Clinicopathologic and Molecular Characteristics of Resected Thoracic Mass Lesions in the Pediatric Population: A 25-Year Institutional Experience From a Tertiary Care Center

The study protocol was approved by the Institutional Review Board at Mayo Clinic Rochester, Minnesota. Cases were retrieved from the surgical pathology archives of a tertiary general hospital after a retrospective search of all resections from patients aged ≤21 years at the time of surgery during a 25-year period (1993–2018). Site keywords searched included lung, bronchus, trachea, mediastinum, pleura, and chest. Tumors outside the thoracic cavity (eg skin, subcutaneous, and extrathoracic soft tissue neoplasms) were not included. Small biopsies (transbronchial biopsies, imaging guided biopsies, etc.) were also not included, as the focus of this study was pediatric thoracic resection specimens likely to be encountered by the general surgical pathologist. Diagnostic hematoxylin-eosin (H&E)–stained slides were reviewed for nonmetastatic lesions, and the diagnoses were confirmed by a thoracic pathologist. Clinical information was obtained from the electronic medical record. Descriptive statistics were employed to analyze the data where appropriate.

For primary neoplasms, formalin-fixed paraffin-embedded (FFPE) tissue specimens collected within the last 10 years were considered for molecular testing. A representative FFPE tissue block was selected for cases that had sufficient tissue for analysis. Tumor tissue was scraped from unstained FFPE slides and used for DNA and RNA extraction. The extracted nucleic acid was tested using a lab-developed comprehensive genomic profiling assay utilizing the Illumina TruSight Oncology 500 (TSO500) high-throughput, hybrid-capture next-generation sequencing chemistry (San Diego, California). The assay is comprised of a DNA and RNA subpanel. The DNA subpanel detects tumor mutational burden (TMB) and microsatellite instability (MSI) status, and identifies single-nucleotide variants (SNVs) and small deletion-insertions (delins) in 514 genes, and gene-level amplification in 59 genes. The RNA subpanel detects fusions involving 55 genes and splice variants in EGFR, MET, and AR. Sequencing was performed on the Illumina NovaSeq 6000 instrument (San Diego, California) and the resulting data were processed through the TSO500 v2.1 Local App.

TMB status is determined by the number of somatic mutations per megabase (mut/Mb) in the assessed coding regions, having >50× coverage, within the 1.94-Mb DNA panel. TMB values <10 mut/Mb are categorized as TMB-Low and those ≥10 mut/Mb as TMB-High. MSI status is determined by the percentage of unstable microsatellite sites, and a status of microsatellite stable is assigned to samples with <20% unstable sites and MSI-High with ≥20% unstable sites. SNVs and delins are detected at variant allele frequency of 2%. Gene-level amplification is detected at 2.2×. Fusions are detected at ≥3 supporting reads and splice variants at ≥10 supporting reads. An in-house–developed visualization tool developed using data from a TSO500 intermediate file with fold change values for all the 514 genes assessed for small variants in the DNA subpanel was used to determine chromosomal gains and losses. Qiagen Clinical Insights Interpret-One (QCI-II, Hilden, Germany) was used for variant classification. Pathogenic and likely pathogenic results were highlighted in this study regardless of Association for Molecular Pathology/American Society of Clinical Oncology/College of American Pathologists tier.

During the 25-year study period, 385 surgical resection specimens for intrathoracic space-occupying mass lesions were identified in the surgical pathology archives, corresponding to 373 individual pediatric patients (12 patients had 2 lesions resected). Overall, 176 resection cases corresponded to nonneoplastic mass lesions, cysts, or malformations; 169 to metastatic neoplasms; and 40 to primary neoplasms. The ratio of thoracic nonneoplastic mass lesions, cysts, or malformations to primary thoracic tumors was 5.9:1, and the ratio of metastatic thoracic tumors to primary thoracic tumors was 5.6:1. The 12 patients that had 2 surgical resections included 8 patients with a metastatic neoplasm and a separate resection of a nonneoplastic lesion, 2 patients with primary thoracic neoplasms and separate metastatic lesions of nonthoracic origin, 1 patient with a primary thoracic neoplasm and a separate resection of a nonneoplastic condition, and 1 patient with 2 resections harboring nonneoplastic lesions.

