Pleuropulmonary Blastoma: A Report on 350 Central Pathology–Confirmed Pleuropulmonary Blastoma Cases by the International Pleuropulmonary Blastoma Registry

Abstract

Background

Pleuropulmonary blastoma (PPB) has 3 subtypes on a tumor progression pathway ranging from type I (cystic) to type II (cystic/solid) and type III (completely solid). A germline mutation in DICER1 is the genetic cause in the majority of PPB cases.

Methods

Patients confirmed to have PPB by central pathology review were included, and their clinical characteristics and outcomes were reported. Germline DICER1 mutations were sought with Sanger sequencing.

Results

There were 435 cases, and a central review confirmed 350 cases to be PPB; 85 cases (20%) were another entity. Thirty-three percent of the 350 PPB cases were type I or type I regressed (type Ir), 35% were type II, and 32% were type III or type II/III. The median ages at diagnosis for type I, type II, and type III patients were 8, 35, and 41 months, respectively. The 5-year overall survival (OS) rate for type I/Ir patients was 91%; all deaths in this group were due to progression to type II or III. OS was significantly better for type II versus type III (P=.0061); the 5-year OS rates were 71% and 53%, respectively. Disease-free survival (DFS) was also significantly better for type II versus type III (P=.0002); the 5-year DFS rates were 59% and 37%, respectively. The PPB type was the strongest predictor of outcome. Metastatic disease at the diagnosis of types II and III was also an independent unfavorable prognostic factor. Sixty-six percent of the 97 patients tested had a heterozygous germline DICER1 mutation. In this subset, the DICER1 germline mutation status was not related to the outcome.

Conclusion

Cystic type I/Ir PPB has a better prognosis than type II, and type II has a better outcome than type III. Surveillance of DICER1 carriers may allow the earlier detection of cystic PPB before its progression to type II or III PPB and thereby improve outcomes.

Introduction

Pleuropulmonary blastoma (PPB) is the most common primary malignancy of the lung in childhood.¹ It was initially proposed to be a distinct entity in 1988 a dysembryonic and nosologic equivalent to neuroblastoma and other organ-based solid malignancies of early childhood.^2,3 Established in 1988, the International Pleuropulmonary Blastoma Registry (IPPBR) has collected and centrally reviewed pathology from patients with PPB. Priest et al⁴ described PPB as the sentinel disease of a distinctive hereditary syndrome (Online Mendelian Inheritance in Man #601200). In 2009, Hill et al⁵ identified heterozygous germline mutations in DICER1 as the first known genetic cause for this syndrome.

Three pathologic types or stages in the evolution of PPB have been defined: type I or purely cystic PPB, type II or cystic/solid PPB, and type III or purely solid PPB. The progression of type I to types II and III is well documented.^6–8 Not all cystic type I PPBs are destined to progress to the more malignant types. These “nonprogressed/regressed” cystic cases are designated as type I regressed (type Ir).^6,8 The clinical course of smaller numbers of PPB patients has been described previously.^6,7,9–14 The current larger report presents data from 350 PPB cases, which allowed a statistically robust analysis of survival and prognostic factors for PPB. Central review proved critical to this effort because 20% of the cases were not PPB. In addition, a comparison of our demographics with the Surveillance, Epidemiology, and End Results (SEER) program suggests that the IPPBR captures a large fraction of the total pool of cases. Finally, our study is also the first to evaluate the role of germline DICER1 mutations in the clinical course of PPB.

Materials and Methods

The IPPBR is a collaboration of Children’s Hospitals and Clinics of Minnesota, the Washington University Medical Center (St. Louis, Mo), and the Children’s National Medical Center (Washington, District of Columbia). Registry activities were approved by the institutional review board at each institution. The study is registered at ClinicalTrials.gov (NCT01464606).

