola et al13 2011 12 Single-center; surgical – Echocardiography; pulmonary *** N/A *** database vein score (continued)

Pulmonary Vein Stenosis in Infants: A Systematic Review, Meta-Analysis, and Meta-Regression 39 THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 198

Table III. Continued Newcastle-Ottawa Scale Age at Primary method(s) Eligible diagnosis, of diagnosis; Selection, Comparability, Outcome, Authors Year cases Characteristics of study mean (mo) definition of PVS of 3 stars of 2 stars of 3 stars Only 1 included PVS case Amin et al41 2009 1 Preterm infant with BPD and 5.0 Echocardiography, turbulent *** N/A *** PH flow; computed tomography, hypoplasia; catheterization, angiography; no definition Bahena et al42 2006 1 Infant with pulmonary artery 9.0 Magnetic resonance ** N/A ** originating from the aorta angiography; no definition Bartolome et al43 2001 1 Preterm infant with heart 7.0 Catheterization, angiography *** N/A *** failure Benjamin et al44 2009 1 Extremely preterm infant with 6.0 Echocardiography; no definition *** N/A *** PH Bonello et al45 2015 1 Preterm infant with BPD 10.0 Echocardiography; no definition *** N/A ** Bravo-valenzuela 2015 1 Infant with craniofacial – Catheterization; angiography *** N/A *** et al46 malformations and severe PH Cromme-Dijkhuis 1995 1 Infant with CHF, presumed cor 2.0 Diagnosed at surgery ** N/A *** et al47 triatriatum Fujii et al48 2014 1 Infant with cyanosis and 11.0 Catheterization, increased *** N/A *** severe PH pressures; computed tomography, angiography Geggel et al49 1984 1 Term infant 7.5 Catheterization, angiography *** N/A *** Gowda et al50 2014 1 Single-center, Trisomy 21 7.0 Catheterization, angiography ** N/A *** infant Jean-St.-Michel 2016 1 Single-center, heart – Unknown; no definition ** N/A *** et al51 transplant with diagnosis of PV stenosis Khositseth et al52 2010 1 Infant with multiple 6.0 Computed tomography; ** N/A * cardiopulmonary catheterization; angiography malformations Lai et al53 1994 1 Infant with ASD; multiple 7.0 Echocardiography; turbulent *** N/A ** pulmonary infections flow/gradient; catheterization; angiography Li et al54 2012 1 Infant with unroofed coronary 8.0 Diagnosis at surgery for ** N/A ** sinus unroofed coronary sinus Massaro et al55 2008 1 Infant with failure to wean 4.0 catheterization; angiography *** N/A *** from ECMO Mendelsohn 1993 1 Single-center; infants 8.4 Unknown; no definition ** N/A *** et al56 undergoing intraoperative/ catheter-based stenting Moura et al57 2003 1 Preterm infant with duodenal 1.1 Echocardiography, turbulent *** N/A *** atresia, cyanosis flow/gradient; catheterization, angiography Mueller et al58 2010 1 Infant with pulmonary 0.7 Echocardiography; no definition *** N/A *** haemorrhage Mühler et al59 1991 1 Infant with tachycardia, PH 2.5 Catheterization, angiography *** N/A *** Pacifico et al60 1985 1 Single-center; infants 4.0 Cardiac catheterization, *** N/A *** undergoing surgical repair angiography Park et al61 1974 1 Infant with CHF 11.5 Catheterization; pressure *** N/A *** Sade et al62 1974 1 Infant with respiratory 8.0 Catheterization; increased *** N/A *** disease, FTT wedge pressures, angiography Seale et al63 2006 1 Single-center; infants treated 1.5 Catheterization, angiographic ** N/A *** with cutting balloon discrete stenosis procedures for PVS Smith64 2012 1 Preterm infant with BPD 3.9 Echocardiography, elevated flow *** N/A *** velocities Urban et al65 1996 1 Infant with respiratory failure, 2.8 Echocardiography, turbulent *** N/A *** PH flow; catheterization, pressures van Son et al66 1995 1 Single-center; infants – Catheterization; angiography ** N/A *** underwent surgical therapy Webber et al67 1992 1 Preterm infant with FTT 1.8 Echocardiography; continuous, *** N/A *** turbulent flow (1.5 m/s)

– , not able to calculate; BPD, bronchopulmonary dysplasia; ECMO, extracorporeal membrane oxygenation; FTT, failure to thrive; HLHS, hypoplastic left heart syndrome; PAPVC, partial anomalous pulmonary venous connection; PH, pulmonary hypertension; PPHN, persistent pulmonary hypertension; PVS, pulmonary vein; TAPVC, total anomalous pulmonary venous connection. †Excluded those undergoing procedures other than sutureless marsupialization, surgical dilation, or balloon angioplasty, as well as those with acquired PVS or total occlusive disease.

