379 unduplicated lung nodule CT images
Image segmentation aims to recognize the voxel information and external contour of the region of interest. In medical imaging, segmentation is mainly used to segment organs or lesions to quantitatively analyze relevant clinical parameters and provide further guidance for follow-up diagnosis and treatment. For example, target delineation is crucial for surgical image navigation and tumor radiotherapy guidance.
Lung segmentation plays a crucial role in medical images for lesion detection, including thorax extraction (removes artifacts) and lung extraction (identifies the left and right lungs). Several threshold techniques, such as the threshold method [ 69 ], iterative threshold [ 70 ], Otsu threshold [ 71 ], and adaptive threshold [ 72 , 73 ], have been investigated for lung segmentation. Few research groups have investigated segmentation methods based on region and 3D region growth [ 74 , 75 ]. Kass et al. [ 76 ] first introduced the active contour model, and Lan et al. [ 77 ] applied the active contour model for lung segmentation. These techniques are manual segmentation and have many disadvantages, such as being relatively slow, prone to human error, scarcity of ground truth, and class imbalance.
Several deep learning approaches have been investigated for lung segmentation. Wang et al. [ 78 ] developed a multi-view CNN (MV-CNN) for lung nodule segmentation, with an average DSC of 77.67% and an average ASD of 0.24 for the LIDC-IDRI dataset. Unlike conventional CNN, MV-CNN integrates multiple input images for lung nodule identification. However, it is difficult for MV-CNN to process 3D CT scans. Thus, a 3D CNN was developed to process volumetric patterns of cancerous nodules [ 79 ]. Sun et al. [ 80 ] designed a two-stage CAD system to segment lung nodules and FP reduction automatically. The first stage aims to identify and segment the nodules, and the second stage aims to reduce FP. The system was tested on the LIDC-IDRI dataset and evaluated by four experienced radiologists. The system obtained an average F1_score of 0.8501 for lung nodule segmentation.
In 2020, Cao et al. [ 81 ] developed a dual-branch residual network (DB-ResNet) that simultaneously captures the multi-view and multi-scale features of nodules. The proposed DB-ResNet was evaluated on the LIDC-IDRI dataset and achieved a DSC of 82.74%. Compared to trained radiologists, DB-ResNet provides a higher DSC.
In 2021, Banu et al. [ 82 ] proposed an attention-aware weight excitation U-Net (AWEU-Net) architecture in CT images for lung nodule segmentation. The architecture contains two stages: lung nodule detection based on fine-tuned Faster R-CNN and lung nodule segmentation based on the U-Net with position attention-aware weight excitation (PAWE) and channel attention-aware weight excitation (CAWE). The AWEU-Net obtained DSC of 89.79% and 90.35%, IoU of 82.34%, and 83.21% for the LUNA16 and LIDC-IDRI datasets, respectively.
Dutta [ 83 ] developed a dense recurrent residual CNN (Dense R2Unet) based on the U-Net and dense interconnections. The proposed method was tested on a lung segmentation dataset, and the results showed that the Dense R2UNet offers better segmentation performance than U-Net and ResUNet.
Table 2 shows the recently developed lung nodule segmentation techniques. Among these approaches, SVM systems obtained an accuracy range of 92.6–98.1%, CNN-based systems obtained a specificity range of 77.67–91%, ResNet models obtained a DSC range of 82.74–98.1%, and U-Net segmentation systems achieved an accuracy range of 82.2–99.27%, precision range of 46.61–98.2%, recall range of 21.43–96.33%, and F1_score range of 24.64–99.1%, respectively. The DenseNet201 system obtained an accuracy of 97%, a sensitivity of 96.2%, a specificity of 97.5%, an AUC of 0.968, and an F1_score of 96.1%. Several segmentation methods, including SVM, Dense R2UNet, 3D Attention U-Net, Dense R2UNet, Res BCDU-Net, U-Net FSL, U-Net CT, U-Net PET, U-Net PET/CT, CNN, and DenseNet201, achieved high accuracy results (over 94%).
Lung nodule segmentation approaches.
Reference | Year | Method | Imaging | Datasets | Results |
---|---|---|---|---|---|
[ ] | 2013 | Support vector machine (SVM) | CT images | Shiraz University of Medical Sciences | Accuracy: 98.1% |
[ ] | 2014 | Lung nodule segmentation | CT images | 85 patients | Accuracy: >90% |
[ ] | 2015 | SVM | CT images | 193 CT images | Accuracy: 94.67% for benign tumors; Accuracy: 96.07% for adhesion tumor |
[ ] | 2015 | Bidirectional chain coding combined with SVM | CT images | LIDC | Accuracy: 92.6% |
[ ] | 2015 | Convolutional networks (ConvNets) | CT images | 82 patients | DSC: 68% ± 10% |
[ ] | 2017 | Multi-view convolutional neural networks (MV-CNN) | CT images | LIDC-IDRI | DSC: 77.67% |
[ ] | 2017 | Two-stage CAD | CT images | LIDC-IDRI | F1-score: 85.01% |
[ ] | 2017 | 3D Slicer chest imaging platform (CIP) | CT images | LIDC | median DSC: 99% |
[ ] | 2017 | Deep computer aided detection (CAD) | CT images | LIDC-IDRI | Sensitivity: 88% |
[ ] | 2018 | 3D deep multi-task CNN | CT images | LUNA16 | DSC: 91% |
[ ] | 2018 | Improved U-Net | CT images | LUNA16 | DSC: 73.6% |
[ ] | 2018 | Incremental-multiple resolution residually connected network (MRRN) | CT images | TCIA | DSC: 74% ± 0.13 |
MSKCC | DSC: 75%±0.12 | ||||
LIDC | DSC: 68%±0.23 | ||||
[ ] | 2018 | U-Net | hematoxylin-eosin-stained slides | 712 lung cancer patients operated in Uppsala Hospital, Stanford TMA cores | Precision: 80% |
[ ] | 2019 | Mask R-CNN | CT images | LIDC-IDRI | Average precision:78% |
[ ] | 2020 | 3D-UNet | CT images | LUNA16 | DSC: 95.30% |
[ ] | 2020 | Dual-branch Residual Network (DB-ResNet) | CT images | LIDC-IDRI | DSC: 82.74% |
[ ] | 2021 | End-to-end deep learning | CT images | 1916 lung tumors in 1504 patients | Sensitivity: 93.2% |
[ ] | 2021 | 3D Attention U-Net | COVID-19 CT images | Fifth Medical Center of the PLA General Hospital | Accuracy: 94.43% |
[ ] | 2021 | Improved U-Net | CT images | LIDC-IDRI | Precision: 84.91% |
[ ] | 2021 | Attention-aware weight excitation U-Net (AWEU-Net) | CT images | LUNA16 | DSC: 89.79% |
LIDC-IDRI | DSC: 90.35% | ||||