PPB cases were included if the central pathology review by one of the study pathologists (D.A.H. and L.P.D.) confirmed PPB. PPB cases included in this report were diagnosed from 1962 to 2012. Data were abstracted from medical records obtained by the IPPBR after participant-informed consent. Surgical, chemotherapy, and radiation decisions were made by local treating physicians. The age at diagnosis was defined as the age at the initial diagnostic surgical procedure. Ages at progression, recurrence, and/or new metastasis were defined as the ages at the first confirmation of each event. The largest diameter of the cyst or mass, whether unilateral, bilateral, or multifocal, was abstracted from medical records or available imaging studies at diagnosis. Race, ethnicity, and achievement of local control are not reported because of incomplete information in the medical and surgical records. Regimens were recorded, but because they varied substantially, this report does not attempt to evaluate their relative efficacy. Verification of disease and survival status was obtained from the local treating institution or from the patient or patient’s family on an annual basis.

Type I PPB is defined as a cystic lesion whose interface with the adjacent lung parenchyma is generally abrupt from normal-appearing distal airspaces or alveoli to cysts formed by more or less delicate septa. Within the septa, a layer of small immature cells with or without rhabdomyoblastic differentiation resides beneath the low cuboidal epithelial cells; the immature cells with a cambium layer–like appearance are present either as a continuous ribbon of subepithelial cells or as discontinuous foci. Microscopic thickening or expansion of the septa by foci of embryonal rhabdomyosarcoma (ERMS) or spindle cell or fibrosarcoma-like areas is also considered within the spectrum of type I PPB. The other pattern of type I PPB is characterized by the same cystic architecture, but the septa are completely devoid of primitive small cells and/or rhabdomyoblasts. Purely cystic tumors that lack a primitive cell component are classified as type Ir PPB, which signifies regression or nonprogression. Type II PPB and type III PPB are similar in their histologic features within the solid areas, which consist of a collage of primitive sarcomatous patterns, including ERMS, spindle cell or fibrosarcoma-like areas, blastemal islands surrounded by primitive mesenchyme, cartilaginous nodules with fetal to sarcomatous features, and anaplastic cells with large, bizarre nuclei and atypical mitotic figures. Anaplasia is defined according to the Wilms tumor criteria of marked nuclear enlargement, hyperchromatism, and atypical, multipolar mitotic figures.¹⁵ The anaplastic cells often overexpress TP53 in the nuclei. Type II PPB contains residual cystic areas, whereas type III PPB is purely solid. The solid components of type II PPB and type III PPB are indistinguishable from each other in small biopsies, and this can prevent the differentiation of type II from type III. In some of the current cases, a large subsequent specimen or radiographic information allowed classification as type II or type III. Type II/III cases are those cases for which the precise classification of type II or III is not possible.

DICER1 Mutation: Sanger Sequencing

Germline DNA was extracted from leukocytes or saliva. Sanger sequencing was performed on all coding exons and intron-exon junctions in the DICER1 gene with standard primers and protocols.⁵ Sequence traces were analyzed for mutations with standard methods.⁵ SIFT was used to assess the potential significance of predicted novel amino acid substitutions.¹⁶

Statistical Analysis

To determine the representativeness of the PPB demography in the IPPBR, we queried the SEER 18 registries for cases from 2000 to 2011 (both sexes, 0-19 years old) with International Classification of Diseases for Oncology (version 3) codes 8972/3 and 8973/3 (both PPB). SEER 18 captures 26.4% of the US population. To compare our demographics with SEER, we tallied IPPBR type I, type II, type II/III, and type III cases (type Ir cases were excluded) diagnosed in the same time interval and for the same age group referred from hospitals in the SEER 18 catchment areas.