40 Backes et al July 2018 ORIGINAL ARTICLES

Prognostic Features in Meta-Regression Table IV. Aggregate data synthesis for included studies At the study level (Figure 4; available at www.jpeds.com), greater Study characteristics (N = 48) Number of studies (%) mortality rates were associated with infants presenting with

Type of report 3- or 4-vein involvement (b = 1.87, Q1 = 5.64, Q2 = 9.65, Case series (>1 PVS case) 20 (42) t2 = 0.00; P < .01). We observed no evidence of association of Single case reports (single PVS case) 28 (58) Location of study* mortality rates according to the proportion of infants with bi- US 23 (48) lateral involvement, <6 months of age at diagnosis, prema- Canada 4 (8) ture birth, chronic lung disease, genetic anomalies, or primary Europe 13 (27) Eastern Asia 5 (10) receipt of surgical intervention (data available upon request) Other† 3 (6) across studies. Included patients Entire cohort 29 (60) Partial cohort 19 (40) Discussion Patient characteristics (N = 185) Number of Cases (%) Premature birth (<37 wk of gestational age) 67 (36) Because most institutions have limited experience with PVS, Small for gestational age 17 (9) Bronchopulmonary dysplasia 34 (18) the data presented in this study should be useful due to the Associated conditions size of the cohort with primary PVS. Although these data Trisomy 21 11 (6) suggest that primary PVS during infancy has a poor progno- Other‡ 6 (3) Age at presentation sis, lack of harmonized collection, entry criteria, definitions, <3 mo 36 (20) and surveillance protocols across studies limit interpreta- 3-<6 mo 57 (31) tion. Given the large variability and widely ranging heteroge- 6-<9 mo 49 (27) 9-<12 mo 15 (8) neity of the studies from which the data were drawn, pooled Not specified 28 (15) results should be interpreted with caution. However, the data Symptoms at presentation help identify sources of variability in results across studies, rep- Pulmonary hypertension 101 (55) Cardiopulmonary failure§ 75 (41) resent the best available evidence for clinical care and family- Failure to thrive 19 (10) centered counseling, and can be used to inform sample-size Multiple respiratory infections 5 (3) calculations for future epidemiologic studies. Number of veins involved at diagnosis 1 25 (14) Traditionally, primary PVS is characterized by an unrelent- 2 47 (25) ing, progressive course that is unresponsive to treatment and 3 28 (15) is usually fatal.1,7,12,18 To that end, our observation of worse prog- 4 44 (24) Unable to determine¶ 41 (22) nosis based on severity of presentation (severe disease) aligns Primary intervention with the available literature, including 1-year mortality rates Vascular surgery: approaching 80% for those with 3 or 4 vein disease.1,4-6,18,21,29,68,69 Resection/reimplantation 5 (3) Patch venoplasty 23 (12) However, in contrast to previous studies that reported no mor- 1,6 Sutureless marsupialization 40 (22) tality in patients with 1 or 2 stenosed veins, our observa- Stent placement (surgical implantation) 3 (2) tion of mortality rates approaching one-third and one-half of Dilation of ostia 7 (4) Not specified 12 (7) patients with 1- or 2-vein disease is noteworthy. Although we Catheterization: accounted for the number of pulmonary veins involved, the Balloon angioplasty only 12 (7) severity and extent of stenosis are also key determinants of Balloon angioplasty and stent placement 11 (6) 21 Transplantation (heart/lung) 7 (4) outcomes. Consistent use of PVS “scoring systems,” includ- Lobectomy (lung) 2 (1) ing assessment of “upstream” stenosis within the lung paren- Palliative care/no intervention 52 (24) chyma, will provide a more robust characterization of ¶ Unable to determine 25 (14) pulmonary venous disease and may offer important prognos- Data are shown as n (% of cases). tic information.3,13,21 *According to location of first author. 6,7,11,12 †New Zealand, Mexico, and Brazil (1 each). Consistent with previous reports, an appreciable ‡Lung mesenchymal dysplasia (1), septo-optic dysplasia (1), Smith-Lemli-Opetz syn- number of infants had no evidence of disease on initial drome (1), Trisomy 16q and Trisomy 18p (1, same case), Cutis marmorata telangiectatica congenita (1), 47 + long arm of chromosome 1 (1). echocardiography. Echocardiography can underestimate the §Cardiopulmonary failure defined as any of the following: dyspnea, cyanosis, tachypnea, right degree of PVS because blood flow is redistributed toward less- ventricular dysfunction. ¶Unable to link infant(s) with associated outcome or intervention. obstructed portions of the lung, resulting in decreased flow through stenotic veins.70 Alternatively, pulmonary veins may proportion of infants with involvement of 3 or 4 veins in epoch have been normal initially, and stenosis developed over time.6,7,11 2 (46.6%) vs epoch 1 (56.4%; P = .26). More infants were di- Even though many studies continue to use echocardiography agnosed with echocardiography only in epoch 2 than in epoch in the diagnosis of PVS, potential advantages of cardiac com- 1 (54.4% vs 8.1%; P < .01). Among infants undergoing surgery, putedtomographyormagneticresonanceimaginginclude we observed an increase in the incidence of a “sutureless” repairs better delineation of pulmonary venous anatomy and blood in epoch 2 (38.5%) vs epoch 1 (10.3%; P < .01, Table VII [avail- flow distribution.21 To answer questions on timing and nature able at www.jpeds.com]). of disease progression, more uniform methods of diagnosis

