Genotyping and cytomorphological subtyping of lung adenocarcinoma based on liquid‐based cytology

Artificial Intelligence Recognition Model Using Liquid-Based Cytology Images to Discriminate Malignancy and Histological Types of Non-Small-Cell Lung Cancer

INTRODUCTION

Artificial intelligence image recognition has applications in clinical practice. The purpose of this study was to develop an automated image classification model for lung cancer cytology using a deep learning convolutional neural network (DCNN).

METHODS

Liquid-based cytology samples from 8 normal parenchymal (N), 22 adenocarcinoma (ADC), and 15 squamous cell carcinoma (SQCC) surgical specimens were prepared, and 45 Papanicolaou-stained slides were scanned using whole-slide imaging. The final dataset of 9,141 patches consisted of 2,737 N, 4,756 ADC, and 1,648 SQCC samples. Densenet-121 was used as the DCNN to classify N versus malignant (ADC+SQCC) and ADC versus SQCC images. AdamW optimizer and 5-fold cross-validation were used in the training.

RESULTS

For malignancy prediction, the sensitivity, specificity, and accuracy were 0.97, 0.85, and 0.94, respectively, in the patch-level classification, and 0.92, 0.88, and 0.91, respectively, in the case-level classification. For SQCC prediction, the sensitivity, specificity, and accuracy were 0.86, 0.91, and 0.90, respectively, in the patch-level classification and 0.73, 0.82, and 0.78, respectively, in the case-level classification.

CONCLUSION

The DCNN model performed excellently in predicting malignancy and histological types of lung cancer. This model may be useful for predicting cytopathological diagnosis in clinical situations by reinforcing training.

Cytological characteristics of histological types of lung cancer by cytomorphometric and flow cytometric analyses using liquid-based cytology materials

BACKGROUND

Distinguishing the histological types of lung cancer is essential for determining treatment strategies in clinical practice. In this study, cytomorphological characteristics and proliferative activities were compared among histological types of lung cancer by cytomorphometric and flow cytometric analyses using liquid-based cytology (LBC) samples.

METHODS

Scraped LBC samples from 73 surgically resected specimens were collected between August 2018 and November 2019. Papanicolaou-stained and paired Ki-67-stained slides were used for cytomorphometric analyses. Another sample for each case was analyzed using a flow cytometric system (LC-1000). The cell proliferation index (CPIx) was calculated to evaluate proliferative activity.

RESULTS

In total, 73 cases, including cases of adenocarcinoma (n = 53), squamous cell carcinoma (n = 14), small cell carcinoma (n = 1), large cell neuroendocrine carcinoma (NEC; n = 3), and pleomorphic carcinoma (n = 2) were evaluated. Small cell carcinoma and large cell NEC were categorized into a single group, NEC. The adenocarcinoma group tended to have a larger nuclear area and longer perimeter than other histological types. The NEC group had a considerably higher Ki-67 labeling index and significantly higher CPIx than other histological types (p = .030). A significant positive correlation was observed between the Ki-67 labeling index and CPIx for all cases (r = 0.362, p = .002).

CONCLUSION

The Ki-67 labeling index and flow cytometric analyses focus on proliferative activity for the distinction of histological types of lung cancer, thereby guiding clinical decision-making.

Guidelines for Handling of Cytological Specimens in Cancer Genomic Medicine

Rapid advances are being made in cancer drug therapy. Since molecularly targeted therapy has been introduced, personalized medicine is being practiced, pathological tissue from malignant tumors obtained during routine practice is frequently used for genomic testing. Whereas cytological specimens fixed mainly in alcohol are considered to be more advantageous in terms of preservation of the nucleic acid quality and quantity. This article is aimed to share the information for the proper handling of cytological specimens in practice for genomic medicine based on the findings established in "Guidelines for Handling of Cytological Specimens in Cancer Genomic Medicine (in Japanese)" published by the Japanese Society of Clinical Cytology in 2021. The three-part practical guidelines are based on empirical data analyses; Part 1 describes general remarks on the use of cytological specimens in cancer genomic medicine, then Part 2 describes proper handling of cytological specimens, and Part 3 describes the empirical data related to handling of cytological specimens. The guidelines indicated proper handling of specimens in each fixation, preparation, and evaluation.

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.

Cytomorphometric and flow cytometric analyses using liquid-based cytology materials in subtypes of lung adenocarcinoma

BACKGROUND

The histological classifications of invasive lung adenocarcinoma subtypes are considered to predict patient prognosis after surgical treatment. The objectives of this study were to evaluate cytomorphological characteristics and proliferative activities among the histological predominant patterns by performing cytomorphometric and flow cytometric analyses using liquid-based cytology materials.