Patients with no follow-up were excluded from the analysis of prognostic factors and outcome data. We used the nonparametric Wilcoxon rank-sum test¹⁷ to compare median values for 2 groups of IPPBR patients. We compared frequency distributions between groups with chi-square analysis. For the time-to-event analysis, we analyzed overall survival (OS) and disease-free survival (DFS), which considered events to be the diagnosis of a recurrence or metastasis of the first PPB tumor, the occurrence of a second primary PPB tumor, or death. We estimated OS and DFS with Kaplan-Meier curves. We estimated relative hazard rates with the Cox proportional hazards model and assessed statistical significance with the likelihood ratio test. We considered 16 variables as potential predictors of outcome: sex, age, date of birth, PPB type, DICER1 mutation status, primary chemotherapy, primary radiotherapy, distant metastases, laterality, focality, pleural effusion, tumor spillage, tumor size, positive hilar nodes, anaplasia, and pneumothorax at diagnosis. We used the false discovery rate¹⁸ to adjust nominal P values for multiple testing. For analyses of predictors of outcome, we combined the small number of type II/III cases with type III cases.

Results

Three hundred fifty of the 435 cases submitted to the IPPBR were centrally confirmed to be PPB cases, whereas 85 cases (20%) were not PPB (Supporting Table 1 [see online supporting information]) during the same time interval. The PPB registration rate accelerated over time and included cases from 42 countries, although most were from the United States (Supporting Figures 1 and 2 [see online supporting information]).

IPPBR Demography Versus SEER 18

In SEER 18, there were 18 males and 19 females (37 total) diagnosed with PPB from 2000 to 2011. In the same interval and catchment area, there were 13 males and 12 females (25 total) diagnosed with PPB in the IPPBR (Supporting Table 2 [see online supporting information]). Thus, in the SEER 18 catchment area, it appears that 25 of 37 PPB cases (67.6%) were captured by the IPPBR. This figure could be as high as 84% [25/(0.8 3 37)] if, similarly to the IPPBR experience, approximately 20% of SEER 18 PPB cases were another disease. In addition, the distribution of cases by age was very similar and peaked in the 1- to 4-year age group (Supporting Table 2 [see online supporting information]).

Cystic PPB Types I and Ir

Central pathology review confirmed 115 of the 350 cases (33%) to be cystic PPB; 77% of these cases were type I, and 23% were type Ir (Table 1 and Supporting Table 3 [see online supporting information]). Type Ir was noted to be distinct from type I by the IPPBR pathologists (L.P.D. and D.A.H.) only after 2000, and this was described in 2008 ⁶; this may explain the small number of type Ir cases in this report. The IPPBR pathologists reviewed and confirmed each type Ir case with the criteria described in that report.⁶ It is possible that some older type I cases would have been reclassified as type Ir upon further review. Such a review could not be completed because the pathology specimens for many earlier cases were no longer available. Males were more commonly affected than females for both type I (57% vs 43%) and type Ir (73% vs 27%). The median age was significantly lower at the diagnosis of type I versus the diagnosis of type Ir (8 vs 46.5 months, P=.0001; Table 1 and Fig. 1). Most type I cases were diagnosed during the first year of life (62%), and almost all (97%) had been diagnosed by age 3 in contrast to types Ir, II, and III (Fig. 1 and Supporting Table 4 [see online supporting information]). Prenatal ultrasound detected lung cysts in 7 cases at a gestational age of 31 to 35 weeks (Supporting Table 5 [see online supporting information]). Pneumothorax at diagnosis was found in 30% of type I and type Ir PPB cases (Table 1). Most cysts were unilateral (74%), half were unifocal, and most were larger than 5 cm (55% > 5 cm; 25% < 5 cm; 20%, size unknown; Supporting Table 3 [see online supporting information]).

Table 1.