Pulmonary Vein Stenosis in Infants: A Systematic Review, Meta-Analysis, and Meta-Regression 41 THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 198

Figure 2. Mortality forest plot. Forest plot with pooled mortality prevalence, CI, and tabulated I-squared value among all included studies.

42 Backes et al July 2018 ORIGINAL ARTICLES

Risk/benefit ratios are likely to be continuous in nature and dependent on factors beyond age at time of diagnosis. This review provides an evaluation of mortality, but additional outcomes (morbidities, quality of life) were beyond the scope of available data. Although we describe initial treatments for PVS, repeat procedures are frequently needed.5 Limited numbers precluded additional subgroup analysis based on type of intervention. Consistent with previous studies,3,13 patient-related factors (eg, prematurity, genetic anomalies) were not associated with greater risk of adverse outcomes; however, important subgroups were small. The retrospective nature of available studies increases the risk of selection bias. Observed differences between case-level and study-level data on the influ- ence of laterality on mortality are not unexpected79 and reinforce the importance of individual patient data. Although we performed a comprehensive literature search, data not indexed in electronic databases (“gray literature”) were not included.80 Because mechanisms for PVS remain largely unknown, we attempted to identify common themes in autopsy findings; however, lack of concordance between reported pathologic findings precluded summative observations. In light of the limitations of available data, healthcare providers managing infants with PVS are encouraged to document and publish their results to further the collective knowledge. Studies of primary PVS during infancy are highly variable in their methodological quality and estimates of clinical outcomes and although this is the largest known meta-analysis among infants with primary PVS, estimates of prognosis remain uncertain. The present study confirms that multicenter, interdisciplinary collaborations, including alignment of key Figure 3. Survival curves. Kaplan-Meijer survival curves based outcome measurements, are needed to answer questions beyond upon A, number of veins involved; and B, laterality (unilat- the scope of available data. ■ eral vs bilateral) at presentation. We thank Alison Gehred, MS, Research Librarian at Nationwide Chil- dren’s Hospital for her assistance in the database searches, the physi- cians who provided follow-up information, and the patients, including Lincoln Ray Hockenberry-Habedank. surveillance, with prespecified cut-off values for defining PVS, 2,21 are needed. Submitted for publication Nov 7, 2017; last revision received Jan 8, 2018; Evidence on longer-term benefits of treatments for primary accepted Feb 13, 2018 2,31,36,71-73 PVS is lacking. Although we observed no differences Reprint requests: Carl H. Backes, MD, Department of Pediatrics, Nationwide in outcomes among infants with severe disease provided sur- Children’s Hospital, 700 Children’s Dr, Columbus, OH 43205. E-mail: [email protected] gical intervention vs nonintervention/medical therapy, base- line heterogeneity among enrolled patients, variation in type of intervention, and inconsistency in completeness of data re- References ported among studies, limit data interpretations. In light of 1. Holt DB, Moller JH, Larson S, Johnson MC. Primary pulmonary vein steno evidence of improved survival across epochs, careful con- nosis. Am J Cardiol 2007;99:568-72. sideration of alternative treatment modalities is warranted, with 2. Seale AN, Daubeney PEF. Pulmonary vein stenosis–novel strategies for the goal of providing more evidence-based approaches to PVS a challenging and resistant condition? J Thorac Cardiovasc Surg care.74-77 2016;151:618-20. 3. Kalfa D, Belli E, Bacha E, Lambert V, di Carlo D, Kostolny M, et al. Primary To address questions beyond the scope of available data, col- pulmonary vein stenosis: outcomes, risk factors, and severity score in a laborations among research groups to align key measure- multicentric study. Ann Thorac Surg 2017;104:182-9. ments and outcomes that are important to parents, healthcare 4. Mahgoub L, Kaddoura T, Kameny AR, Lopez Ortego P, Vanderlaan RD, professionals, and policymakers alike, will be necessary. A col- Kakadekar A, et al. Pulmonary vein stenosis of ex-premature infants with laborative multicenter PVS registry study, led by a team from pulmonary hypertension and bronchopulmonary dysplasia, epidemiology, and survival from a multicenter cohort. Pediatr Pulmonol The Hospital for Sick Children in Toronto (PVS Network) will 2017;52:1063-70. address contemporary treatment practices using standard- 5. Latson LA, Prieto LR. Congenital and acquired pulmonary vein steno- ized outcome measures.78 sis. Circulation 2007;115:103-8.

