Clinicopathological characteristics and molecular analysis of lung cancer associated with ciliated muconodular papillary tumor/bronchiolar adenoma

Ciliated muconodular papillary tumor/bronchiolar adenoma (CMPT/BA) is a recently introduced benign lung tumor. It remains unclear whether CMPT/BA is associated with a specific type of lung cancer (LC). We studied the clinicopathological characteristics and genetic profiles of the coexisting primary LC and CMPT/BA (LCCM) cases. We identified eight LCCM (0.4%) from the resected Stage 0–III primary LC (n = 1945). The LCCM cohort was male‐dominant (n = 8), elderly (median 72 years old), and most were smokers (n = 6). In addition to the adenocarcinoma (n = 8), we detected two squamous cell carcinomas and one small cell carcinoma—in some cases, multiple cancer. The target sequence/whole exome sequence (WES) revealed no shared mutations between CMPT/BA and LC. One exceptional case was invasive mucinous adenocarcinoma harboring an HRAS mutation (I46N, c.137T>A), but it was likely to be a single nucleotide polymorphism based on variant allele frequency (VAF). Other driver mutations in LC included EGFR (InDel, n = 2), BRAF(V600E) (n = 1), KRAS (n = 2), GNAS (n = 1), and TP53 (n = 2). BRAF(V600E) was the most frequent mutation in CMPT/BA (60%). In contrast, LC showed no specific trend in driver gene mutations. In conclusion, our study revealed differences in the gene mutation profiles of CMPT/BA and LC in coexisting cases, suggesting mostly independent clonal tumorigenesis of CMPT/BA from LC.


INTRODUCTION
Ciliated muconodular papillary tumor (CMPT) is a new disease concept of the lung that was first reported by Ishikawa et al. 1 in 2002. CMPT is characterized by papillary proliferation of ciliated cells, mucous cells, and basal cells, forming a mucus-rich nodular lesion in the lung. 2 Recent genetic studies have revealed that CMPT has a high frequency of driver mutations, including BRAF and EGFR, 3,4 supporting neoplasmic rather than a reactive change. 5 In addition, all reported CMPT cases, mainly from Asia, have been successfully treated without recurrence by mostly reduction surgery. 2,6 Chang et al. proposed a new concept of bronchiolar adenoma (BA), which includes CMPT but expands to a nonclassical form often found in the peripheral lung. Nonclassical lesions show flattened (nonpapillary) proliferation of a bland bilayered bronchiolar-type epithelium containing a continuous layer of basal cells. Ciliated muconodular papillary tumor/bronchiolar adenoma (CMPT/BA) was adopted as a benign tumor in the 5th edition of the World Health Organization (WHO) classification. 7,8 However, CMPT/BA has two aspects of pathological findings. Ciliated cells and mild nuclear atypia suggest that it is benign; however, micropapillary or papillary proliferation patterns are consistent with those of lung adenocarcinoma. 9,10 CMPT/BA, especially in frozen specimens, is more likely to be misdiagnosed as mucus-rich invasive mucinous adenocarcinoma (IMA). 11,12 On CT images, CMPT/BA occurs as a nodule in the peripheral site of the lung, with a gland glass shadow as mucus around the solid nodule. 13 The finding of a part-solid gland glass nodule does not exclude the possibility of lung adenocarcinoma. In addition, although FDG-PET uptake in CMPT/BA is relatively low, early stage lung cancer (LC) also has low uptake; therefore, it is difficult to differentiate between the two nodules. 14 Recently, several case reports have suggested a potential malignant transformation of CMPT/BA [15][16][17] and the role of KRAS, HRAS, and AKT mutations was discussed. Thus, CMPT/BA has several similarities to LC, suggesting its potential as a precancerous lesion. The relationship between CMPT/BA and LC has not yet been studied, especially using comparative genomic profiling. Therefore, we collected coexisting cases of CMPT/BA and primary LC and compared the genetic features of each lesion by target sequencing and whole exome sequencing (WES) to explore their relationship.