Demographic, treatment and outcome data for cystic PPB Type-I and Type-Ir

		Type I n (%)	Type Ir n (%)	Total Cystic PPB n (%)
Total		89 (77%)	26 (23%)	115 (100%)
Age at diagnosis	median months (range)	8 (0–114)	46.5 (7–546)	12 (0–546)
Sex	female	38 (43%)	7 (27%)	45 (39%)
	male	51 (57%)	19 (73%)	70 (61%)
Pneumothorax	no	20 (22%)	4 (15%)	24 (21%)
	yes	29 (33%)	5 (19%)	34 (30%)
	unknown	40 (45%)	17 (66%)	57 (49%)
Anaplasia	yes	4 (4%)	0 (0%)	4 (3%)
	no	85 (96%)	26 (100%)	111 (97%)
DICER1a	positive	17/28 (61%)	4/6 (67%)	21/34 (62%)
	negative	11/28 (39%)	2/6 (33%)	13/34 (38%)
Treatment	surgery only	52 (58%)	23 (88%)	75 (65%)
	surgery with chemotherapy	29 (33%)	3 (12%)	32 (28%)
	surgery with unknown	8 (9%)	0 (0%)	8 (7%)
Follow-up	median months (range)	59.9 (0–477)	55.3 (0–472)	58.9 (0–477)
Recurrence or progressionb	recurrence (to Type I or Ir)	5 (6%)	1 (4%)	6 (5%)
	progression to Type II and III	9 (10%)	2 (8%)	11 (10%)
	Progression after surgery onlyc	7/52 (12%)	1 (4%)	8/75 (11%)
	Progression after surgery with chemotherapyc	2/29 (7%)	1 (33%)	3/32 (9%)
Survival	alive	84 (94%)	26 (100%)	110 (96%)
	dead	5 (6%)	0 (0%)	5 (4%)
	5-year OS % (95% CI)	89 (80-99)	100	91% (83-99)
	5-year DFS % (95% CI)	79 (69-91)	93 (80-100)	82% (73-92)

^aThe DICER1 percentage was calculated only for the evaluated patients.

^bThe median time to progression was 23 mo (range, 3-53 mo).

^cThe progression percentage was calculated for patients treated with surgery only and for patients treated with surgery and chemotherapy.

Figure 1.

Age at diagnosis. Abbreviation: PPB, pleuropulmonary blastoma.

Primary surgical resection was performed in all 115 cases of type I PPB and type Ir PPB with no perioperative mortality. In addition to surgery, 36% of type I cases and 12% of type Ir cases were treated with various postoperative chemotherapy regimens (Table 1). The DFS and OS for type I were similar to those for type Ir (Fig. 2a,b). Types I and Ir had a 5-year DFS of 82% (95% confidence interval, 73%-92%) and a 5-year OS of 91% (95% confidence interval, 83%-99%). After surgical resection, with or without chemotherapy, cystic PPB progressed to type II or III in 11 cases (10%) at a median of 23 months (range, 3-53 months) after diagnosis. Critically, all type I PPB deaths (5 cases) followed progression to type II or III (Table 1). The strategy of surgery only versus surgery plus chemotherapy resulted in similar risks of progression (11% vs 9%; Table 1); chemotherapy had no effect on OS for PPB types I and Ir (Table 2).

Figure 2.

(a,c) Overall survival and (b,d) disease-free survival since Dx. Abbreviation: Dx, diagnosis.

Table 2.

Prognostic Cox hazard model^a

	Disease-Free Survival			Overall Survival
Prognostic Factors	Type I & Ir	Type II	Type II/III & III	Type I & Ir	Type II	Type II/III & III
Upfront Chemotherapy	0.11	0.06	0.00007b	0.63	0.29	0.17
Laterality	0.12	0.41	0.94	0.13	0.96	0.12
Pleural Effusion	0.13	0.12	0.33	0.2	0.13	0.27
Gender	0.14	0.08	0.33	0.92	0.03	0.16
Tumor Spillage	0.17	0.01c	0.7	0.47	0.08	0.57
Focality	0.17	0.41	0.56	0.46	0.38	0.52
Date of Birth ≥ 2002	0.25	0.06	0.41	0.25	0.12	0.25
DICER1 Mutation	0.26	0.94	0.63	ND	0.37	0.91
Intrathoracic Nodes	0.47	0.77	0.86	0.13	0.2	0.74
Upfront Radiation	0.48	0.58	0.32	0.66	0.38	0.36
Lesion Size	0.52	0.04	0.84	0.25	0.26	0.6
Pneumothorax	0.55	0.02c	0.002b	0.32	0.04	0.01c
Age at PPB diagnosis (4 steps)	0.67	0.03	0.38	0.61	0.1	0.42
Anaplasia	0.12	0.46	0.38	0.23	0.68	0.28
Distant Metastasis	(None)	0.002d	0.0002b	(None)	0.002d	0.002d

Abbreviations: ND, not determined; PPB, pleuropulmonary blastoma.