Pulmonary Vein Stenosis in Infants: A Systematic Review, Meta-Analysis, and Meta-Regression 43 THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 198

6. Breinholt JP, Hawkins JA, Minich LA, Tani LY, Orsmond GS, Ritter S, et al. 29. Charlagorla P, Becerra D, Patel PM, Hoyer M, Darragh RK. Congenital Pulmonary vein stenosis with normal connection: associated cardiac ab- pulmonary vein stenosis: encouraging mid-term outcome. Pediatr Cardiol normalities and variable outcome. Ann Thorac Surg 1999;68:164-8. 2016;37:125-30. 7. Drossner DM, Kim DW, Maher KO, Mahle WT. Pulmonary vein steno- 30. Devaney EJ, Chang AC, Ohye RG, Bove EL. Management of congenital sis: prematurity and associated conditions. Pediatrics 2008;122:e656-61. and acquired pulmonary vein stenosis. Ann Thorac Surg 2006;81:992-5. 8. Devaney EJ, Ohye RG, Bove EL. Pulmonary vein stenosis following repair discussion 995-996. of total anomalous pulmonary venous connection. Semin Thorac 31. Driscoll DJ, Hesslein PS, Mullins CE. Congenital stenosis of individual Cardiovasc Surg Pediatr Card Surg Annu 2006;51-5. pulmonary veins: clinical spectrum and unsuccessful treatment by 9. Hickey EJ, Caldarone CA. Surgical management of post-repair pulmo- transvenous balloon dilation. Am J Cardiol 1982;49:1767-72. nary vein stenosis. Semin Thorac Cardiovasc Surg Pediatr Card Surg Annu 32. Heching HJ, Turner M, Farkouh-Karoleski C, Krishnan U. Pulmonary vein 2011;14:101-8. stenosis and necrotising enterocolitis: is there a possible link with 10. Krishnan U, Feinstein JA, Adatia I, Austin ED, Mullen MP, Hopper RK, necrotising enterocolitis? Arch Dis Child Fetal Neonatal Ed 2014;99:F282- et al. Evaluation and management of pulmonary hypertension in chil- 5. dren with bronchopulmonary dysplasia. J Pediatr 2017;188:24-34. 33. Holcomb RG, Tyson RW, Ivy DD, Abman SH, Kinsella JP. Congenital pul- 11. Swier NL, Richards B, Cua CL, Lynch SK, Yin H, Nelin LD, et al. Pul- monary venous stenosis presenting as persistent pulmonary hyperten- monary vein stenosis in neonates with severe bronchopulmonary dys- sion of the newborn. Pediatr Pulmonol 1999;28:301-6. plasia. Am J Perinatol 2016;33:671-7. 34. Kanter KR, Kirshbom PM, Kogon BE. Surgical repair of pulmonary venous 12. Seale AN, Webber SA, Uemura H, Partridge J, Roughton M, Ho SY, et al. stenosis: a word of caution. Ann Thorac Surg 2014;98:1687-91. discus- Pulmonary vein stenosis: the UK, Ireland and Sweden collaborative study. sion 1691-1682. Heart 2009;95:1944-9. 35. Laux D, Rocchisani MA, Boudjemline Y, Gouton M, Bonnet D, Ovaert 13. Viola N, Alghamdi AA, Perrin DG, Wilson GJ, Coles JG, Caldarone CA. C. Pulmonary hypertension in the preterm infant with chronic lung disease Primary pulmonary vein stenosis: the impact of sutureless repair on sur- can be caused by pulmonary vein stenosis: a must-know entity. Pediatr vival. J Thorac Cardiovasc Surg 2011;142:344-50. Cardiol 2016;37:313-21. 14. Stroup DF, Berlin JA, Morton SC, Olkin I, Williamson GD, Rennie D, et al. 36. Mendeloff EN, Spray TL, Huddleston CB, Bridges ND, Canter CB, Mallory Meta-analysis of observational studies in epidemiology: a proposal for GB Jr. Lung transplantation for congenital pulmonary vein stenosis. Ann reporting. Meta-analysis Of Observational Studies in Epidemiology Thorac Surg 1995;60:903-6. discussion 907. (MOOSE) group. JAMA 2000;283:2008-12. 37. Presbitero P, Bull C, Macartney FJ. Stenosis of pulmonary veins with ven- 15. PROSPERO. International prospective register of systemic reviews. https:// tricular septal defect. A cause of premature pulmonary hypertension in www.crd.york.ac.uk/PROSPERO/. Accessed November 16, 2017. infancy. Br Heart J 1983;49:600-3. 16. Higgins JPT, Lasserson T, Chandler J, Tovey D, Churchill R. Method- 38. Santoro G, Formigari R, Mazzera E, Ballerini L. [Intraoperative stent im- ological Expectations of Cochrane Intervention Reviews. 2016. plantation in congenital stenosis of pulmonary veins]. G Ital Cardiol http://community.cochrane.org/mecir-manual. Accessed August 1, 2017. 1996;26:201-5. 