[ ] | 2021 | Dense Recurrent Residual Convolutional Neural Network(Dense R2U CNN) | CT images | LUNA | Sensitivity: 99.4% ± 0.2% |
[ ] | 2021 | Modified U-Net in which the encoder is replaced with a pre-trained ResNet-34 network (Res BCDU-Net) | CT images | LIDC-IDRI | Accuracy: 97.58% |
[ ] | 2021 | Hybrid COVID-19 segmentation and recognition framework (HMB-HCF) | X-Ray images | COVID-19 dataset from 8 sources * | Accuracy: 99.30% |
[ ] | 2021 | Clinical image radionics DL (CIRDL) | CT Images | First Affiliated Hospital of Guangzhou Medical University | Sensitivity: 0.8763 |
[ ] | 2021 | 2D & 3D hybrid CNN | CT scans | 260 patients with lung cancer treated | Median DSC: 0.73 |
[ ] | 2022 | Few-shot learning U-Net (U-Net FSL) | PET/CT images | Lung-PET-CT-DX TCIA | Accuracy: 99.27% ± 0.03 |
U-Net CT | Accuracy: 99.08% ± 0.05 | ||||
U-Net PET | Accuracy: 98.78% ± 0.06 | ||||
U-Net PET/CT | Accuracy: 98.92% ± 0.09 | ||||
CNN | Accuracy: 98.89% ± 0.08 | ||||
Co-learning | Accuracy: 99.94% ± 0.09 | ||||
[ ] | 2022 | DenseNet201 | CT images | Seoul St. Mary’s Hospital dataset | Sensitivity: 96.2% |
COVID-19 dataset from 8 sources *: COVID-19 Radiography Database, Pneumonia (virus) vs. COVID-19 Dataset, Covid-19 X-Ray images using CNN Dataset, COVID-19 X-ray Images5 Dataset, COVID-19 Patients Lungs X-Ray Images 10,000 Dataset, COVID-19 Chest X-Ray Dataset, COVID-19 Dataset, Curated Chest X-Ray Image Dataset for COVID-19.
Lung nodule detection is challenging because its shape, texture, and size vary greatly, and some non-nodules, such as blood vessels and fibrosis, have a similar appearance to lung nodules that often appear in the lungs. The processing includes two main steps: lung nodule detection and false-positive nodule reduction. Over the past few decades, researchers worldwide have extensively investigated machine learning and deep learning-based approaches for lung nodule detection. Chang et al. [ 106 ] applied the support vector machine (SVM) for nodules classification in ultrasound images. Nithila et al. [ 107 ] developed a lung nodule detection model based on heuristic search and particle clustering algorithms for network optimization. In 2005, Zhang et al. [ 108 ] developed a discrete-time cellular neural network (DTCNN) to detect small (2–10 mm) juxtapleural and non-pleural nodules in CT images. The method obtained a sensitivity of 81.25% at 8.29 FPs per scan for juxtapleural nodule detection and a sensitivity of 83.9% at 3.47 FPs per scan for non-pleural nodule detection.
Hwang et al. [ 109 ] investigated the relationship between CT and commercial CAD to detect lung nodules. They also studied LDCT images with three reconstruction kernels (B, C, and L) from 36 human subjects. The sensitivities of 82%, 88%, and 82% for the nodules of B, C, and L were obtained for all images. Experimental results showed that CAD sensitivity could be elevated by combining data from 2 different kernels without radiation exposure. Young et al. [ 110 ] studied the effects on the performance of a CAD-based nodule detection model by reducing the CT dose. The CAD system was evaluated on the NLST dataset and obtained sensitivities of 35%, 20%, and 42.5% at the initial dose, 50% dose, and 25% dose, respectively. Tajbakhsh et al. [ 111 ] studied massive training ANN (MTANN) and CNN for lung nodule detection and classification. MTANN and CNN obtained AUCs of 0.8806 and 0.7755, respectively. MTANN performs better than CNN for lung nodule detection and classification.
Liu et al. [ 112 ] developed a cascade CNN for lung nodule detection. The transfer learning model was applied to train the network to detect nodules, and a non-nodule filter was introduced to the detection network to reduce false positives (FP). The proposed architecture effectively reduces FP in the lung nodule detection system. Li et al. [ 65 ] developed a lung nodule detection method based on a faster R-CNN network and an FP reduction model in thoracic MR images. In this study, a faster R-CNN was developed to detect lung nodules, and an FP reduction model was developed to reduce FP. The method was tested on the FAHGMU dataset and obtained a sensitivity of 85.2%, with 3.47 FP per scan. Cao et al. [ 113 ] developed a two-stage CNN (TSCNN) model for lung nodule detection. In the first stage, a U-Net based on ResDense was applied to detect lung nodules. A 3D CNN-based ensemble learning architecture was proposed in the second stage to reduce false-positive nodules. The proposed model was compared with three existing models, including 3DDP-DenseNet, 3DDP-SeResNet, and 3DMBInceptionNet.
Several 3D CNN models have been developed for lung nodule detection [ 114 , 115 , 116 ]. Perez et al. [ 117 ] developed a 3D CNN to automatically detect lung cancer and tested the model on the LIDC-IDRI dataset. The experimental results showed that the proposed method provides a recall of 99.6% and an AUC of 0.913. Vipparla et al. [ 118 ] proposed a multi-patched 3D CNN with a hybrid fusion architecture for lung nodule detection with reduced FP. The method was tested on the LUNA16 dataset and achieved a competition performance metric (CPM) of 0.931. Dutande et al. [ 119 ] developed a 2D–3D cascaded CNN architecture and compared it with existing lung nodule detection and segmentation methods. The results showed that the 2D–3D cascaded CNN architecture obtained a DCM of 0.80 for nodule segmentation and a sensitivity of 90.01% for nodule detection. Luo et al. [ 120 ] developed a 3D sphere representation-based center-point matching detection network (SCPM-Net) consisting of sphere representation and center-point matching components. The SCPM-Net was tested on the LUNA16 dataset and achieved an average sensitivity of 89.2% at 7 FPs per image for lung nodule detection. Franck et al. [ 121 ] investigated the effects on the performance of deep learning image reconstruction (DLIR) techniques on lung nodule detection in chest CT images. In this study, up to 6 artificial nodules were located within the lung phantom. Images were generated using 50% ASIR-V and DLIR with low (DL-L), medium (DL-M), and high (DL-H) strengths. No statistically significant difference was obtained between these methods ( p = 0.987, average AUC: 0.555, 0.561, 0.557, and 0.558 for ASIR-V, DL-L, DL-M, and DL-H).