Cases and clinicopathological review
We collected eight primary LC cases with the coexistence of CMPT/BA from the pathological record files at the National Cancer Center Hospital after approval by the institutional review board (IRB No. 2015-289). The subjects were 1945 patients with resected LC, excluding Stage IV, and 25 patients with resected CMPT/BA between January 2015 and December 2018. All CMPT/BA lesions were morphologically reviewed and confirmed by researchers, including an experienced pathologist (M. H. and N. M.). In challenging cases, we performed immunohistochemical staining of BRAF V600E (to identify the cilia and the possibility of a gene mutation) and p40/p63 (to identify the basal cells). Their clinicopathological background, including age at diagnosis, sex, smoking history, histological types based on the 5th edition of the WHO classification of thoracic tumors, and pathological stages based on the 8th edition of the Union for International Cancer Control (UICC) for International Cancer Control TNM staging, was retrospectively collected from medical charts.
Target sequencing of mutation profiles in CMPT/BA and LC Formalin-fixed paraffin-embedded (FFPE) specimens of CMPT/BA and LC (10 µm thick, serial five sections) were macroscopically dissected from the CMPT/BA and cancer areas separately, then proceeded to the following sequence after deparaffinization and dehydration ( Figure 1). Serial hematoxylin-eosin-stained slides were reviewed before and after sectioning, and the quantity and quality of the tumor tissue were confirmed before genetic testing. According to the manufacturer's instructions, DNA was isolated using the MagMAX FFPE DNA/RNA Ultra Kit (Applied Biosystems). Ion Torrent Libraries were used for the Ion AmpliSeq Cancer Hotspot Panel V2 (Thermo Fisher Scientific) for approximately 2800 COSMIC mutations from 50 oncogenes and tumor suppressor genes. This analysis was performed using a sequencing assay. The Cancer Hotspot Panel V2 assay detects LUNG CANCER ASSOCIATED WITH CMPT/BA mutations in selected hotspot regions of 50 genes. Extracted DNA was quantified using a Qubit dsDNA HS Assay Kit (Thermo Fisher Scientific). Libraries were constructed in 21 PCR cycles using the Ion AmpliSeq Cancer Hotspot Research Panel v2: Chef Ready Kit (Thermo Fisher Scientific) on Ion Chef instruments (Thermo Fisher Scientific) according to the manufacturer's recommendations. The pooled libraries prepared by Ion Chef were diluted to an equimolar concentration of 50 pmol for template preparation. Template preparation, enrichment of beads containing the template, and chip loading of the templated beads were performed using Ion Chef equipment and Ion 510 and Ion 520 chips (Thermo Fisher Scientific) according to the manufacturer's instructions. The Ion530 Kit -Chef (Thermo Fisher Scientific) was used according to the manufacturer's instructions. Sequencing was performed using an Ion S5XL system (Thermo Fisher Scientific).

Whole exome sequencing
Snap-frozen cancerous lung tissues suitable for genome analysis of six LC specimens from eight patients were available from the National Cancer Center Biobank. WES was performed for these six patients. Exome sequencing was conducted using 2.5 μg of cancerous DNA isolated from snap-frozen tissues using the SureSelect Human All Exon exome capture kit (version 4 or 5; Agilent Technologies Japan) on the Illumina HiSeq. 2000 platform (Illumina). The reads were aligned against the reference human genome (UCSC, human genome 19; Hg19) using the Burrows-Wheeler Aligner Multi-Vision software package. As duplicate reads were generated during the PCR amplification process, duplicated paired-end reads aligned to the same genomic positions were marked using Picard. Somatic SNVs were called using the Mutect program, which applies a Bayesian classifier to allow the detection of somatic mutations with a low allele frequency. Somatic indel F I G U R E 1 Schematic procedure of target sequence determination. Formalin-fixed paraffin-embedded sections were dissected from lung cancer (11 lesions) and CMPT/BA (eight legions) tissues, and 50 cancer-related gene mutations were examined by NGS target sequencing (Ion AmpliSeq Hot Spot Panel v2). mutations were identified using the GATK somatic indel detector (BROAD Institute; https://www.broadinstitute. org/gatk/). Similarly, the observation of the variants was supported by visual examination using the IGV.

Clinicopathological characteristics
We identified eight patients with primary LC and concurrent CMPT/BA. As shown in Table 1, the patients' age ranged from 66 to 81 years at the time of surgery, with a median age of 72. Six patients were male and six were smokers. Three patients had synchronous multiple LCs; consequently, 11 primary LC lesions were included in this cohort. The most frequent histological type of LC was adenocarcinoma (n = 8), including the papillary, lepidic, acinar, and micropapillary subtypes, and one IMA. There were two squamous cell carcinomas and one small cell carcinoma ( Figure 2). Most CMPT/BA lesions were diagnosed incidentally, except for two cases detected with an additional tumor on CT and FDG-PET (Supporting Information: Figure S1).
Radiological differential diagnoses of these small nodules were intrapulmonary metastases or concurrent multiple LC. Two patients underwent lobectomy, two underwent segmentectomy, and one underwent wedge resection. Three patients underwent a lobectomy and wedge resection. The size of CMPT/BAs ranged from 0.3 to 1.1 cm in diameter (median, 0.5 cm). Compared to primary LC without CMPT/BA, lung cancer associated with CMPT/BA (LCCM) was more common in elderly men and heavy smokers, with a higher proportion of squamous and small cell carcinomas (Supporting Information: Table S1). The pathological stages did not show a consistent trend.