METHODS

Cytological samples fixed by liquid-based cytology preservatives from 53 surgically-resected lung adenocarcinoma specimens were obtained between August 2018 and November 2019. The Papanicolaou-stained and paired Ki-67-stained slides were analyzed for calculating nuclear morphology (nuclear area, nuclear perimeter and nuclear circularity) and Ki-67 labeling index using software. The cell proliferation index (CPIx) was calculated and cellular information including cell cycle stage of tumor cells was obtained by flow cytometry.

RESULTS

The 53 cases included papillary (n = 29), acinar (n = 8), lepidic (n = 5), and solid (n = 4) subtypes, and invasive mucinous adenocarcinoma (n = 7) were also included. In the lepidic pattern, nuclear area (79.6 ± 28.8 μm² ) and perimeter (34.1 ± 6.1 μm) were relatively larger and longer than those of the other predominant patterns. The Ki-67 labeling index of the solid pattern (27.9 ± 12.5%) was highest compared with those of other predominant patterns. There were statistically significant differences in the lepidic versus solid patterns and the papillary versus solid patterns (p = .013 and p = .039, respectively). The calculated mean CPIx of the lepidic and the acinar patterns were approximately two-fold higher than those of the other predominant patterns.

CONCLUSION

By revealing the differences of cytomorphological characteristics, these methodologies might be used for diagnosing cytopathological materials using digital cytopathology.

Special Staining of the Liquid-Based Cytopathology Test in Bronchoalveolar Lavage Fluid for Diagnosis of Invasive Pulmonary Aspergillosis with Nonneutropenic Patients

In recent years, various biomarkers have been gradually applied on bronchoalveolar lavage (BAL) fluid for the diagnosis of invasive pulmonary aspergillosis (IPA). The objective of this study is to assess the value of the liquid-based cytopathology test (LCT) for improving the identification of IPA in BAL fluid from possible IPA patients, following special staining with periodic acid-Schiff staining (PAS) or Grocott's methenamine silver (GMS). A total of 47 consecutive possible IPA patients who underwent bronchoscopy with BAL fluid from January 2017 to December 2018 were included. 45 people had a pair of BAL fluid specimens and 2 patients had two BAL fluid specimens. The 49 pairs of BAL fluid specimens were processed for culture, tuberculosis acid fast staining smear, direct microbial smear, and LCT with special staining (PAS and GMS), respectively. Then, we compared the sensitivity and specificity of PAS and GMS in BAL fluid in high-risk patients. Among 47 possible IPA patients, 25 patients had proven/probable IPA, and 11 patients had other invasive fungal diseases. The sensitivity of GMS was higher than that of PAS (92.11% versus 81.58%; P = 0.175). The specificity of GMS was 81.82%, which was higher than that of PAS (81.82% versus 72.73%; P = 0.611). The negative predictive value (NPV) for PAS and GMS were 53.33% and 75.00%, respectively. The positive predictive value (PPV) for PAS and GMS were 91.18% and 94.59%, respectively. This study showed that special staining of LCT in BAL fluid may be a novel method for the diagnosis of IPA, and the GMS of LCT had higher sensitivity and specificity, which was superior to PAS.

Subtyping and EGFR mutation testing from blocks of cytological materials, based on liquid-based cytology for lung cancer at bronchoscopic examinations

BACKGROUND

Liquid-based cytology (LBC) allows immunohistochemistry (IHC), fluorescence in situ hybridization, and molecular testing to be performed in fixed cell materials. We examined the feasibility of subtyping and EGFR mutation testing of bronchoscopic samples from patients with lung cancer using cell blocks (CB) based on LBC fixation (LBC-CB).

METHODS

We included 35 consecutive patients with peripheral lung nodules who underwent endobronchial ultrasonography with a guide sheath in our hospital. Thirty of these patients were diagnosed with lung cancer by obtaining cytological samples. Cytological subtyping was performed with IHC using LBC-CB, and the Cobas EGFR Mutation Test ver. 2 was performed using extracted genomic DNA from the LBC-CB, formalin-fixed paraffin-embedded (FFPE) tissue, and matched plasma.

RESULTS

Of the 30 cases, 25 were classified cytomorphologically as adenocarcinoma (ADC, n = 17) and squamous-cell carcinoma (SQCC, n = 8). The remaining five cases were classified by IHC as favor ADC (n = 3) and favor SQCC (n = 2) according to the WHO criteria. In the final ADC group (n = 20), EGFR mutations on the LBC-CB were identified in eight cases (40%; 1 exon 19 deletion, 6 L858R, and 1 L861Q). Mutations in FFPE samples were identified in seven cases (35%) at the same site in each case. Plasma EGFR mutations were identified in four cases (20%) at the same site. The CB detection rate was higher than for FFPE and plasma.

CONCLUSION

LBC-CB is suitable for subtyping and EGFR mutation testing in lung cancers.