17. Balk EM, Chung M, Hadar N, Patel K, Yu WW, Trikalinos TA, et al. Ac- 39. Stewart AD, Calder AL, Neutze JM, James AH, Brandt PW. Stenosis of curacy of data extraction of non-English language trials with google trans- pulmonary veins in Down syndrome. J Paediatr Child Health 1992;28:192- late [Internet]. Rockville (MD): Agency for Healthcare Research and Quality 5. (US); 2012. 40. van de Laar I, Wessels M, Frohn-Mulder I, Dalinghaus M, de Graaf B, van 18. Song MK, Bae EJ, Jeong SI, Kang IS, Kim NK, Choi JY, et al. Clinical char- Tienhoven M, et al. First locus for primary pulmonary vein stenosis maps acteristics and prognostic factors of primary pulmonary vein stenosis or to chromosome 2q. Eur Heart J 2009;30:2485-92. atresia in children. Ann Thorac Surg 2013;95:229-34. 41. Amin R, Kwon S, Moayedi Y, Sweezey N. Pulmonary vein stenosis: case 19. Wells GA, Shea B, O’Connell D, Peterson J, Welch V, Losos M, et al. The report and literature review. Can Respir J 2009;16:e77-80. Newcastle-Ottawa Scale (NOS) for assessing the quality of nonrandomised 42. Bahena EP, Calderón-Colmenero J, Ramírez S, Montes JAG, Meave A, studies in meta-analyses. 2014. http://www.ohri.ca/programs/clinical_ Erdmenger J, et al. Origen anomalo de la arteria pulmonar derecha y epidemiology/oxford.asp. Accessed September 15, 2017. estenosis de vena pulmonar superior izquierda. Arch Cardiol Mex 20. Margulis AV, Pladevall M, Riera-Guardia N, Varas-Lorenzo C, Hazell L, 2006;76:80-2. Berkman ND, et al. Quality assessment of observational studies in a drug- 43. Bartolome FB. Estenosis congenita de venas pulmonares: eﬁcacia a largo safety systematic review, comparison of two tools: the Newcastle- plazo tras dilatacion intraluminal. Rev Esp Cardiol (Engl Ed) 2001;54:1111- Ottawa Scale and the RTI item bank. Clin Epidemiol 2014;6:359-68. 2. 21. Lo Rito M, Gazzaz T, Wilder TJ, Vanderlaan RD, Van Arsdell GS, Honjo 44. Benjamin JT, Hamm CR, Zayek M, Eyal FG, Carlson S, Manci E. Ac- O, et al. Pulmonary vein stenosis: severity and location predict quired left-sided pulmonary vein stenosis in an extremely premature infant: survival after surgical repair. J Thorac Cardiovasc Surg 2016;151:657- a new entity? J Pediatr 2009;154:459, e451. 66, e652. 45. Bonello B, Trivedi KR, Fraisse A. Multiple and aggressive pulmonary vein 22. Doi SA, Barendregt JJ, Khan S, Thalib L, Williams GM. Advances in the transcatheter interventions as bridge to transplantation in primary meta-analysis of heterogeneous clinical trials I: the inverse variance het- diffuse pulmonary vein stenosis. Catheter Cardiovasc Interv 2015;86:E190- erogeneity model. Contemp Clin Trials 2015;45:130-8. 3. 23. Barendregt JJ, Doi SA, Lee YY, Norman RE, Vos T. Meta-analysis of preva- 46. Bravo-Valenzuela NJ, Silva GR, Varella MP. Pulmonary vein stenosis in lence. J Epidemiol Community Health 2013;67:974-8. a newborn: a commonly overlooked diagnosis. Case Rep Cardiol 24. Higgins JP, Thompson SG, Deeks JJ, Altman DG. Measuring inconsis- 2015;2015:870257. tency in meta-analyses. BMJ 2003;327:557-60. 47. Cromme-Dijkhuis AH, Bogers AJ, Hess J. Isolated pulmonary vein ste- 25. Barendregt JJ, Doi SA. MetaXL user guide, version 5.3. EpiGear Interna- nosis in complex congenital heart disease, simulating cor triatriatum by tional Pty Ltd; 2016. cardiac catheterization and transoesophageal echocardiography. Eur Heart 26. Borenstein M, Hedges LV, Higgins JPT, Rothstein HR. Meta-regression. J 1995;16:287-8. In: Introduction to meta-analysis. John Wiley & Sons, Ltd; 2009. p. 187- 48. Fujii T, Tomita H, Kise H, Fujimoto K, Kobayashi K, Watanabe Y, et al. 203. An infant with primary pulmonary vein stenosis, associated with fatal oc- 27. Bini RM, Cleveland DC, Ceballos R, Bargeron LM Jr, Paciﬁco AD, Kirklin clusion of intraparenchymal small pulmonary veins. J Cardiol Cases JW. Congenital pulmonary vein stenosis. Am J Cardiol 1984;54:369-75. 2014;9:3-7. 28. Chakrabarti S, Mittal R, Gnanapragasam JP, Martin RP. Acquired ste- 49. Geggel RL, Fried R, Tuuri DT, Fyler DC, Reid LM. Congenital pulmo- nosis of normally connected pulmonary veins. Cardiol Young 2007;17:322- nary vein stenosis: structural changes in a patient with normal pulmo- 7. nary artery wedge pressure. J Am Coll Cardiol 1984;3:193-9.