Table 3 shows recently developed lung nodule detection approaches using deep learning techniques. Among these approaches, the co-learning feature fusion CNN obtained the best accuracy of 99.29%, which is higher than other lung nodule detection approaches. Several networks, including 3D Faster R-CNN with U-Net-like encoder, YOLOv2, YOLOv3, VGG-16, DTCNN-ELM, U-Net++, MIXCAPS, and ProCAN, obtained good accuracy (>90%) of lung nodule detection.
Lung nodule detection approaches.
Reference | Year | Method | Imaging | Datasets | Results |
---|---|---|---|---|---|
[ ] | 2016 | 3D CNN | CT images | LUNA16 | Sensitivity: >87% at 4 FPs/scan |
[ ] | 2016 | 2D multi-view convolutional networks (ConvNets) | CT images | LIDC-IDRI | Sensitivity: 85.4% at 1 FPs/scan, 90.1% at 4 FPs/scan |
[ ] | 2016 | Thresholding method | CT images | JSRT | Accuracy: 96% |
[ ] | 2017 | Computer aided detection (CAD) | LDCT | NLST | Mean sensitivity: 74.1% |
[ ] | 2017 | 3D CNN | LDCT | KDSB17 | Accuracy: 87.5% |
[ ] | 2017 | 3D Faster R-CNN with U-Net-like encoder | CT scans | LUNA16 | Accuracy: 81.41%; |
LIDC-IDRI | Accuracy: 90.44% | ||||
[ ] | 2018 | Single-view 2D CNN | CT scans | LUNA16 | metric score: 92.2% |
[ ] | 2018 | DetectNet | CT scans | LIDC | Sensitivity: 89% |
[ ] | 2018 | 3D CNN | CT scans | KDSB17 | Sensitivity: 87%; |
[ ] | 2018 | Novel pulmonary nodule detection algorithm (NODULe) based on 3D CNN | CT scans | LUNA16 | CPM score: 94.7% |
LIDC-IDRI | Sensitivity: 94.9% | ||||
[ ] | 2018 | Deep neural networks (DNN) | PET images | 50 lung cancer patients, & 50 patients without lung cancer | Sensitivity: 95.9% |
ultralow dose PET | Sensitivity: 91.5% | ||||
[ ] | 2018 | FissureNet | 3DCT | COPDGene | AUC: 0.98 |
U-Net | AUC: 0.963 | ||||
Hessian | AUC: 0.158 | ||||
[ ] | 2018 | DFCN-based cosegmentation (DFCN-CoSeg) | CT scans | 60 NSCLC patients | Score: 0.865 ± 0.034; |
PET images | Score: 0.853 ± 0.063; | ||||
[ ] | 2018 | Multi-scale Gradual Integration CNN (MGI-CNN) | CT scans | LUNA16, V1 dataset includes 551,065 subjects; V2 dataset includes 754,975 subjects | CPM: 0.908 for the V1 dataset, 0.942 for the V2 dataset; |
[ ] | 2018 | Deep fully CNN (DFCNet) | CT scans | LIDC-IDR | Accuracy: 84.58% |
CNN | Accuracy: 77.6% | ||||
[ ] | 2018 | Deep learning–based automatic detection algorithm (DLAD) | CT scans | Seoul National University Hospital | Sensitivity: 69.9% |
[ ] | 2018 | SVM classifier coupled with a least absolute shrinkage and selection operator (SVM-LASSO) | CT scans | LIDC-IDRI | Accuracy: 84.6% |
[ ] | 2019 | CNN | CT scans | LIDC-IDR | Sensitivity: 88% at 1.9 FPs/scan; 94.01% at 4.01 FPs/scan |
[ ] | 2019 | 3D CNN | LDCT | LUNA16 and Kaggle datasets | Average metric: 92.1% |
[ ] | 2019 | Deep learning model (DLM) based on DCNN | Chest radiographs (CXRs) | 3500 CXRs contain lung nodules & 13,711 normal CXRs | Sensitivity: 76.8% |
[ ] | 2019 | Two-Step Deep Learning | CT scans | Nagasaki University Hospital | Sensitivity of 79.6% with sizes ≤0.6 mm; Sensitivity of 75.5% with sizes ≤0.7 mm; |
[ ] | 2019 | Faster R-CNN network and false positive (FP) | CT scans | FAHGMU | Sensitivity: 85.2% |
[ ] | 2019 | YOLOv2 with Asymmetric Convolution Kernel | CT scans | LIDC-IDRI | Sensitivity: 94.25% |
[ ] | 2019 | VGG-16 network | CT scans | LIDC-IDRI | Accuracy: 92.72% |
[ ] | 2019 | Noisy U-Net (NU-Net) | CT scans | LUNA16 | Sensitivity: 97.1% |
[ ] | 2019 | CAD using a multi-scale dot nodule-enhancement filter | CT scans | LIDC | Sensitivity: 87.81% |
[ ] | 2019 | Co-Learning Feature Fusion CNN | PET-CT scans | 50 NSCLC patients | Accuracy: 99.29% |
[ ] | 2019 | Convolution networks with attention feedback (CONAF) | Chest radiographs | 430,000 CXRs | Sensitivity: 78% |
[ ] | 2019 | Recurrent attention model with annotation feedback (RAMAF) | Chest radiographs | 430,000 CXRs | Sensitivity: 74% |
[ ] | 2020 | Two-Stage CNN (TSCNN) | CT scans | LUNA16 & LIDC-IDRI | CPM: 0.911 |
[ ] | 2020 | Deep Transfer CNN and Extreme Learning Machine (DTCNN-ELM) | CT scans | LIDC-IDRI & FAH-GMU | Sensitivity: 93.69%; |
[ ] | 2020 | U-Net++ | CT scans | LIDC-IDRI | Sensitivity: 94.2% at 1 FP/scan, 96% at 2 FPs/scan |
[ ] | 2020 | MSCS-DeepLN | CT scans | LIDC-IDRI & DeepLN | |
[ ] | 2020 | Multi-scale CNN (MCNN) | CT scans | LIDC-IDRI | Accuracy: 93.7% ± 0.3 |
[ ] | 2021 | Lung Cancer Prediction CNN (LCP-CNN) | CT scans | U.S. NLST | Sensitivity: 99%; |
[ ] | 2021 | Automatic AI-powered CAD | CT scans | 150 images include 340 nodules | mean sensitivity: 82% for second-reading mode, 80% for concurrent-reading mode |
[ ] | 2021 | DNA-derived phage nose (D2pNose) using machine learning and ANN | CT scans | Pusan National University | Detection accuracy: >75%; Classification accuracy: >86% |