Nonneoplastic mass lesions and congenital space occupying mass lesions (ie cysts and lung malformations) were the most common type of resected lesions encountered, representing 176 of 385 cases (46%; Table 1). These lesions were a cyst or malformation in 58 of 176 patients (33%; median age 8 months, range 1 day–21 years, interquartile range [IQR]: 3.42 years), most commonly CPAMs or bronchial atresia and pulmonary sequestrations (Table 1; Figure 1, A through D). As expected, congenital malformations and cystic lesions were resected in younger patients when compared to patients with neoplasms and other nonneoplastic mass lesions and these differences were statistically significant (Mann-Whitney U test, P < .001, and P < .001, respectively).

The remaining 118 of 176 nonneoplastic mass lesions (67%) had various pathologic diagnoses (Table 1; Figure 2, A through F), most commonly necrotizing granulomas, intrapulmonary lymph nodes, and organizing pneumonia. Median age of these patients was 1 year (range 3 months–21 years). Most of these nonneoplastic mass lesions (n = 64) were detected in patients being monitored for a prior history of malignancy. Infectious etiologies were confirmed in many nonneoplastic mass lesions (n = 43), most commonly Histoplasma spp., mycobacteria, and invasive aspergillosis. Specific etiological infectious organisms and their corresponding histopathologic patterns are described in detail in Supplemental Tables 1 and 2 (see supplemental digital content, containing 3 tables at https://meridian.allenpress.com/aplm in the November 2024 table of contents.).

Metastatic neoplasms were the second most common type of mass lesion encountered, and comprised 169 of the 385 lesions (43.9%; Table 2). Median age at resection was 16 years (range 1.4–21 years). There was no statistically significant difference between the age at diagnosis of primary pulmonary neoplasms versus metastatic neoplasms (P = .39). The most common metastatic neoplasm was osteosarcoma, followed by Ewing sarcoma (Table 2).

An additional 18 of 385 resected lesions (4.7%) likely represented treated metastatic lesions with complete response to chemotherapy (Table 1). Median age at resection was 17.5 years (range 2–20 years). A wide array of nonneoplastic findings were present consistent with therapy effect or tumor bed, including necrosis (9 of 18, 50%), fibrosis with histiocytic reaction (n = 4, 22%), nodular ossification/calcification (n = 3, 17%), and fibroinflammatory nodules with organizing hemorrhage (n = 2, 11%).

Primary thoracic neoplasms were the most uncommon type of mass lesion occurring in this series, observed in 40 of 373 patients (11%). Primary neoplasms occurred in 20 males and 20 females, ranging from 6 months to 21 years of age (median 17 years; range 6 months–21 years). The clinicopathologic features of the studied cases are detailed in Table 3.

Primary extrapulmonary neoplasms occurred in 16 of 373 patients (4%; Table 3). Median age was 16.5 years (range 3–21 years). Presenting symptoms included back pain (n = 8), chest pain (n = 6), and wheezing (n = 2). In 2 patients with neural tumors, the neoplasms were incidentally discovered on imaging for trauma or upper respiratory infection. Eleven of 16 (69%) primary extrapulmonary neoplasms were malignant (Table 3), most commonly representing sarcomas (6 chest wall Ewing sarcomas, 1 mediastinal monophasic synovial sarcoma) and mediastinal germ cell tumors (1 seminoma, 1 yolk sac tumor).

Primary lung neoplasms occurred in 24 of 373 patients (6%; Table 3 and Figure 3, A through F). Median age was 17.6 years (range 6 months–21 years). Presenting symptoms included obstructive pneumonia (n = 9), cough (n = 8), wheezing (n = 4), chest pain (n = 4), and Cushing syndrome (n = 3 carcinoid tumor patients). Five of the 24 lung neoplasms (21%) were incidentally discovered on imaging procedures performed for other reasons, including 3 patients being monitored for a prior history of a malignancy, and 2 patients being monitored in the setting of hereditary neoplastic syndromes.

