Machine learning-based gene alteration prediction model for primary lung cancer using cytologic images.

BACKGROUND: Understanding the gene alteration status of primary lung cancers is important for determining treatment strategies, but gene testing is both time-consuming and costly, limiting its application in clinical practice. Here, potential therapeutic targets were selected by predicting gene alterations in cytologic specimens before conventional gene testing. METHODS: This was a retrospective study to develop a cytologic image-based gene alteration prediction model for primary lung cancer. Photomicroscopic images of cytology samples were collected and image patches were generated for analyses. Cancer-positive (n = 106) and cancer-negative (n = 32) samples were used to develop a neural network model for selecting cancer-positive images. Cancer-positive cases were randomly assigned to training (n = 77) and validation (n = 26) data sets. Another neural network model was developed to classify cancer images of the training data set into 4 groups: anaplastic lymphoma kinase (ALK)-fusion, epidermal growth factor receptor (EGFR), or Kirsten rat sarcoma viral oncogene homologue (KRAS) mutated groups, and other (None group), and images of the validation data set were classified. A decision algorithm to predict gene alteration for cases with 3 probability ranks was developed. RESULTS: The accuracy and precision for selecting cancer-positive patches were 0.945 and 0.991, respectively. Predictive accuracy for the EGFR and KRAS groups in the validation data set was ~0.95, whereas that for the ALK and None groups was ~0.75 and ~ 0.80, respectively. Gene status was correctly predicted in the probability rank A cases. The model extracted characteristic conventional cytologic findings in images and a novel specific feature was discovered for the EGFR group. CONCLUSIONS: A gene alteration prediction model for lung cancers by machine learning based on cytologic images was successfully developed.

Positive bag lavage cytology during thoracoscopic surgery for lung cancer is a significant predictor of locoregional recurrence.

OBJECTIVES: Advances in thoracoscopic surgery have made skin incisions smaller, but there are concerns about cancer cell contamination during sample extraction. We performed retrieval bag lavage cytology (BLC) during thoracoscopic surgery to evaluate the risk of cancer dissemination and the prognostic influence of BLC status. METHODS: BLC was investigated in 893 patients who underwent thoracoscopic lobectomy or segmentectomy for lung cancer between 2013 and 2018. The clinicopathological features and prognosis were compared between the BLC-positive and BLC-negative groups. RESULTS: Forty-nine patients (5.5%) were positive for BLC. BLC correlated with pleural invasion (49.0% vs. 12.9%, P < 0.001); however, BLC was positive in 3.3% of cases without pleural invasion. Multivariate analysis revealed that tumor size, lymph node metastasis, lymphatic and pleural invasion were predictive factors for positive BLC. Prognosis was poorer in the BLC-positive group than in the BLC-negative group (5-year overall survival, 73.6% vs. 90.2%, P < 0.001); nevertheless, positive BLC was not an independent prognostic factor. The locoregional recurrence rate was higher among BLC-positive patients than among BLC-negative patients, whereas there was no significant difference in the distant recurrence rate. Positive BLC was associated with locoregional recurrence (hazard ratio 1.87, P = 0.044) and the correlation was stronger in stage I lung cancer. There were no cases of extraction bag breakage or port-site recurrence. CONCLUSIONS: BLC positivity was correlated with the risk of locoregional recurrence in patients with surgically resected lung cancer, although it was not an independent prognostic factor. Careful manipulation is essential for extracting specimens from the thoracic cavity.

Prognosis of non-small-cell lung cancer patients with positive pleural lavage cytology.

OBJECTIVES: Positive pleural lavage cytology (PLC) is considered as a precursor condition of pleural dissemination (PD) or malignant pleural effusion (PE), and one of the poor prognostic factors in surgically resected non-small-cell lung cancer (NSCLC) patients. Although PD and PE are classified as M1a, PLC does not contribute to the tumour, node and metastasis (TNM) classification of the Union Internationale Contre le Cancer. This study aimed to evaluate the prognostic effect of positive PLC status in surgically resected NSCLC patients compared with PD and/or PE. We also aimed to consider the contribution of positive PLC status to the TNM classification. METHODS: We reviewed 1572 consecutive patients with completely resected NSCLC, and analysed the relationship between PLC status, other clinicopathological factors and prognosis. The survival rates of 45 patients with PD and/or PE were also investigated. RESULTS: Positive preresection PLC (pre-PLC) status was observed in 56 patients. Pre-PLC status was significantly associated with other clinicopathological factors. Positive pre-PLC patients exhibited a worse 5-year overall survival (50.6%) compared with negative pre-PLC patients (78.0%), but better survival than PD and/or PE patients (21.0%). Prognosis of positive pre-PLC patients was equal to that of pT3, negative pre-PLC patients; survival equality was observed when patients were stratified according to pN0, pN1 and pN2. CONCLUSIONS: Positive pre-PLC had the significant prognostic effect in surgically resected NSCLC patients. However, it is not a contraindication for surgical resection, unlike PD and/or PE. Our data suggest that positive pre-PLC should be classified as pT3 in next TNM classification.