[ ] | 2021 | Capsule network-based mixture of experts (MIXCAPS) | CT scans | LIDC-IDRI | Sensitivity: 89.5%; |
[ ] | 2021 | CNN with attention mechanism | CT scans | LUNA16 | Specificity: 98.9% |
[ ] | 2021 | Deep learning image reconstruction (DLIR) | CT scans | LIDC-IDRI | AUC of 0.555, 0.561, 0.557, 0.558 for ASIR-V, DL-L, DL-M, DL-H |
[ ] | 2021 | 2D-3D cascaded CNN | CT scans | LIDC-IDRI | Sensitivity: 90.01% |
[ ] | 2022 | 3D sphere representation-based center-points matching detection network (SCPM-Net) | CT scans | LUNA16 | Average sensitivity: 89.2% |
[ ] | 2022 | YOLOv3 | CT scans | RIDER | Accuracy: 95.17% |
[ ] | 2022 | 3D Attention CNN | CT scans | LUNA16 | CPM: 0.931 |
[ ] | 2022 | Progressive Growing Channel Attentive Non-Local (ProCAN) network | CT scans | LIDC-IDRI | Accuracy: 95.28% |
In recent years, investigators have studied various deep learning techniques to improve the performance of lung nodule classification [ 160 , 161 , 162 , 163 , 164 , 165 , 166 , 167 , 168 , 169 , 170 , 171 , 172 , 173 ]. The sensitivity and specificity of the SIFT-based classifier and SVM in the classification of pulmonary nodules reached 86% and 97% [ 160 ], 91.38%, and 89.56% [ 163 ], respectively. The accuracy, sensitivity, and specificity of multi-scale CNN and multi-crop CNN in lung nodule classification were 90.63%, 92.30%, and 89.47% [ 164 ], respectively, and 87%, 77%, and 93% [ 170 ], respectively. The accuracy of deep-level semantic networks and multi-scale CNN in lung nodule classification were 84.2% [ 167 ] and 86.84% [ 168 ], respectively. The CAD system developed by Cheng et al. [ 169 ] achieved the best accuracy of 95.6%, sensitivity of 92.4%, and specificity of 98.9% in the classification of pulmonary nodules.
The comparative study results showed that the sensitivity and specificity of CNN and DBN for pulmonary nodule classification are 73.40% and 73.30%, 82.20%, and 78.70%, respectively [ 165 ]. Another comparative study showed that the sensitivity and specificity of CNN and ResNet in the classification of nodules are 76.64% and 89.50%, 81.97%, and 89.38%, respectively [ 171 ]. The combined application of CNN and RNN achieved accuracy, sensitivity, and specificity of 94.78%, 94.66%, and 95.14%, respectively, in classifying pulmonary nodules [ 172 ].
In 2019, Zhang et al. [ 174 ] used an ensemble learner of multiple deep CNN in CT images and obtained a classification accuracy of 84% for the LIDC-IDRI dataset. The proposed classifier achieved better performance than other algorithms, such as SVM, multi-layer perceptron, and random forests.
Sahu et al. [ 175 ] proposed a lightweight multi-section CNN with a classification accuracy of 93.18% for the LIDC-IDRI dataset to improve accuracy. The proposed architecture could be applied to select the representative cross sections determining malignancy that facilitate the interpretation of the results.
Ali et al. [ 176 ] developed a system based on transferable texture CNN that consists of nine layers to extract features automatically and classify lung nodules. The proposed method achieved an accuracy of 96.69% ± 0.72%, with an error of 3.30% ± 0.72% and a recall of 97.19% ± 0.57%, respectively.
Marques et al. [ 177 ] developed a multi-task CNN to classify malignancy nodules with an AUC of 0.783. Thamilarasi et al. [ 178 ] proposed an automatic lung nodule classifier based on CNN with an accuracy of 86.67% for the JSRT dataset. Kawathekar et al. [ 179 ] developed a lung nodule classifier using a machine-learning technique with an accuracy of 94% and an F1_score of 92% for the LNDb dataset.
More recently, Radford et al. [ 180 ] proposed deep convolution GAN (DCGAN), Chuquicusma et al. [ 181 ] applied DCGAN to generate realistic lung nodules, and Zhao et al. [ 182 ] applied Forward and Backward GAN (F&BGAN) to classify lung nodules. The F&BGAN was evaluated on the LIDC-IDRI dataset and obtained the best accuracy of 95.24%, a sensitivity of 98.67%, a specificity of 92.47%, and an AUC of 0.98.
Table 4 shows the recently developed traditional and deep learning-based techniques for classifying lung nodules. Among these methods, CNN variants obtained an accuracy range of 83.4–99.6%, a specificity range of 73.3–95.17%, a sensitivity range of 73.3–96.85%, and an AUC range of 0.7755–0.9936, respectively. Several methods achieved high classification accuracy (>95%), including F&BGAN, Inception_ResNet_V2, ResNet152V2, ResNet152V2+GRU, CSO-CADLCC, ProCAN, Net121, ResNet50, DITNN, and optimal DBN with an opposition-based pity beetle algorithm. DCNN systems obtained a sensitivity of 89.3% [ 183 ] and an accuracy of 97.3% [ 184 ]. The classifier was developed based on the VGG19 and CNN models and achieved accuracy, sensitivity, specificity, recall, F1_score, AUC, and MCC above 98%.
Lung nodule classification approaches.