Most primary lung neoplasms were malignant (14 of 24; 58%). Carcinoid tumors were the most common primary lung neoplasms (10 of 24; 42%). Three of the 10 carcinoid tumors were atypical (30%), and most were endobronchial (8 of 10; 80%). Small (<1 cm) adenocarcinomas occurred in 2 patients (Table 3), both of which arose in the setting of prior malignancy (chondroblastic osteosarcoma and metastatic appendiceal well-differentiated neuroendocrine tumor). The patient with osteosarcoma had received systemic chemotherapy, while the patient with metastatic neuroendocrine tumor had not. In one patient the adenocarcinoma was a minimally invasive adenocarcinoma with a lepidic component spanning 8 mm and 3 mm of invasion. In the second patient, the adenocarcinoma was a 4-mm moderately differentiated (papillary predominant) invasive tumor which demonstrated spread through alveolar spaces (Figure 4, A through C). Neither of the patients with adenocarcinoma had a known predisposing genetic syndrome. One other primary neoplasm occurred in the setting of prior malignancy, which was a pulmonary hamartoma found on follow-up imaging in a patient with Wilms tumor. Syndromic associations included a case of paraganglioma which occurred in a patient with familial paraganglioma syndrome, and a myofibroma that occurred in a patient with generalized familial infantile myofibromatosis. Other primary neoplasms included 3 cases of inflammatory myofibroblastic tumor (2 endobronchial), 2 pleuropulmonary blastomas, 2 cases of pulmonary Langerhans cell histiocytosis (PLCH), and 1 case each of a sclerosing pneumocytoma and an endobronchial squamous papilloma. The 2 cases of PLCH occurred in adolescents with a history of cigarette smoking.

Overall, clinical follow-up information was available for most patients (38 of 40; 95%) with primary thoracic neoplasms, with a median postresection clinical follow-up interval of 70.2 months (range 1–387 months, IQR: 78.5 months; Table 3). No patients with primary pulmonary neoplasms died of disease during the follow-up period. Most patients with primary lung neoplasms were alive without evidence of disease at last clinical follow-up (20 of 23, 87%; median follow-up 70.8 months, range 1–273 months, IQR: 58.4 months). The 2 patients with PLCH had evidence of disease at last clinical follow-up (42 and 100 months after diagnosis), and 1 patient with typical carcinoid tumor died of an unrelated cause 4 months after diagnosis. Most patients with primary extrapulmonary thoracic neoplasms were alive without evidence of disease at last clinical follow-up (12 of 15; 80%; median follow-up 81.6 months, range 6.5–388 months, IQR: 172.7 months). Three of 15 patients (20%) died of disease at a median of 14 months (range 13–32 months). The patients that died had Ewing sarcoma (2) and synovial sarcoma (1).

Eleven of the primary tumors had adequate tissue available to be evaluated by next-generation sequencing, which included 6 carcinoid tumors (2 atypical), and 1 each of paraganglioma, schwannoma, inflammatory myofibroblastic tumor, synovial sarcoma, and pleuropulmonary blastoma. Detailed molecular results are summarized in Supplemental Table 3. All tumors were microsatellite stable and had low tumor mutational burden (range 0–2.4 mut/Mb). No gene-level amplifications were identified. Three carcinoid tumors (2 typical, 1 atypical) had copy number alterations (CNAs) including loss of chromosomes 13, 21, 22, and 15q, and gains of 7 and 8. One typical carcinoid tumor had ARID1A c.6747dupA p.E2250fs*28, and 1 atypical carcinoid tumor had DNMT1 c.445 + 1G>A. The carcinoid tumors with CNAs and those with mutations were mutually exclusive. One typical carcinoid tumor had no genetic abnormalities detected. The pleuropulmonary blastoma showed DICER1 c.5126A>G p.D1709G. The inflammatory myofibroblastic tumor showed no definitively pathogenic mutations. The paraganglioma demonstrated mutations in SDHD c.167delA p.H56fs*30 and VHL c.598C>T p.R200W. The schwannoma had ATM c.7865C>G p.A2622G, which is a variant of unknown significance. In addition to the SS18::SSX1 fusion detected clinically by a separate reverse transcriptase polymerase chain reaction test performed at the time of diagnosis, the synovial sarcoma showed CNAs including losses of chromosomes 3, 11q, 22, and X. No other gene fusions or splice variants were detected in any tumor type, but fusion testing failed in a case of atypical carcinoid and inflammatory myofibroblastic tumor.