Targeted sequencing of mutation hotspots
We compared the genetic alterations between CMPT/ BA and concurrent primary LC to determine whether CMPT/BA and LC share molecular events. The molecular results of the target sequence test are summarized in Figure 3. BRAF (V600E) was the most frequently mutated gene, detected in four CMPT/BA and one lung adenocarcinoma (Cases 1-4 and 8). KRAS (G12V) mutation was detected in one CMPT/BA and one invasive mucinous adenocarcinoma (Cases 6 and 5). The other mutation found only in CMPT/BA was AKT1 (Case 6). EGFR mutations included exon 19 deletion (E746_A750del) and exon 20 insertion (N771delinsKG) in adenocarcinoma (Cases 1 and 2). TP53 (R248W, missense, Case 3; and V157F, missense, Case 8) and GNAS mutations were found only in LC. Among the 11 LCs, only two had the same mutation (EGFR, nonframeshift deletion). In Case 5, an HRAS mutation (I46N, c.137T>A) was consistently detected in both CMPT/BA and the two LCs, with an allele frequency of over 50% for each specimen. In Case 7, TP53 mutations were shared between T A B L E 1 Characteristics of patients with lung cancer associated with CMPT/BA.

Case
Age/sex CMPT/BA and LC. However, the mutation types differed (missense mutations in LC and nonsense mutations in CMPT/BA). The driver gene mutation profile was quite different between CMPT/BA and LC (Supporting Information: Figure S2).

Whole exome sequencing
Whole exome sequencing was performed on the six LC lesions (Figure 4). BRAF mutations were not found in any patient. In Case 1, EGFR exon 19 deletions (E746_A750del) were detected, resulting in the target sequence. All the patients had TP53 mutations. In addition, KEAP1 and FBXW7 mutations were found in Case 6, and KMT2D mutations was found in Case 8.

DISCUSSION
Our first comprehensive study included all types of LC with CMPT/BA in serial primary LCs and the largest cohort of LCCM (n = 8) from Asia. We reported variations in the histological types of LC in LCCM. CMPT/BA was not only associated with adenocarcinoma but also with other histological types, including squamous cell carcinoma and small cell carcinoma (Table 1). Similar to the previous most extensive series of CMPT/BA reported by Chang et al. at Memorial Sloan Kettering Cancer Center, the most common histological type was adenocarcinoma, but not limited to one histological type, suggesting an independent pathway between CMPT/BA and LC (Table 2). 7 The incidence of primary LC was very low (0.4%). Compared with the AAH-AIS-invasive adenocarcinoma sequence (17%, NCCH data), the incidence of CMPT/BA in primary LC was low. In this study, two CMPT/BA nodules were incidentally detected by preoperative CT. In the previous study, CMPTs often show small nodules, including grand glass shadow in the peripheral lung on CT, and generally have low FDG uptake on PET. 13,14 Preoperative radiological diagnosis for additional nodules included intrapulmonary metastases or concurrent multiple LC. It is generally hard to predict the possibility of CMPT/BA on CT. Clinicopathologically, LCCM was more common in men (male:female = 6:2) and smokers (smoker:nonsmoker = 6:2). However, this result may be confounded by the sex trend of LC in Japan (2:1 maleto-female ratio). 18 The high incidence of smokers in the LCCM might be related to male dominance because the risk of cigarette smoking is higher in men. 19 The association between CMPT/BA and smoking-related LCs, such as squamous cell carcinoma and small cell  20 When diagnosing CMPT/BA in small specimens, especially frozen sections, it cannot be easily differentiated from mucinous adenocarcinoma. 21 We agree with their suggestion that mucinous adenocarcinoma might be coincident with CMPT/BA and suspect the possibility of under-recognition of this rare phenomenon. Our study had one IMA and coexistence of CMPT/BA, but the two lesions were geographically independent. Histological evaluation revealed no significant changes. The incidence of IMA is low in LC; therefore, it is expected to accumulate in similar cases.
Concerning genetic alterations, our study revealed a low frequency of shared mutations between CMPT/ BA and concurrent primary LCs. At baseline, four of the eight CMPT/BAs (50.0%) had a BRAF (V600E) mutation, which is the most common mutation in CMPT/BA. Our genetic test results for CMPT/BA are F I G U R E 4 Genetic mutation profiles of lung cancer (LC) associated with CMPT/BA detected by whole-exon sequence (WES). WES using fresh frozen samples revealed that TP53 was the most common mutation in LC with CMPT/BA, and one mutation each in EGFR, KMT2D, KEAP1, and FBXW7. No BRAF mutation was detected. Ad, adenocarcinoma; CA, carcinoma; Sm, small cell carcinoma; Sq, squamous cell carcinoma. LUNG CANCER ASSOCIATED WITH CMPT/BA consistent with those of previous reports. 5 Compared to solitary CMPT/BA series without LC, CMPT/BA with LC (including our cohort) showed slightly different genetic characteristics. (Supporting Information: Table S2) According to previous reports 3-5,7,22-31 and our case series, the incidence of BRAF/EGFR/KRAS mutations in solitary CMPT/BA was 32%/16%/14%: on the other hand, 60%/0%/20% in CMPT/BA with LC, respectively. Although the number of cases is small, the incidence of BRAF mutations tends to be higher in CMPT/BA with LC than in solitary CMPT/BA. CMPT/BA in Case 6 had KRAS and AKT1 mutation with 4.4% and 5.9% variant allele frequency (VAF), respectively. This co-mutation is comparatively rare. Liu et al. reported BRAF and AKT1 mutation for CMPT in western patients. 3 Furthermore, the co-mutation of KRAS and U2FA was reported in CMPT with LC. 7 The molecular studies of CMPT/BA reported so far have used whole sections of the lesions of interest, so there is no information on which type of cell the mutations were detected, except for the BRAF V600E mutation. In the BRAF V600E mutation case, mutant BRAF-specific immunostaining revealed positivity on both luminal and apical cells. 7,32 The same genetic mutations are likely expressed in both cells. Further studies, including in situ single-cell studies, are warranted.
T A B L E 2 Summary of CMPT/BA and primary lung cancer associated with CMPT/BA.