Reference | Year | Method | Imaging | Datasets | Results |
---|---|---|---|---|---|
[ ] | 2014 | FF-BPNN | CT scans | LIDC | Sensitivity: 91.4% |
[ ] | 2015 | Multi-scale CNN | CT scans | LIDC-IDRI | Accuracy: 86.84% |
[ ] | 2015 | CAD using deep features | CT scans | LIDC-IDRI | Sensitivity: 83.35% |
[ ] | 2015 | Deep belief network (DBN) | CT scans | LIDC | Sensitivity: 73.4% |
[ ] | 2015 | CNN | CT scans | LIDC | Sensitivity:73.3% |
[ ] | 2015 | Fractal | CT scans | LIDC | Sensitivity:50.2% |
[ ] | 2015 | Scale-invariant feature transform (SIFT) | CT scans | LIDC | Sensitivity: 75.6% |
[ ] | 2016 | Intensity features +SVM | CT scans | DLCST | Accuracy: 27.0% |
[ ] | 2016 | Unsupervised features+SVM | CT scans | DLCST | Accuracy: 39.9% |
[ ] | 2016 | ConvNets 1 scale | CT scans | DLCST | Accuracy: 84.4% |
[ ] | 2016 | ConvNets 2 scale | CT scans | DLCST | Accuracy: 85.6% |
[ ] | 2016 | ConvNets 3 scale | CT scans | DLCST | Accuracy: 85.6% |
[ ] | 2017 | Multi-crop CNN | CT scans | LIDC-IDRI | Accuracy: 87.14% |
[ ] | 2017 | Deep 3D DPN | CT scans | LIDC-IDRI | Accuracy: 88.74% |
[ ] | 2017 | Deep 3D DPN+ GBM | CT scans | LIDC-IDRI | Accuracy: 90.44% |
[ ] | 2017 | Massive-training ANN (MTANN) | CT scans | LDCT | AUC: 0. 8806 |
[ ] | 2017 | CNN | CT scans | LDCT | AUC: 0.7755 |
[ ] | 2017 | Wavelet Recurrent Neural Network | Chest X-Ray | Japanese Society Radiology and Technology | Sensitivity: 88.24% |
[ ] | 2017 | Multi-crop convolutional neural network (MC-CNN) | CT scans | LIDC-IDRI | Sensitivity: 77% |
[ ] | 2018 | Topology-based phylogenetic diversity index classification CNN | CT scans | LIDC | Sensitivity: 90.70% |
[ ] | 2018 | Transfer learning deep 3D CNN | CT scans | Institution records | Accuracy: 71% |
[ ] | 2018 | CNN | CT scans | Kaggle Data Science Bowl 2017 | Sensitivity: 87% |
[ ] | 2018 | Feature Representation Using Deep Autoencoder | CT scans | ELCAP | Accuracy: 93.9% |
[ ] | 2018 | Multi-view multi-scale CNN | CT scans | LIDC-IDRI & ELCAP | overall classification rates: 92.3% for LIDC-IDRI; overall classification rates: 90.3% for ELCAP |
[ ] | 2018 | Wavelet-Based CNN | CT scans | 448 images include four categories | Accuracy: 91.9% |
[ ] | 2018 | Deep ConvNets | CT scans | LIDC-IDRI | Accuracy: 98% |
[ ] | 2018 | Forward and Backward GAN (F&BGAN) | CT scans | LIDC-IDRI | Sensitivity: 98.67% |
[ ] | 2019 | Ensemble learner of multiple deep CNN | CT scans | LIDC-IDRI | Accuracy: 84.0% |
[ ] | 2019 | Lightweight Multi-Section CNN | CT scans | LIDC-IDRI | Accuracy: 93.18% |
[ ] | 2019 | Deep hierarchical semantic CNN (HSCNN) | CT scans | LIDC | Sensitivity: 70.5% |
[ ] | 2019 | Multi-view knowledge-based collaborative (MV-KBC) | CT scans | LIDC-IDRI | Accuracy: 91.60% |
[ ] | 2019 | 3D CNN | CT scans | LIDC | Sensitivity: 66.8% |
[ ] | 2019 | DCNN | CT scans | 46 images from interventional cytology | Sensitivity: 89.3% |
[ ] | 2019 | 3D MixNet | CT scans | LIDC-IDRI & LUNA16 | Accuracy: 88.83% |
[ ] | 2019 | 3D MixNet +GBM | CT scans | LIDC-IDRI & LUNA16 | Accuracy: 90.57% |
[ ] | 2019 | 3D CMixNet+ GBM | CT scans | LIDC-IDRI & LUNA16 | Accuracy: 91.13 |
[ ] | 2019 | 3D CMixNet+ GBM+Biomarkers | CT scans | LIDC-IDRI & LUNA16 | Accuracy: 94.17% |
[ ] | 2019 | Deep Learning with Instantaneously Trained Neural Networks (DITNN) | CT scans | Cancer imaging Archive (CIA) | Accuracy: 98.42% |
[ ] | 2020 | DCNN | CT scans | LIDC | Accuracy: 97.3% |
[ ] | 2020 | CNN | CT scans | LIDC | Sensitivity: 93.4% |
[ ] | 2020 | 2.75D CNN | CT scans | LUNA16 | AUC: 0.9842 |
[ ] | 2020 | Two-step Deep Network (TsDN) | CT scans | LIDC-IDRI | Sensitivity: 88.5% |
[ ] | 2020 | Transferable texture CNN | CT scans | LIDC-IDRI & LUNGx | Accuracy: 96.69% ± 0.72% |
[ ] | 2020 | Taguchi-Based CNN | X-ray & CT images | 245,931 images | Accuracy: 99.6% |
[ ] | 2021 | Optimal Deep Belief Network with Opposition-based Pity Beetle Algorithm | CT scans | LIDC-IDRI | Sensitivity: 96.86% |
[ ] | 2021 | Multi-task CNN | CT scans | LIDC-IDRI | AUC: 0.783 |
[ ] | 2021 | CNN | CT scans | JSRT | Accuracy: 86.67% |
[ ] | 2021 | Inception_ResNet_V2 | CT scans | LC25000 | Accuracy: 99.7% |
[ ] | 2021 | VGG19 | CT scans | LC25000 | Accuracy: 92% |
[ ] | 2021 | ResNet50 | CT scans | LC25000 | Accuracy: 99% |
[ ] | 2021 | Net121 | CT scans | LC25000 | Accuracy: 99.4% |
[ ] | 2021 | Improved Faster R-CNN and transfer learning | CT scans | Heilongjiang Provincial Hospital | Accuracy: 89.7% |
[ ] | 2021 | Three-stream network | CT scans | LIDC-IDRI | Accuracy: 98.2% |
[ ] | 2021 | FractalNet | CT scans | LUNA 16 | Sensitivity: 96.68% |
[ ] | 2021 | VGG19+CNN | X-ray & CT images | GitHub | Specificity: 99.5% |
[ ] | 2021 | ResNet152V2 | X-ray & CT images | GitHub | Specificity: 98.4% |
[ ] | 2021 | ResNet152V2+GRU | X-ray & CT images | GitHub | Specificity: 98.7% |
[ ] | 2021 | ResNet152V2+Bi-GRU | X-ray & CT images | GitHub | Specificity: 97.8% |
[ ] | 2022 | Machine learning | CT scans | LNDb | Accuracy: 94% |
[ ] | 2022 | Progressively Growing Channel Attentive Non-Local (ProCAN) | CT scans | LIDC-IDRI | Accuracy: 95.28% |
[ ] | 2022 | CNN-based multi-task learning (CNN-MTL) | CT scans | LIDC-IDRI | Sensitivity: 96.2% |
[ ] | 2022 | Cat swarm optimization-based CAD for lung cancer classification (CSO-CADLCC) | CT scans | Benchmark | Specificity: 99.17% |
[ ] | 2022 | 2-Pathway Morphology-based CNN (2PMorphCNN) | CT scans | LIDC-IDRI | Sensitivity: 96.85% |
Forte et al. [ 209 ] recently conducted a systematic review and meta-analysis of the diagnostic accuracy of current deep learning approaches for lung cancer diagnosis. The pooled sensitivity and specificity of deep learning approaches for lung cancer detection were 93% and 68%, respectively. The results showed that AI plays an important role in medical imaging, but there are still many research challenges.