Thoracic space-occupying mass lesions resected from children at our institution for a 25-year period were unlikely to be primary thoracic neoplasms, as metastatic tumors and nonneoplastic lesions, congenital malformations, or cysts far exceeded the number of primary neoplasms, both occurring more than 5 times more commonly. These data are similar to prior cohorts.^1–3  Metastases and nonneoplastic lesions, congenital malformations, or cysts were resected at almost equal numbers. The metastatic lesions most commonly represented childhood sarcomas. Of the 251 mass lesions resected from 242 children with a history of malignancy in our cohort that were being monitored for recurrence/metastasis, 169 (67%) represented viable metastasis, 18 (7%) represented likely treated metastasis with complete pathologic response, and 64 (25%) represented alternative nonneoplastic etiologies. Therefore, it seems that the risk of pulmonary mass lesions representing metastasis in this clinical setting is high. The nonneoplastic lesions, congenital malformations, or cysts most commonly represented CPAMs or bronchial atresia, sequestrations, granulomas, intrapulmonary lymph nodes, and various types of inflammatory or reactive changes. Importantly, infectious agents were confirmed in about one-third of nonneoplastic mass lesions that were not classified as malformations. Some of these were agents that require urgent initiation of antimicrobial therapy, which can be prolonged, can lead to serious side effects, and may require the use of multi-drug regimens (eg, mycobacteria). Therefore, appropriate testing for infectious etiologies should always be considered for pediatric patients with fibroinflammatory mass lesions.

Primary lung neoplasms represented 24 of 40 (60%) of the resected primary thoracic neoplasms in this cohort. Carcinoid tumor and inflammatory myofibroblastic tumor were the most common primary pulmonary tumors in our series, while Ewing sarcoma was the most common extrapulmonary primary neoplasm. Most primary lung tumors were of low-grade malignancy or benign, but 2 examples of nonmucinous pulmonary adenocarcinoma were observed. Interestingly, neither were associated with preexisting CPAM, and both occurred in the setting of prior malignancy. Cases of non-CPAM–associated nonmucinous lung adenocarcinoma arising in never-smoking pediatric patients with other malignancies have been reported by other authors, and similar to our 2 cases, they have been minimally invasive adenocarcinomas or well-differentiated to moderately differentiated adenocarcinomas.^6,8,9  While systemic chemotherapy is known to predispose pediatric patients to a variety of secondary malignancies,^10–12  including lung cancer, this risk factor was only present in 1 of the 2 patients in our study. Neither of these patients had a known genetic syndrome, but it is possible that an unknown genetic predisposition to neoplasia could be present in one or both patients, given the presence of 2 separate primary malignancies. The apparent increased risk of lung adenocarcinoma could also be due to discovery of incidental indolent tumors via increased surveillance.⁶  Both patients in our study were alive without evidence of recurrent lung adenocarcinoma at last follow-up. Prior series of pulmonary nonmucinous adenocarcinoma–like lesions arising in pediatric cancer patients have likewise shown a good outcome.^6,13  Therefore, it is likely that these tumors are relatively indolent. Unfortunately, the 2 lung adenocarcinomas in our pediatric cohort had inadequate material for molecular testing to determine any similarities or differences to similar appearing tumors occurring in adults, either due to low tumor volume or the age of the archived specimen.