^aCox models of prognostic factors are presented for disease-free survival and overall survival. Types I and Ir and types II/III and III are grouped together because of small numbers. Observed P values significant by the false discovery rate are flagged.

^bP <.01.

^cP <.1.

^dP <.05.

PPB Types II and III

Total 235 type II or III PPB included 53% type II, 9% type II/III, and 38% type III (Table 3 and Supporting Table 6 [see online supporting information]). Males and females were equally affected. These tumors presented at an older age than type I (Fig. 1): type II patients and type III patients were diagnosed at median ages of 35 and 41 months, respectively (Table 3). None of the patients with type III and only 1 (6-month-old) patient with type II were younger than 1 year; 95% were diagnosed by 6.8 years (Supporting Table 4 [see online supporting information]). Rare cases were found in adolescents; the oldest patient was diagnosed at the age of 36 years.¹⁹ These tumors were large (>10 cm in 123 patients and <10 cm in 73 patients), unilateral (89%), and unifocal (65%; Supporting Table 6 [see online supporting information]). Distant metastases at diagnosis were found in 20 cases (9%). The status of thoracic lymph nodes was missing in most cases (71%), but 9 cases (4%) had pathologically confirmed regional nodal metastases (Supporting Table 6 [see online supporting information]). Pleural effusion was found in 77% of the cases (85/111) with available data; pneumothorax at diagnosis was more common in cystic/solid type II patients (40/62 or 65%) versus type II/III patients and type III patients (6/30 or 20%). Various degrees of anaplasia were present in 68% of patients (Table 3).

Table 3.

Demographic, Treatment, and Outcome Data for PPB Types II and III^a

		Type II n (%)	Type II/III n (%)	Type III n (%)	Total n (%)
Total		124 (53%)	21 (9%)	90 (38%)	235 (100%)
Age at diagnosis	median months (range)	35 (6–431)	36 (18–75)	41 (19–147)	37 (6–431)
Sex	female	62 (50%)	9 (43%)	47 (52%)	118 (50%)
	male	60 (48%)	12 (57%)	43 (48%)	115 (49%)
	unknown	2 (2%)	0 (0%)	0 (0%)	2 (1%)
Metastasis at presentation	none	105 (85%)	19 (90%)	75 (83%)	199 (85%)
	present	9 (7%)	1 (5%)	10 (11%)	20 (9%)
	unknown	10 (8%)	1 (5%)	5 (6%)	16 (7%)
Anaplasia	yes	75 (60%)	11 (52%)	64 (71%)	150 (64%)
	no	49 (40%)	10 (48%)	26 (29%)	85 (36%)
DICER1b	positive	22/35 (63%)	3/4 (75%)	18/24 (75%)	43/63 (68%)
	negative	13/35 (37%)	1/4 (25%)	6/24 (25%)	20/63 (32%)
Treatment	surgery only	9 (7%)	1 (5%)	2 (2%)	12 (5%)
	surgery with chemotherapy	109 (88%)	19 (90%)	85 (94%)	213 (91%)
	surgery with unknown	6 (5%)	1 (5%)	3 (3%)	10 (4%)
	radiation	17 (14%)	3 (14%)	27 (30%)	47 (20%)
	ASCT	3 (2%)	1 (5%)	4 (4%)	8 (3%)
Follow-up	median months (range)	58.0 (0–313)	46.0 (0–373)		52.4 (0–373)
Relapse	isolated chest	5	0		5
	Isolated CNS	7	15		22
	chest with CNS	2	2		4
	distant, non-CNS	6	7		13
	total relapse	20 (16%)	24 (27%)		44 (19%)
Survival	5 years OS (95% CI)	71% (62–81%)	53% (43 – 65%)		62 (55–70%)
	5 years DFS (95% CI)	59% (50–70%)	37% (28 – 48%)		48 (42–56%)
	alive	93 (75%)	67 (60%)		160 (68%)
	dead	31 (25%)	44 (40%)		75 (32%)c
Cause of death	non-relapsed	4 (13%)	5 (11%)		9 (12%)
	relapse-related	17 (55%)	23 (52%)		40 (53%)
	unknown	10 (32%)	16 (36%)		26 (35%)