44 Backes et al July 2018 ORIGINAL ARTICLES

50. Gowda S, Bhat D, Feng Z, Chang CH, Ross RD. Pulmonary vein 66. van Son JA, Danielson GK, Puga FJ, Edwards WD, Driscoll DJ. Repair stenosis with Down syndrome: a rare and frequently fatal cause of pul- of congenital and acquired pulmonary vein stenosis. Ann Thorac Surg monary hypertension in infants and children. Congenit Heart Dis 1995;60:144-50. 2014;9:E90-7. 67. Webber SA, de Souza E, Patterson MW. Pulsed wave and color Doppler 51. Jean-St-Michel E, Honjo O, Manlhiot C, McCrindle BW, Dipchand AI. findings in congenital pulmonary vein stenosis. Pediatr Cardiol Surgical approaches to pulmonary vein stenosis in pediatric heart trans- 1992;13:112-5. plant recipients: opportunity for success in a difficult situation. J Heart 68. Balasubramanian S, Rehman M, Gauvreau K, Jenkins KJ. Bilateral disease Lung Transplant 2016;35:1135-7. and early age at presentation are associated with shorter survival in pa- 52. Khositseth A, Siripornpitak S, Laohakunakorn P. Hypoplastic right lung tients with congenital heart disease and intraluminal pulmonary vein ste- associated with right pulmonary vein stenosis and systemic collateral. nosis. Congenit Heart Dis 2012;7:378-86. Congenit Heart Dis 2010;5:76-80. 69. Yun TJ, Coles JG, Konstantinov IE, Al-Radi OO, Wald RM, Guerra V, et al. 53. Lai YC, Wu MH, Hou SH, Wang JK, Lue HC. Unilateral pulmonary vein Conventional and sutureless techniques for management of the pulmo- stenosis with atrial septal defect: report of one case. Zhonghua Min Guo nary veins: evolution of indications from postrepair pulmonary vein ste- Xiao Er Ke Yi Xue Hui Za Zhi 1994;35:319-24. nosis to primary pulmonary vein anomalies. J Thorac Cardiovasc Surg 54. Li Y, An Q, Zhang E. Successful correction of unroofed coronary 2005;129:167-74. sinus with pulmonary vein stenosis. Eur J Cardiothorac Surg 2012;42:e17- 70. Ha JW, Chung N, Yoon J, Jang Y, Kim BO, Cho SY, et al. Pulsed wave 8. and color Doppler echocardiography and cardiac catheterization find- 55. Massaro AN, Kanter JP, Scavo L, Short BL. Pulmonary vein stenosis di- ings in bilateral pulmonary vein stenosis. J Am Soc Echocardiogr agnosed after failure to wean from extracorporeal membrane oxygen- 1998;11:393-6. ation. Pediatr Cardiol 2008;29:238-40. 71. Balasubramanian S, Marshall AC, Gauvreau K, Peng LF, Nugent AW, Lock 56. Mendelsohn AM, Bove EL, Lupinetti FM, Crowley DC, Lloyd TR, Fedderly JE, et al. Outcomes after stent implantation for the treatment of con- RT, et al. Intraoperative and percutaneous stenting of congenital pul- genital and postoperative pulmonary vein stenosis in children. Circ monary artery and vein stenosis. Circulation 1993;88:II210-7. Cardiovasc Interv 2012;5:109-17. 57. Moura C, Moreira J, Silva JS, Monterroso J. Obstruction to the pulmo- 72. Peng LF, Lock JE, Nugent AW, Jenkins KJ, McElhinney DB. Compari- nary venous flow in an infant. Rev Port Cardiol 2003;22:841-3. son of conventional and cutting balloon angioplasty for congenital and 58. Mueller GC, Dodge-Khatami A, Weil J. First experience with a new drug- postoperative pulmonary vein stenosis in infants and young children. Cath- eluting balloon for the treatment of congenital pulmonary vein stenosis eter Cardiovasc Interv 2010;75:1084-90. in a neonate. Cardiol Young 2010;20:455-8. 73. Azakie A, Lavrsen MJ, Johnson NC, Sapru A. Early outcomes of primary 59. Mühler E, Engelhardt W, Grabitz RG, Messmer BJ, von Bernuth G. [Con- sutureless repair of the pulmonary veins. Ann Thorac Surg 2011;92:666- genital pulmonary vein stenosis as a rare cause of pulmonary hyperten- 71. discussion 671-662. sion]. Klin Padiatr 1991;203:137-40. 74. Rehman M, Jenkins KJ, Juraszek AL, Connor JA, Gauvreau K, Muneeb 60. Pacifico AD, Mandke NV, McGrath LB, Colvin EV, Bini RM, Bargeron M, et al. A prospective phase II trial of vinblastine and methotrexate in LM Jr. Repair of congenital pulmonary venous stenosis with living multivessel intraluminal pulmonary vein stenosis in infants and chil- autologous atrial tissue. J Thorac Cardiovasc Surg 1985;89:604- dren. Congenit Heart Dis 2011;6:608-23. 9. 75. Dragulescu A, Ghez O, Quilici J, Fraisse A. Paclitaxel drug-eluting stent 61. Park SC, Neches WH, Lenox CC, Zuberbuhler JR, Siewers RD, Bahnson placement for pulmonary vein stenosis as a bridge to heart-lung trans- HT. Diagnosis and surgical treatment of bilateral pulmonary vein ste- plantation. Pediatr Cardiol 2009;30:1169-71. nosis. J Thorac Cardiovasc Surg 1974;67:755-61. 76. Zhu J, Ide H, Fu YY, Teichert AM, Kato H, Weisel RD, et al. Losartan ame- 62. Sade RM, Freed MD, Matthews EC, Castaneda AR. Stenosis of indi- liorates “upstream” pulmonary vein vasculopathy in a piglet model of pul- vidual pulmonary veins. Review of the literature and report of a surgi- monary vein stenosis. J Thorac Cardiovasc Surg 2014;148:2550-7. cal case. J Thorac Cardiovasc Surg 1974;67:953-62. 77. Quinonez LG, Gauvreau K, Borisuk M, Ireland C, Marshall AM, Mayer 63. Seale AN, Daubeney PE, Magee AG, Rigby ML. Pulmonary vein steno- JE, et al. Outcomes of surgery for young children with multivessel pul- sis: initial experience with cutting balloon angioplasty. Heart 2006;92:815- monary vein stenosis. J Thorac Cardiovasc Surg 2015;150:911-7. 20. 78. Vanderlaan RD, Caldarone CA. PVS Network. 2017. http://www. 64. Smith SC, Rabah R. Pulmonary venous stenosis in a premature infant with pvsnetwork.org. Accessed November 1, 2017. bronchopulmonary dysplasia: clinical and autopsy findings of these newly 79. Riley RD, Lambert PC, Abo-Zaid G. Meta-analysis of individual partici- associated entities. Pediatr Dev Pathol 2012;15:160-4. pant data: rationale, conduct, and reporting. BMJ 2010;340:c221. 65. Urban S, Kaulitz R, Kallfelz HC, von der Hardt H, Muller KM, Poets CF. 80. Schmucker CM, Blumle A, Schell LK, Schwarzer G, Oeller P, Cabrera L, Respiratorische Insuffizienz im Sauglingsalter aufgrund kongenitaler et al. Systematic review finds that study data not published in full text Lungenvenenstenosen mit Lungenvenenhypoplastie. Monatsschr articles have unclear impact on meta-analyses results in medical re- Kinderheilkd 1996;144:263-6. search. PLoS ONE 2017;12:e0176210.

