Stepwise addition of genetic changes correlated with histological change from “well-differentiated” to “sarcomatoid” phenotypes: a case report

The patient was a 54-year-old man. He complained of hemosputum, and chest X-ray showed a mass shadow in the right upper lung field (Fig. 1a). He had a history of smoking 20 cigarettes per day for 30 years. Chest computed tomography revealed an irregularly-shaped mass in segment 1 of the right lung (Fig. 1b). Bronchoscopic biopsy led to a diagnosis of adenocarcinoma.

Under a diagnosis of lung cancer (cT2aN0M0), he underwent right upper lobectomy and lymph node dissection. Bilateral legs had been swelling preoperatively, which subsided spontaneously after the surgery. Pleural lavage cytology led to a diagnosis of disseminated adenocarcinoma (pT2aN0M1a, stage IV); histopathological analysis showed that tumor was composed of three different histological areas; portion A: well-differentiated adenocarcinoma, portion B: poorly differentiated adenocarcinoma, portion C: sarcomatoid cancer, and portion B was a major component (Fig. 1c-g). As the sarcomatoid cancer component accounted for less than 10% of the entire tumor area, the cancer did not meet the diagnostic criteria for pleomorphic carcinoma. To examine whether different histological components resulted from genetic divergence, tumor cells of each different histological area were collected from formalin-fixed paraffin-embedded tissues by using laser capture microdissection (Fig. 1d-g). A panel targeting the exon of 53 lung cancer-associated genes was established in-house, as we previously reported [1‐4]. Subsequently targeted sequencing of those 53 lung cancer-related genes was performed. The sequencing data were processed using standard Ion Torrent Suite software running on a Torrent Server (Thermo Fisher Scientific). Variants calling were performed using an Ion Reporter Server System (Thermo Fisher Scientific), and peripheral blood DNA was used as a control to detect somatic mutations in tumours using filtration of “confident somatic variants” in a Tumour-Normal pair pipeline [4, 5]. This study was approved by the institutional review board, and the patient provided written informed consent. In total, there were 26 mutations found in the cancer (Fig. 2a and Additional file 1: Table S1), and mutations with an allele fraction ≥30% were determined as significant mutations (Table 1). Common missense mutations in TP53 and KRAS genes were detected in the portion A, B and C (Table 1, Fig. 2a). These mutations were validated by Sanger sequencing (Fig. 2b). The positions and patterns of the amino acid changes and base-pair substitutions in those co-mutations were consistent (Table 1).

For further inference of phylogenies and estimation of evolutionary distances, the Neighbor-Joining method was implemented to cluster the nonsilent mutations and the phylogenetic tree was constructed (Fig. 3a and b) [6]. The ‘ape’ and ‘phangon’R (3.2.3 on linux) packages were utilized for these analyses. As a result, sarcomatoid component clusters together with poorly differentiated adenocarcinoma and that cluster segregates away from well-differentiated adenocarcinoma (Fig. 3a). From the phylogenetic tree, it can be inferred, under the assumption of normal cells without any mutations being roots, that well-differentiated adenocarcinoma first branched off from the roots, and that another branch evolved through poorly differentiated adenocarcinoma into sarcomatoid carcinoma (Fig. 3b). According to the branch length, which indicates the evolutionary distance, the distance between the poorly differentiated adenocarcinoma and the sarcomatoid carcinoma is relatively short.

Postoperative adjuvant chemotherapy was performed, but the patient developed lung and brain metastases and died of cancer 5 months postoperatively.