Abbreviations: ASCT, high-dose chemotherapy autologous stem cell rescue; CI, confidence interval; CNS, central nervous system; DFS, disease-free survival; OS, overall survival.

^aBecause of the small number of type II/III cases, they are combined with type III cases for both relapse and survival data.

^bThe DICER1 percentage was calculated only for the evaluated patients.

^cOne patient with limited data was excluded from the survival analysis.

^dThe causes of nonrelapse mortality were congestive heart failure (n=3), infection (n=2), respiratory failure (n=2), second cancer (glioblastoma multiforme, n=1), and thrombocytopenia-induced bleeding (n=1).

All cases had surgical resection, but whether local control was achieved was not consistently reported. Chemotherapy regimens varied over the years and mostly followed rhabdomyosarcoma or soft tissue sarcoma protocols (for a partial list, see Supporting Table 7 [see online supporting information]). High-dose chemotherapy and stem cell rescue were used as part of the primary treatment in only 8 cases (3%). Radiation therapy as part of the primary therapy was used for 20% of the type II and III patients (Table 3) and had no impact on survival (Table 2). The outcome was significantly better for type II versus types II/III and III with respect to DFS (P=.0002) and OS (P=.0061; Fig. 2c,d); the 5-year-DFS rates were 59% vs 37%, and the 5-year OS rates were 71% vs 53% (Table 3). The better outcome for type II PPB was due to fewer events, which resulted also in a longer median follow-up by 12 months, in comparison with types II/III and III, which had worse outcomes and more events (Table 3).

The relapse/progression rate was lower for type II versus types II/III and III (Fig. 3). After the exclusion of type II/III, the outcomes for type II remained significantly better than those for type III with respect to both DFS(P=4.93 × 10⁻⁵) and OS (P=.0028; Supporting Fig. 3a,b [see online supporting information]). CNS metastasis with or without local chest relapse constituted the majority of relapses (26/44 cases or 59%), and it was followed by isolated chest relapse and distant relapse (Table 3). Sixteen prognostic factors were evaluated, and only PPB type (type III was worse than type II) and the presence of distant metastatic disease at presentation were associated with a statistically significantly unfavorable prognosis (Table 2 and Fig. 4). Pneumothorax had a significant adverse affect on 1 measure only (DFS for types II/III and III), but it had no significant affect on all other outcomes; this suggests that pneumothorax may not be an important prognostic factor. Although most events occurred within the first 3 years after the diagnosis, rare late events were seen (Fig. 3). Relapse was the most common cause of death, but 9 patients (12% of all deaths) died from other causes.

Figure 3.

Relapse/progression analysis. Abbreviation: Dx, diagnosis

Figure 4.

Metastatic disease at diagnosis has significantly poor prognosis. a. Survival analysis. b. Relative hazard of metastatic disease. M0 = no metastasis at diagnosis, M1 = presence of metastasis at diagnosis, Dx = diagnosis, RH = relative hazard.

DICER1 Evaluation

Ninety-seven PPB patients were evaluated for germline DICER1 mutations, and 64 (66%) had a heterozygous, deleterious DICER1 mutation. Those harboring a DICER1 mutation were equally distributed among the 3 pathologic subtypes (Tables 1 and 3). We could not identify clinical differences between patients who had a detectable germline DICER1 mutation and patients lacking a mutation. Similarly, a germline DICER1 mutation had no prognostic impact (Table 2).