Pulmonary Vein Stenosis in Infants: A Systematic Review, Meta-Analysis, and Meta-Regression 45 THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 198

Figure 1. Flow diagram for article selection. Flow diagram showing the process for identiﬁcation and selection of articles for inclusion. ASD, atrial septal defect; VSD, ventricular septal defect.

Figure 4. Meta-regression analysis. Meta-regression plot for % 3- and 4-vein involvement.

45.e1 Backes et al July 2018 ORIGINAL ARTICLES

Table I. Search strategies and terms used PubMed/Medline “pulmonary vein stenosis”* “pulmonary veno-occlusive disease”* “pulmonary vein atresia”* “pulmonary vein” AND “stenosis” “pulmonary vein” AND “atresia” “pulmonary vein stenosis” AND “congenital” “pulmonary vein stenosis” AND “infant” “pulmonary vein stenosis” AND “newborn” “pulmonary veno-occlusive disease” AND “congenital” “pulmonary veno-occlusive disease” AND “infant” “pulmonary veno-occlusive disease” AND “newborn” “pulmonary vein atresia” AND “congenital” “pulmonary vein atresia” AND “infant” “pulmonary vein atresia” AND “newborn” SCOPUS/Web of Science “pulmonary vein stenosis” combined with “congenital” “pulmonary vein stenosis” combined with “infant” “pulmonary vein stenosis” combined with “newborn” “pulmonary vein stenosis” combined with “children” “pulmonary veno-occlusive disease” combined with “congenital” “pulmonary veno-occlusive disease” combined with “infant” “pulmonary veno-occlusive disease” combined with “newborn” “pulmonary veno-occlusive disease” combined with “children” “pulmonary vein atresia” combined with “congenital” “pulmonary vein atresia” combined with “infant” “pulmonary vein atresia” combined with “newborn” “pulmonary vein atresia” combined with “children”

*The singular terms “pulmonary vein stenosis,” “pulmonary vein atresia,” and “pulmonary veno-occlusive disease” were searched with limitations: humans-only, infants (0-1 year of age), and children (0-18 years of age).