Clinicopathological findings of non-small-cell lung cancer with high serum progastrin-releasing peptide concentrations.

Although progastrin-releasing peptide (proGRP) is used as a serum tumor marker for small cell lung cancer (SCLC), high serum pro-GRP concentrations are observed in some non-small-cell lung cancers (NSCLCs). The characteristics of these NSCLCs are not well known. To determine the clinicopathological features of NSCLC in patients with elevated serum proGRP concentrations, serum proGRP values were assessed in 654 advanced lung cancer patients, and positive (>46pg/mL) NSCLC specimens were subjected to cytological and histopathological reevaluation. Serum proGRP concentrations were positive in 34 of 421 NSCLC patients (8.1%) and 186 of 233 SCLC patients (80%). Histological subtypes of the 34 NSCLC patients at diagnosis were 20 adenocarcinomas, 5 squamous cell carcinomas, 4 large cell carcinomas, and 5 large cell neuroendocrine carcinomas. Six of 27 cytology specimens contained characteristic neuroendocrine morphology. Immunohistochemical analysis showed that 11 of 17 tumors were positive for neuroendocrine markers (64.7%). Twenty of 34 serum proGRP-positive NSCLC patients received platinum-based chemotherapy, and the response rate was 55.0%. These results suggest that serum proGRP-positive NSCLCs may have neuroendocrine differentiation. In addition, serum proGRP-positive NSCLCs may have clinical characteristics that are different from other NSCLCs.

Implications for differential diagnosis of lung cancer-associated lymphadenopathy in lymphoepithelioid cell lymphoma (Lennert's lymphoma) arising simultaneously with lung cancer: a case report.

BACKGROUND: Lymphoepithelioid cell lymphoma (LCL) is a rare morphologic variant of peripheral T-cell lymphoma, and its cytologic features have not been well characterized. We describe details from fine needle aspiration cytology (FNAC) of LCL in a patient simultaneously suffering from lung cancer, in whom extensive lymph node metastasis was suspected clinically. CASE: A 54-year-old man had a lung nodule diagnosed as an adenocarcinoma by biopsy. 18F-fluoro-deoxyglucose positron emission tomography showed high uptake in the lung nodule as well as interlobar, supraclavicular and axillary lymph nodes. FNAC from interlobar and supraclavicular lymph nodes revealed abundant lymphoid cells intermingled with epithelioid cell clusters. Most lymphoid cells were small, with teardrop-shaped nuclei. Occasionally, large lymphoid cells with hyperconvoluted nuclei and prominent nucleoli were observed. An extensive sarcoid reaction was suspected on cytology, and lobectomy was performed. LCL with lung adenocarcinoma was diagnosed on the immunohistochemical findings. CONCLUSION: Detailed observation of lymphoid cells with FNAC is important even in patients with lung cancer and massive regional lymphadenopathy. Presence ofa teardrop nuclear shape and nuclear irregularities of lymphoid cells provides important information for cytologic diagnosis of LCL when epithelioid cell clusters are evident.

Discriminant model for cytologic distinction of large cell neuroendocrine carcinoma from small cell carcinoma of the lung.

BACKGROUND: To establish cytologic criteria for pulmonary large cell neuroendocrine carcinoma (LCNEC), we developed and evaluated a discriminant model for cytologic differential diagnosis between LCNEC and small cell lung carcinoma (SCLC). METHODS: Aspiration cytologic and/or imprint smears from 29 LCNEC cases were reviewed in comparison with 26 SCLC cases. We selected the following parameters for assessment: background, cellular arrangement, cell clusters, cell cohesion, arrangements, cell dimensions areas, the presence of cytoplasm and/or prominent nucleoli, nuclear features, mitosis, naked nuclei, and nuclear streaking. To demonstrate the utility of differences in frequencies of cytologic parameters for LCNECs and SCLCs, a discriminant model was developed and evaluated. RESULTS: Among the cytologic parameters investigated, large clusters (consisting of >or=60 tumor cells) with tight cohesion and small tumor cells (showing

Association of cytologic micropapillary clusters in cytology samples with lymphatic spread in clinical stage I lung adenocarcinomas.

OBJECTIVE: Cytologic micropapillary clusters (MPCs) are round, three-dimensional and cohesive clusters of lung cancer cells with a pseudopapillary configuration. Recently, we demonstrated that MPCs from early stage lung adenocarcinomas can be considered as useful aids to assessing prognosis. We here demonstrate that stage I lung adenocarcinomas found to be positive for MPCs in preoperative cytologic approaches are high risk for lymphatic spread. METHODS: The clinicopathologic characteristics, metastatic status of dissected lymph nodes, vascular infiltration and presence of MPCs in preoperative cytologic specimens in 209 patients with clinical stage I lung adenocarcinomas undergoing complete surgical resection during 2004-2006 were reviewed. RESULTS: Thirty-eight patients (18%) had positive MPC findings; 21 patients with clinical stage IA and 17 with stage IB. Significant associations with postoperative stages IA and IB, frequent lymph node metastasis and venous infiltration on pathologic examination were observed (P<0.05). In particular, 50% of clinical stage I patients with MPCs demonstrated advance in the postoperative stage of disease. CONCLUSIONS: MPCs may be a manifestation of aggressive behavior, as evidenced by frequent lymph node metastasis, of clinical stage I lung adenocarcinomas. Preoperative cytologic detection of MPCs, therefore, should alert the surgeon to the probability of lymph node metastases.