Discussion

It may be difficult for even experienced pathologists to correctly identify a lesion as PPB, so a central pathology review has helped to clarify questionable cases. It seems that for approximately 20% of cases submitted to IPPBR pathologists, the diagnosis was another entity (Supporting Table 1 [see online supporting information]), and these cases have been excluded from the analysis. This demonstrates the critical importance of central pathology review for rare tumors, and it increases the reliability of the current report of 350 centrally confirmed PPB cases. In addition, we now show that the high capture rate and the age and sex distributions of cases in the IPPBR in comparison with SEER 18 suggest that the IPPBR cohort is indeed representative of this disease population.

PPB may be associated with a unique set of disorders,⁴ and the genetic basis of the PPB familial syndrome is the heterozygous loss-of-function mutation of DICER1.⁵ In addition to PPB, this syndrome includes cystic nephroma (CN), ovarian Sertoli-Leydig cell tumors, ciliary body medulloepithelioma, nodular hyperplasia and differentiated carcinoma of the thyroid gland, pituitary blastoma, pineoblastoma, nasal chondromesenchymal hamartoma (NCMH), and ERMS.^20,21 PPB, CN, NCMH, and cervical ERMS have a unique juxtaposition of mesenchymal cells beneath an intact epithelium that can transform into high-grade sarcoma.^4,6,22–24 One of the reproducible features of the transformed mesenchymal cells in these neoplasms is the presence of specific missense mutations in the DICER1 RNase IIIb domain, which are seen in type II and III PPB, CN, NCMH, and cervical ERMS.^22,23,25,26 A loss-of-function mutation in the germline DICER1 allele with a somatic second hit results in decreased processing of the 5p mature micro-RNA (miRNA) from its precursor miRNA, whereas processing of the 3p-miRNA is unaffected, and this results in a bias toward increased 3p-miRNA.^27,28 The loss of 5pmiRNAs such as the tumor-suppressing let-7 family miRNA may lead to de-repression of oncofetal genes and uncontrolled proliferation. Our largest study of 47 PPB tumors suggests that the tumorigenic mechanism of DICER1 includes 2 hits but a not complete loss of function.²⁵ The mechanism by which 34% of PPB patients without a germline DICER1 mutation develop PPB is under investigation, but 2-hit somatic mutations have been found in some cases,²⁵ and others may have deletions, deep intronic mutations, mosaicism, or epigenetic changes. We found no clinical difference between PPB patients with or without germline DICER1 mutations; ongoing studies may clarify whether germline, somatic mutations or large deletions of DICER1 have clinical implications.

PPB type I develops at a much younger age (97% before the age of 3 years) than types II and III. Pulmonary cysts can be found prenatally at a gestational age of 31 to 35 weeks and even as early as a gestational age of 23 weeks (Supporting Table 5 [see online supporting information]). In the newborn or older infant, purely cystic type I PPB is readily mistaken for one of several congenital cystic lesions of the lungs such as congenital pulmonary airway malformation (CPAM). Such lesions are more likely to be PPB if there is a family history of PPB or associated tumors, but ultimately, a pathological examination is required.⁸ Type Ir was originally recognized in older relatives of PPB patients, but cysts with these features can be found in very young children⁶ (see online supporting information for an expanded description of type Ir). The age range of type Ir is much larger, and a lung cyst in an older individual with DICER1 or a relative of a PPB patient is most likely type Ir. The diagnostic workup requires computed tomography of the chest, and interrogation for associated diseases such as CN is recommended.29 Because metastatic disease has never been found with type I PPB, there is no need for metastatic workup. The survival of both type I PPB patients and type Ir PPB patients is 91%, but 10% may later progress to type II or III. Importantly, death after type I PPB occurred only after progression to type II or III. Whether the use of chemotherapy can prevent progression remains unresolved. This report and another from Italy⁹ failed to show a statistical benefit from chemotherapy for type I PPB in contrast to prior studies from our group.^6,7 Critically, PPB type Ir may not be benign because 2 patients progressed to type II or III PPB as late as 53 months after their diagnosis, and this indicates the need for follow-up, especially in a young child.