This study extensively surveys papers published between 2014 and 2022. Table 2 , Table 3 and Table 4 demonstrate that deep learning-based lung imaging systems have achieved high efficiency and state-of-the-art performance for lung nodule segmentation, detection, and classification using existing medical images. Compared to reinforcement and supervised learning techniques, unsupervised deep learning techniques (such as CNN, Faster R-CNN, Mask R-CNN, and U-Net) are more popular methods that have been used to develop convolutional networks for lung cancer detection and false-positive reduction.
Previous studies have shown that CT is the most widely used imaging tool in the CAD system for lung cancer diagnosis. Compared to 2D CNN, 3D CNN architectures provide more promising usefulness in obtaining representative features of malignant nodules. To this day, only a few works on 3D CNN for lung cancer diagnosis have been reported.
Deep learning techniques have achieved good performance in segmentation and classification. However, deep learning techniques still have many unsolved problems in lung cancer detection. First, clinicians have not fully acknowledged deep learning techniques for everyday clinical exercise due to the lack of standardized medical image acquisition protocols. The unification of the acquisition protocols could minimize it.
Second, deep learning techniques usually require massive annotated medical images by experienced radiologists to complete training tasks. However, it is costly and time consuming to collect an enormous annotated image dataset, even performed by experienced radiologists. Several methods were applied to overcome the scarcity of annotated data. For example, transfer learning is a possible way to solve the training problem of small samples. Another possible method is the computer synthesis of images, such as the generation of confrontation networks. Inadequate data will inevitably affect the accuracy and stability of predictions. Therefore, improving prediction accuracy using weak supervision, transfer learning, and multi-task learning with small labeled data is one of the future research directions.
Third, the clinical application of deep learning requires high interpretability, but current deep learning techniques cannot effectively explain the learned features. Many researchers have applied visualization and parameter analysis methods to explain deep learning models. However, there is still a certain distance from the interpretable imaging markers required by clinical requirements. Therefore, investigating the interpretable deep learning method will be a hot spot in the medical image field.
Fourth, developing the robustness of the prediction model is a challenging task. Most deep learning techniques work well only for a single dataset. The image of the same disease may vary significantly due to different acquisition parameters, equipment, time, and other factors. This led to poor robustness and generalization of existing deep learning models. Thus, improving the model structure and training methods by combining brain cognitive ideas and improving the generalization ability of deep learning is one of the key future directions.
Finally, some of the current literature has little translation into applicability in clinical practice due to the lack of experience of non-medical investigators in choosing more relevant clinical outcomes. Most deep learning techniques were developed by non-medical professionals with little or no oversight of radiologists, who, in practice, will use these resources when they become more widely available. As a result, some performance metrics, such as accuracy, AUC, and precision, which have little meaningful clinical application, continue to be used and are often the only summary outcomes reported by some studies. Instead, investigators should always strive to report more relevant clinical parameters, such as sensitivity and specificity, because they are independent of the prevalence of the disease and can be more easily translated into practice.
In the future, investigators should pay more attention to the following research directions: (1) develop new convolutional networks and loss functions to improve the performance; (2) weak supervised learning, using a large number of incomplete, inaccurate, and ambiguous annotation data in the existing medical records to achieve model training; (3) bring prior clinical knowledge into model training; (4) radiologists, computer scientists, and engineers need to work more closely to develop more realistic and sensitive models and add more meaning to the research field; (5) single disease identification to complete disease identification. In clinical examination, only a few cases need to solve one well-defined problem. For example, clinicians can detect pulmonary nodules in LDCT and check whether there are other abnormalities, such as emphysema. Solving multiple problems with one network will not reduce performance in specific tasks. In addition, deep learning can be explored in some areas where the medical mechanism is not precise, such as large-scale lung image analysis using deep learning, which is expected to make diagnosing lung diseases more objective.
This paper reviewed recent achievements in deep learning-based approaches for lung nodule segmentation, detection, and classification. CNN is one of the most widely used deep learning techniques for lung disease detection and classification, and CT image datasets are the most frequently used imaging datasets for training networks. The article review was based on recent publications (published in 2014 and later). Experimental and clinical trial results demonstrate that deep learning techniques can be superior to trained radiologists. Deep learning is expected to effectively improve lung nodule segmentation, detection, and classification. With this powerful tool, radiologists can interpret images more accurately. Deep learning algorithm has shown great potential in a series of tasks in the radiology department and has solved many medical problems. However, it still faces many difficulties, including large-scale clinical verification, patient privacy protection, and legal accountability. Despite these limitations, with the current trend and rapid development of the medical industry, deep learning is expected to generate a greater demand for accurate diagnosis and treatment in the medical field.
The author would like to thank the reviewers for their critical comments to improve the manuscript significantly.
This research was funded by the International Science and Technology Cooperation Project of the Shenzhen Science and Technology Commission (GJHZ20200731095804014).
Informed consent statement, data availability statement, conflicts of interest.
The author declares no conflict of interest.
Publisher’s Note: MDPI stays neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Research News
July 18, 2024
Hackensack Meridian Health
CDI scientists publish groundbreaking papers on lung cancer breath test and immune system’s role in fighting diseases. These discoveries could revolutionize cancer detection and treatment, potentially improving patient outcomes.
Scientists at the Center for Discovery and Innovation (CDI) have made exciting progress in cancer research. They’ve published two important papers that could change how we detect and treat cancer.
The first paper talks about a new way to test for lung cancer using a person’s breath. This test could be a game-changer because it’s quick, easy, and doesn’t hurt. The researchers found that certain chemicals in a person’s breath might show if they have lung cancer. This could help doctors find cancer earlier, which is really important for successful treatment.
The second paper is about how our bodies fight off infections and cancer. The scientists discovered that a part of our immune system, called neutrophils, can change to help fight diseases better. This finding could lead to new ways to boost our body’s natural defenses against cancer and infections.
Dr. David Perlin, who leads the CDI, says these discoveries are a big deal. They show how the CDI is working hard to find new ways to help patients. The research team used advanced technology and worked together with other experts to make these breakthroughs.
The breath test for lung cancer is especially exciting because it could be a simple way to check for cancer without invasive procedures. It might even help find other types of cancer in the future.
The study on neutrophils is also important because it gives us new ideas about how to make our immune systems stronger. This could lead to better treatments for both cancer and infections.
Overall, this research is a big step forward in understanding and fighting diseases. It shows how scientists are always working to find new ways to keep us healthy.