Table II. Outcomes of interest Infant characteristics • Sex • Premature birth (<37 wk of gestation) • IUGR status • Chronic lung disease • Age at diagnosis (<6 mo) • Concomitant congenital heart disease (left-to-right shunt, critical heart disease)19 • Genetic anomalies, including Trisomy 21 Diagnostic characteristics • Primary imaging modality used in the initial diagnosis (echocardiography, catheterization, computed topography, magnetic resonance imaging, or any combination thereof) • Definition of PVS • Cardiac hemodynamics • Pulmonary vein characteristics at diagnosis: aggregate number of involved veins, location of stenosis (right upper, right lower, left upper, left lower), unilateral vs bilateral, and right-sided vs left-sided Treatment characteristics • Type of treatment: surgery, catheter-based, transplantation • Nature of surgery: resection, patch angioplasty, stent implantation • Extent of surgery, including number of veins included at initial intervention • Autopsy findings

IUGR, intrauterine growth restriction.

Pulmonary Vein Stenosis in Infants: A Systematic Review, Meta-Analysis, and Meta-Regression 45.e2 THE JOURNAL OF PEDIATRICS • www.jpeds.com Volume 198

Table V. Pooled mortality prevalence based upon number of veins involved at presentation and laterality Meta-analysis pooled mortality prevalence Prevalence Lower CI Upper CI No of cases No of studies I2 LFK index 1-Vein involvement 20.6% 8.2% 36.4% 25 16 0.0 1.70 2-Vein involvement 45.7% 30.1% 61.8% 46 20 21.7 0.60 3-Vein involvement 77.1% 62.0% 89.4% 28 19 0.0 −0.98 4-Vein involvement 81.1% 69.4% 90.6% 44 20 0.0 −0.49 1- or 2-Vein involvement 37.5% 26.4% 49.2% 72 30 10.4 0.60 3- or 4-Vein involvement 79.9% 70.7% 87.7% 72 29 0.0 −0.49 Unilateral involvement 28.0% 16.0% 41.5% 50 27 18.1 3.26 Bilateral involvement 76.2% 67.5% 83.9% 89 32 0.0 −0.64

Table VI. Outcomes among infants undergoing reintervention Authors Year Interventions Outcome Time to death* Chakrabarati, et al28 2007 Venoplasty, venoplasty Expired – Holt, et al1 2007 TVBD + stent, TVBD, lung transplant Alive N/A TVBD, Venoplasty Expired 2.5 mo TVBD + stent, lung transplant Expired 108 mo Mendeloff, et al36 1995 TVBD, lung transplant Alive N/A TVBD + stent, lung transplant Alive N/A Bonello, et al45 2015 Multiple TVBD with or without stent, heart-lung transplant Alive N/A Bravo-Valenzuela, et al46 2015 Marsupialization, TVBD Expired 4 mo Geggel, et al49 1984 TVBD, TVBD, venoplasty Alive N/A Lai, et al53 1994 TVBD, venoplasty Alive N/A Mueller, et al,58 † 2010 TVBD, TVBD, TVBD Expired 1.3 mo

–, timing of death not explicitly stated; N/A, not applicable; TVBD, transvenous balloon dilation. *Time to death is described as time (mo) following presentation. †Paclitaxel drug-eluting balloon used.

Table VII. Primary interventions by epoch Epoch 1 (N = 78) Epoch 2 (N = 83) Primary interventions* (1974-2007) (2008-2016) P value Vascular surgery Sutureless repair 8 (10) 32 (39) <.01 Other surgeries† 27 (35) 8 (10) Percutaneous/catheterization Balloon dilation only (no stent) 10 (13) 4 (5) >.99 Balloon dilation + stent implantation 8 (10) 3 (4) Transplant or lobectomy Heart and/or lung transplant 6 (8) 1 (1) .08 Lobectomy 0 (0) 2 (2) Nonintervention 19 (24) 33 (40) .11

Data presented as n (% of column total). *Data among infants with well-defined interventions; some infants may have undergone more than one type of procedure as primary intervention (ie, surgical resection on 1 affected vein and venoplasty on another). †Resection, patch venoplasty, stent implantation, dilation of ostia.

45.e3 Backes et al