Type II PPB and III PPB are diagnosed at an older age than type I PPB. There were no type III patients and only 1 type II patient younger than 1 year of age, and this suggests that a solid pulmonary tumor in a child less than 12 months old is very unlikely to be PPB; other diagnoses such as fetal lung interstitial tumor (FLIT) must be considered.³⁰ Although pleural effusion can be seen in up to 76% of these cases, cytology was never used by us for diagnosis. Type II PPB and type III PPB have metastatic potential to the brain, bone, and, rarely, liver. Thus, chest computed tomography and brain magnetic resonance imaging with a possible bone scan are required.^12,13 Because solid PPB tumors can extend into the thoracic great vessels, preoperative echocardiography may be indicated.³¹ Although patients can have profound cardiorespiratory compromise with lesions involving the entire hemithorax causing mediastinal shift, most respond to early surgery and/or chemotherapy with dramatic improvement (Supporting Fig. 4 [see online supporting information]). Both type II and type III are aggressive malignancies that require chemotherapy soon after the first diagnostic surgery (Supporting Fig. 5 [see online supporting information]).^9,13 Many chemotherapy regimens were used (partial list Supporting Table 7 [see online supporting information]), and each included only small number of patients. Consequently, we did not attempt to evaluate the benefit of any specific regimen. Although doxorubicin may be an important agent for PPB types II and III,⁹ the best regimen for this disease remains unknown. This large data set of 235 cases confirmed that the PPB type was the strongest prognostic factor: outcomes were better for type II versus types II/III and III. Even after the exclusion of patients with type II/III, the outcomes for type II patients remained significantly better than those for type III patients (Supporting Fig. 3 [see online supporting information]). The biological basis for the superior outcomes of type II patients versus type III patients is under investigation at this time. Within each type, the presence of distant metastasis at diagnosis had a statistically significant detrimental effect on DFS and OS. Anaplasia was common in both type II (60%) and type III (71%) but had no prognostic effect in contrast to other childhood sarcomas. Whether the progressive loss of TP53 noted by Pugh et al²⁵ has prognostic implications is not yet known. Gross total resection for adequate local control may have important prognostic implications,^9,14 but because of inconsistent reporting, we could not verify this reliably or use it in the prognostic models. Whether upfront radiation therapy is beneficial remains unclear (Tables 2 and 3).^9,13 The data regarding high-dose chemotherapy and autologous stem cell rescue are even less clear because of the small numbers. For this group, we did not demonstrate an improvement in outcomes in the last 10 years (Table 2). To develop baseline data for future studies, in 2011, the IPPBR opened its current single-arm protocol for patients with PPB (http://ppbregistry.org/).

The limitations of this report are inherent to rare tumor research and the nature of retrospective data submitted to a registry. These limitations include a possible reporting bias, the variability of therapy strategies, and, at times, missing data. These factors limit the conclusions reported here. Treatment is a major determinant of outcome,^9,14 but this varied by case, as dictated by the treating physician, and thus we could not control for it; this limited the validity of the prognostic factors analysis. On the other hand, the use of central pathology review and ongoing yearly monitoring of these patients strengthens the validity of this registry data.

In conclusion, the outcome of purely cystic PPB (types I and Ir) is better than the outcomes of more aggressive types. Type II has better outcomes than type III, although both have high relapse and death rates. The PPB type and the presence of distant metastasis at diagnosis are the most important prognostic factors related to treatment outcomes. In a large subgroup of patients that were tested at any point in the natural history of PPB, the germline DICER1 status had no impact on prognosis. For type II and III patients, surgery and chemotherapy are critical components toward achieving a cure. Current attempts to screen DICER1 mutation carriers for cystic PPB at a young age may permit the early detection of PPB type I.³² Subsequent surgical resection may prevent the progression to types II and III with their higher morbidity and mortality.