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Scientific Reports volume 14 , Article number: 17917 ( 2024 ) Cite this article
570 Accesses
1 Altmetric
Metrics details
Chimeric antigen receptor (CAR) T cells are effective against hematological cancers, but are less effective against solid tumors such as non-small cell lung cancer (NSCLC). One of the reasons is that only a few cell surface targets specific for NSCLC cells have been identified. Here, we report that CD98 heavy chain (hc) protein is overexpressed on the surface of NSCLC cells and is a potential target for CAR T cells against NSCLC. Screening of over 10,000 mAb clones raised against NSCLC cell lines showed that mAb H2A011 bound to NSCLC cells but not normal lung epithelial cells. H2A011 recognized CD98hc. Although CAR T cells derived from H2A011 could not be established presumably due to the high level of H2A011 reactivity in activated T cells, those derived from the anti-CD98hc mAb R8H283, which had been shown to lack reactivity with CD98hc glycoforms expressed on normal hematopoietic cells and some normal tissues, were successfully developed. R8H283 specifically reacted with NSCLC cells in six of 15 patients. R8H283-derived CAR T cells exerted significant anti-tumor effects in a xenograft NSCLC model in vivo. These results suggest that R8H283 CAR T cells may become a new therapeutic tool for NSCLC, although careful testing for off-tumor reactivity should be performed in the future.
Introduction.
Non-small cell lung cancer (NSCLC) is one of the most common causes of cancer deaths worldwide 1 , 2 . Although immune checkpoint blockade therapy has largely improved the prognosis of NSCLC 3 , 4 , advanced NSCLC remains incurable in most cases. New therapeutic options, including CAR T cell therapy, are therefore urgently needed for patients with NSCLC. CAR T cell therapy has shown tremendous efficacy in the treatment of hematological cancers 5 , 6 . Although recent reports have demonstrated that CAR T cells exert an anti-tumor effect against some types of solid tumors 7 , 8 , 9 , 10 , they are still less effective against solid tumors, including NSCLC, than against hematological cancers. One of the major reasons is the lack of cell surface target antigens that are specific for tumor cells. Several cell surface antigens, such as epidermal growth factor receptor 11 , 12 , mesothelin 10 , prostate stem cell antigen 13 , mucin 1 13 , human epidermal growth factor receptor 2 14 , carcinoembryonic antigen 15 , 16 , programmed death-ligand 1 17 , and receptor tyrosine-kinase-like orphan receptor 18 , 19 , 20 , have been tested as targets for CAR T cells intended to treat NSCLC, although these target molecules are not completely tumor specific. Identifying additional target antigens specific for NSCLC is important to develop effective and safe CAR T-cell therapy.
Expression levels of messenger RNA were previously shown to lack sufficient correlation with the abundances of their corresponding proteins 21 . In addition, cancer-specific conformational epitopes formed by post-translational events such as glycosylation or conformational changes may have been missed in screening using transcriptome analysis 22 . Previously, we thoroughly screened for multiple myeloma (MM)-specific monoclonal antibodies (mAbs) among large numbers of mAbs raised against MM cells, and identified two novel mAbs recognizing MM-specific antigens that could not be found by transcriptome analysis 23 , 24 . The first was an mAb that specifically recognized the activated integrin β7, which is constitutively overexpressed in MM cells 23 . The second was R8H283, which had been shown to lack reactivity with CD98hc glycoforms expressed on normal hematopoietic cells and also with some normal tissues 24 . In this study, we applied the same strategy to identify NSCLC-specific cell surface targets and found that H2A011, which recognizes the CD98hc protein, reacted specifically with a subset of NSCLC samples, but not with normal lung epithelial cells. Although T cells transduced with H2A011-derived CAR could not be expanded in vitro, T cells transduced with another anti-CD98hc mAb that we previously reported 24 could be expanded and exerted anti-tumor activity in vitro and in vivo.
We immunized mice with one of five NSCLC cell lines (A549, H1792, H1975, H2228, or HCC827) and generated approximately 10,000 clones of mAbs that bound to the cell line used for immunization. Among them, we selected 573 hybridomas that produced mAbs lacking reactivity to Ep-CAM + lung epithelial cells obtained from normal regions of resected lung tissues. Then, Ep-CAM + NSCLC cells from tumor regions of resected specimens were stained with the candidate mAbs and subjected to flow cytometry analysis. We identified 12 candidate mAbs that bound to Ep-CAM + NSCLC cells in at least one sample. Among them, we focused on H2A011, which reacted most frequently with NSCLC cells (Fig. 1 A). Distinct binding of H2A011 to NSCLC cells was observed in four of the five patients with NSCLC, while H2A011 did not react with any of the five samples of normal lung epithelial cells (Fig. 1 B,C, Supplementary Fig. S1 A and B, Supplementary Table S1 ). H2A011 also reacted with all NSCLC cell lines tested (Supplementary Fig. S1 C).
An anti-CD98hc mAb, H2A011, reacts with NSCLC cells but not with normal lung epithelial cells. ( A ), Strategy for identification of the NSCLC-specific mAb H2A011. ( B ) and ( C ) , Flow cytometry analyses of H2A011 reactivity against CD45 − Ep-CAM + normal lung epithelial cells ( B ) and CD45 − Ep-CAM + tumor cells ( C ) from resected NSCLC tissues. Analyses of live (propidium iodide-negative) cells are shown. Results of staining with isotype control instead of anti-Ep-CAM mAb were used to draw the gate for Ep-CAM + cells. The analysis of cells from patient UPN4 is shown as an example. The results of other patients are shown in Supplementary Fig. S1 . ( D ), Strategy for identifying the antigen recognized by H2A011. ( E ), Flow cytometry plots showing the process of enriching H2A011 + cells in expression cloning of the H2A011 antigen. ( F ), Flow cytometry analysis of the binding of H2A011 or MEM-108 (a known anti-CD98hc mAb) to wild-type (WT) or CD98hc-deficient (CD98hc KO) A549 cells.
The antigen recognized by H2A011 was identified by expression cloning using retroviruses 25 (Fig. 1 D). Specifically, retroviruses carrying a cDNA library generated from A549 cells (H2A011 positive) were used to infect Ba/F3 cells (H2A011 negative), and then cells labeled with H2A011 were enriched by FACS. After the third round of cell sorting, most cells were H2A011 positive. Sequencing of the inserted cDNA revealed that H2A011 recognized CD98hc (also known as SLC3A2) (Fig. 1 E). Consistent with this, H2A011 reactivity was absent in CD98hc knockout (KO) A549 cells that were established using the CRISPR-Cas9 system and confirmed as CD98hc deficient by lack of staining with the known CD98hc-reactive antibody MEM-108 (Fig. 1 F).