No Authors
Published Two of the eight patients had shared gene mutations in CMPT/BA and LC. In Case 5, the mutation shared by CMPT/BA and LC (double primary cancer, IMA, and papillary adenocarcinoma) was an HRAS mutation (position: chr11:533919, p.Ile46Asn, c.137T>A). The number of HRAS mutations in LC is relatively low. 33 HRAS mutations in CMPT/BA have also rarely been reported. 7 However, in Case 5, the allele of each mutation was 50%, which is more likely to be a germline single nucleotide polymorphism (SNP) rather than an oncogenic mutation. A similar mutation has been reported in the NCBI dsSNP database (https://www.ncbi.nlm.nih.gov/snp/rs1564789700). In Case 8, mutations in TP53 were commonly observed. The TP53 mutation found in squamous cell carcinoma was a nonsense mutation with c.1024C>T, whereas the TP53 mutation found in CMPT/BA was a missense mutation with c.799C>T. These should be considered as different types of conversions. A previous study reported that the most common mutation in CMPT/BAassociated adenocarcinoma is KRAS, with no shared mutations in CMPT/BA. 7 Compared to the EGFR mutation studies between AAH and AIS, shared mutations frequently occur in both groups, indicating an initial driving event in early tumor progression. 34 If CMPT/BA is a precursor of specific LCs, genetic alterations should be shared. Our study indicates that many LCCM cases had different molecular spectra between CMPT/BA and LC. At the very least, CMPT/ BA may not transform into a malignancy in common types of LC. However, we do not have enough cases to determine whether CMPT/BA is a potential precursor of IMA owing to its rarity. Considering the several cases of CMPT/BA-related adenocarcinoma, 15,16 the malignant potential of CMPT/BA cannot be completely ruled out. CMPT/BA with mucinous adenocarcinoma may share a KRAS mutation, as recently reported by Han et al. 17 Further studies are warranted to clarify the relationship between CMPT/BA and IMA.
The limitations of this study include the small number of cases, which may have biased the rate of genetic mutations, the fact that many of the CMPT/BA tumors were smaller than 1 cm, the amount of DNA may not have been sufficient, and the lack of analysis of normal tissues, which makes it impossible to show a germline origin. TP53 mutations were detected in Cases 1, 2, 3, 6, 7, and 8 by WES, and in Cases 1, 2, and 6 by hotspot panel testing. TP53 mutations in Cases 2 and 6 were not covered by the Ion AmpliSeq Cancer Hotspot Panel V2 (Thermo Fisher Scientific), whereas the TP53 mutation (S106R, VAF: 45%) in Case 1 was covered. This discrepancy between inspections may result from the amount of DNA extracted with FFPE in Case 1 and the quality of the sample. Differences in fixation (formalin vs freezing) 35 and mutation detection methods may have affected the results. EGFR differentiation in Case 2 had low VAF (6.7%); thus, heterogeneity as a subclonal mutation can also be considered. In addition, the DNA panel used in this study included 50 driver genes for LC and did not contain the entire spectrum of genetic mutations in CMPT/BA and LC. WES showed no BRAF mutations in LC and no common genes were found when compared with CMPT/BA, for which the target sequence was performed. WES was performed on frozen specimens, and one of its limitations is the lack of CMPT/BA specimens.
In conclusion, we report that CMPT/BA was associated with various histological types of LC, and that there was no shared genetic mutation between CMPT/BA and LC, indicating the benign nature of CMPT/BA. Further studies of IMA with CMPT/BA are warranted to further clarify their relationship.