Recurrence patterns in patients with early stage non-small cell lung cancers undergoing positive pleural lavage cytology.

BACKGROUND: Cytologic approaches such as pleural lavage cytology (PLC) are considered as possible aids to assessing prognosis of lung cancers. However, there is some controversy whether radical surgery is warranted based on the positive PLC findings with stage I non-small cell lung cancers (NSCLCs). METHODS: From January 1991 to December 2002, PLC was performed before any manipulation or resection of the lung for 853 consecutive patients who had no macroscopic pleural effusion, dissemination, or diffuse adhesions and who subsequently underwent curative resection for NSCLCs. Results of PLC with reference to clinicopathologic characteristics, adjuvant therapy, 5-year survival, and recurrence patterns were analyzed. RESULTS: PLC findings were positive in 41 patients (4.8%), rates being most frequent with adenosquamous carcinomas and adenocarcinomas. In the positive group, distant metastases (72%) and pleural recurrence (25%) (p = 0.0011) were often observed, and the survival rate was significantly poorer (p < 0.002), even for patients with stage I disease (p = 0.009). As adjuvant therapies in the positive group after resection, 6 patients received hypotonic cisplatin and 15 received a distilled water infusion into the pleural space. Although only 2 patients had pleural recurrence, these therapies did not improve long-term outcome. CONCLUSIONS: PLC is a distinct prognostic factor for early stage lung carcinomas. Thus, we suggest that cytologic examination of PLC should be routine, even for patients with stage I NSCLCs before beginning lung resection. Moreover, curative resection, followed by adjuvant systemic therapy, could be necessary for improvement of outcome.

Cytology of pulmonary adenocarcinomas showing enteric differentiation.

OBJECTIVE: To evaluate cytologic differences between primary pulmonary adenocarcinomas showing enteric differentiation (PAED) primary pulmonary adenocarcinomas of ordinary structure (PAC) and pulmonary metastases from colorectal carcinomas (MCR). STUDY DESIGN: During an 18-year period (1986-2003), cytologic materials were obtained from 5 PAEDs confirmed by pathologic examination of surgically resected specimens at the Cancer Institute Hospital. Aspiration cytologic and/or imprint smears of routine samples stained by the Papanicolaou method from the PAED cases were reviewed in comparison with 10 cases each of PAC showing a tubular pattern and MCR. RESULTS: The aspiration biopsy cytology and imprints showed similar features. Abundant necrotic debris in the background was recognized in a majority of all cases independent of the group. None to slight overlapping of tumor cells and less frequent palisading or glandular arrangements were characteristic features of PAED, significantly differing from MCR. Moreover, differences in nuclear features were evident as follows: in the MCR group, nuclear chromatin was hyperchromatic and coarsely condensed, and there were prominent nucleoli, whereas a slightly hyperchromatic pattern with some small to enlarged nucleoli was typical of PAED and PAC cases. CONCLUSION: Although diagnosis of PAED by routine cytology is difficult due to the features of the lesion, differential diagnosis between PAED, PAC and MCR is a possible using conventional cytologic criteria.

Micropapillary clusters in early-stage lung adenocarcinomas: a distinct cytologic sign of significantly poor prognosis.

BACKGROUND: It is known that patients who have pulmonary adenocarcinomas with a pathologic micropapillary pattern (pMPP) featuring small papillary tufts that lack a central fibrovascular core have a poor prognosis. Although the pMPP initially was identified in surgical material, preoperative detection is desirable from the standpoint of making treatment decisions. Therefore, the authors focused on cytologic features resembling the pMPP in lung adenocarcinomas, with particular reference to the survival of patients with pathologic Stage I disease. METHODS: The authors reviewed clinical course data, preoperative cytologic specimens, and histologic materials from 110 patients with Stage I adenocarcinoma of the lung who presented between 1986 and 1995. Cytology of micropapillary clusters (MPCs) was characterized by round, 3-dimensional, cohesive clusters of neoplastic cells (consisting of > 3 cells and < 20 cells) with a pseudopapillary configuration. Total counts of cohesive clusters that consisted of more than three neoplastic cells on slides and frequencies of MPCs were investigated. RESULTS: All patients (54 females and 56 males) had a preoperative diagnosis of malignancy and underwent complete surgical resection. The patients with Stage I disease were subclassified into an MPC-positive group (n = 41) and an MPC-negative group (n = 69). The 5-year survival rate was 91.3% for patients in the MPC-negative group and 75.6% for patients in the MPC-positive group; this difference was statistically significant. CONCLUSIONS: MPC cytology is a distinct prognostic marker for early-stage lung adenocarcinoma with poor prognosis. The presence of this component, therefore, should alert the clinician to the need for close follow-up.