Overall, although most of the primary thoracic neoplasms in our cohort were malignant, the patients had a good clinical outcome with death from disease occurring in only 3 of 40 patients (7.5%). No patients with primary lung neoplasms died from their disease. All 3 deaths observed in our cohort occurred in patients with extrapulmonary aggressive thoracic sarcomas (Ewing sarcoma and synovial sarcoma), in which some clinicopathologic characteristics have been associated with poor overall survival in prior cohorts (development of metastases, tumor size, and staging group).^14,15  All 3 patients who died from disease in our cohort developed metastatic disease after the initial diagnosis.

Based on our limited cohort of genetic data, primary thoracic neoplasms in pediatric patients seem to have relatively simple genetics, including low tumor mutational burden and microsatellite stability, and frequent presence of one or more tumor-specific pathogenic or likely pathogenic mutations and CNAs. One prior study has shown a significant increase in tumor mutational burden associated with increased age.¹⁶  However, our results are quite limited by low case numbers, and further studies are needed to compare the genetic characteristics between pediatric thoracic neoplasms and their counterparts occurring in adults.

While some pediatric carcinoid tumors showed mutations in genes acting in chromatin-remodeling pathways, similar to those reported in adults,^17,18  others did not harbor pathogenic or likely pathogenic SNVs or delins mutations, and instead demonstrated CNA. Interestingly, none of our pediatric carcinoid tumors showed loss of 11q (including the MEN1 gene locus), which was observed in more than half of adult pulmonary carcinoid tumors in one series.¹⁹  That series did observe recurrent losses of chromosome 13 in adult carcinoid tumors, which were also present in 2 cases in our cohort, but they did not observe any losses of 21 and 22, which were present in some of our pediatric carcinoid tumors.¹⁹  As expected, in 1 case of paraganglioma and 1 case of pleuropulmonary blastoma, we identified mutations in potentially hereditary tumor susceptibility genes (SDHD and DICER1) that have been previously associated with these neoplasms.^20,21  Interestingly, the paraganglioma also harbored a mutation in the von Hippel-Lindau (VHL) tumor suppressor gene. VHL gene alterations have been reported in around 11% of paragangliomas in adults.²⁰  However, to our knowledge, the missense mutation seen in this case has not been previously reported in paragangliomas but has been reported in clear cell renal cell carcinoma.^22,23 ATM mutations (missense and nonsense) are known to be common in schwannomas,²⁴  although the specific mutation observed in our study does not seem to have been reported in that tumor type to date and is considered a variant of unknown significance.

This study is limited by its retrospective nature. In addition, while the total number of space-occupying mass lesions in this series is high, the number of primary neoplasms remains quite low. This small cohort size is an additional study limitation, as is the low number of samples adequate for advanced molecular testing. Unfortunately, as primary neoplasms are infrequently found in this population,²⁵  old archived specimens unsuitable for molecular profiling are a common occurrence. In addition, our study focused solely on resected lesions that are likely to be encountered in the surgical pathology laboratory. Therefore, this should not be interpreted as a comprehensive collection of all pediatric neoplasms that occur in the chest, as our study design clearly introduces selection bias against tumors that were not treated with surgical resection (ie lymphoproliferative disorders, etc.), as well as against nonneoplastic conditions that are biopsied and do not undergo resection.

In summary, this study provides further clinicopathologic data to support the rarity and histologic diversity of primary thoracic neoplasms in children, and arms surgical pathologists with a differential diagnosis of entities that they are likely to encounter when a thoracic resection from a pediatric patient is received. While most pediatric thoracic neoplasms have a good outcome, deaths may occur, especially in the setting of extrapulmonary aggressive thoracic sarcomas. In this limited study, the genetic characteristics of primary pulmonary thoracic neoplasms were relatively simple and generally fit the profile of mutations that have been described in the literature for the specific tumor types observed. While a subset of pediatric carcinoid tumors show mutations similar to those seen in adults, others show CNAs that are somewhat different than those that have been described in adults.