In transcriptome analyses comparing CD45 − CD31 − epithelial cell adhesion molecule (Ep-CAM) + cells 26 from tumor regions with those from normal regions of lung tissues resected from three patients with NSCLC (Supplementary Fig. S2 A, B, Supplementary Table S1 ), CD98hc (SLC3A2) mRNA expression was comparable in two of three patients (Supplementary Fig. S2 C). In addition, we confirmed the expression of CD98hc in publicly available single-cell RNA sequencing data. We extracted the data from Laughney et al., 27 which included the samples from eight lung NSCLC cells and four normal lungs from the Human Lung Cell Atlas dataset 28 . The expression of CD98hc was not significantly different between NSCLC cells and normal lung epithelial cells (Supplementary Fig. S3 ).
Four CAR constructs derived from the variable region of H2A011 were established using either CD28 or 4-1BB as a co-stimulatory molecule (Fig. 2 A). T cells were transduced with each CAR construct and cultured in vitro. However, after 10 d of culture, T cells expressing each CAR were scarcely detected (Fig. 2 B). High levels of H2A011 reactivity in activated T lymphocytes (Fig. 2 C,D) may be a cause of loss of H2A011 CAR T cells.
Successful development of CAR T cells derived not from H2A011, but from another anti-CD98hc mAb, R8H283, that lacks reactivity with CD98hc glycoforms expressed on normal hematopoietic cells. ( A ), Constructs for the CAR derived from H2A011. ( B ), Flow cytometry analysis of H2A011 CAR transduction efficiencies 7 d after CAR transduction. ( C , D ), Flow cytometric analysis of R8H283 or H2A011 reactivity against phytohemagglutinin P (PHA)-activated T cells ( C ) and CD3/CD28-stimulated T cells ( D ). A549 cells were simultaneously stained as a positive control. ( E ), Flow cytometry analysis of the binding of H2A011 or R8H283 to wild-type (WT) or GnTI-deficient (GnTI − ) 293 cells. F, Construct for the CAR derived from R8H283. ( G ), Growth of R8H283 CAR T cells during in vitro culture. The data are presented as means ± standard error of the mean (SEM). *: p < 0.05. ( H ), Representative flow cytometry analysis data of R8H283 CAR transduction efficiencies and CD4/CD8 expression in CAR T cells 7 d after CAR transduction.
Another anti-CD98hc mAb, R8H283, was previously shown to bind myeloma cells but not normal tissues due to differences in CD98hc N-glycosylation 24 . Consistently, R8H283, but not H2A011, reactivity was significantly increased in GnTI-deficient 293cells, which do not have N-acetylglucosaminyltransferase I (GnTI) activity and therefore lack complex N-glycans (Fig. 2 E). A CAR construct derived from the variable region of R8H283 was established using CD28 as a co-stimulatory molecule (Fig. 2 F). T cells were transduced with the CAR construct and cultured in vitro. CAR T cells expressing the R8H283-derived CAR could be expanded, although the expansion of R8H283 CAR T cells was reduced compared to that of control T cells (Fig. 2 G,H). Reactivity of R8H283 was detected in activated T cells but was much lower than that of H2A011 (Fig. 2 C,D). R8H283 CAR T cells spontaneously produced small amounts of cytokines even in the absence of antigen stimulation (Supplementary Fig. S4 ).
R8H283 reacted with CD45 − CD31 − Ep-CAM + NSCLC cells in six of 15 patients (Fig. 3 A,B, Supplementary Fig. S5 A), but did not react with any of the normal lung epithelial cells from ten patients (Fig. 3 C and D and Supplementary Fig. S5 B). These results indicate that R8H283 reactivity is specific for NSCLC cells in a subset of patients. Five of the six NSCLC samples that reacted with R8H283 were squamous cell carcinomas (Fig. 3 A and B , Supplementary Fig. S5 A, Supplementary Table S1 ). In the samples that we were able to analyze in pairs (tumor vs normal epithelial cells), R8H283 reacted with tumor cells but not with normal lung epithelial cells (UPN 7, 10, 11). The results of MEM108 (pan-CD98hc mAb) staining showed that CD98hc protein was expressed at significantly higher levels on NSCLC cells than on normal epithelial cells, hematopoietic cells, and endothelial cells (Fig. 3 E). To further explore the basis for the NSCLC specificity of R8H283, we compared CD98hc in normal lung epithelial cells and NSCLC cells. Whole cell lysates of normal lung epithelial cells or NSCLC cells from a patient were electrophoresed and immunoblotted with polyclonal anti-CD98 antibody (Supplementary Fig. S6 ). The mobility of CD98hc was different in the NSCLC cells compared to normal lung epithelial cells. We also found that the electrophoretic mobilities of the CD98hc species in NSCLC cells and normal lung epithelial cells were still different after the removal of N-glycans by PNGase F treatment. Thus, it was unclear whether the difference in electrophoretic mobility reflected from the difference in glycosylation or the expressed spliced forms.
R8H283 reacted with NSCLC cells in a subset of patients. ( A ), Flow cytometry analyses of R8H283 reactivity against CD45 − CD31 − Ep-CAM + tumor cells in the tumor region of lung tissues resected from a patient with NSCLC (UPN5). ( B ), Representative results of flow cytometry analyses of R8H283 reactivity against CD45 − CD31 − Ep-CAM + tumor cells in the tumor regions of lung tissues resected from patients with NSCLC. Analyses of the other samples are shown in Supplementary Fig. S5 A. ( C ), Flow cytometry analysis of R8H283 reactivity against CD45 − CD31 − Ep-CAM + lung epithelial cells in normal regions of resected lung tissues from a patient with NSCLC (UPN10). ( D ), Representative results of flow cytometry analyses of R8H283 reactivity against CD45 − CD31 − Ep-CAM + lung epithelial cells in unaffected regions of resected lung tissues. Analyses of the other samples are shown in Supplementary Fig. S5 B. ( E ), Flow cytometric analysis of R8H283 or H2A011 reactivity against each cell subset in the normal and tumor regions of the resected lung tissue. The analysis of UPN 7 is shown as a representative of three tested samples.
CAR T cells derived from R8H283, but not those derived from a CD19 antibody (used as a control because they are specific for an irrelevant target), secreted IFN-γand IL-2, and exhibited cytotoxic activity when co-cultured with A549 lung cancer cells, but not when co-cultured with CD98hc-deficient A549 cells established using CRISPR-Cas9 (Fig. 4 A,B). R8H283 CAR T cells produced minimal amounts of cytokines upon co-culture with normal lung epithelial cells purified from normal regions of resected lung tissues of patients with NSCLC (Fig. 4 C–E).
CAR T cells derived from R8H283 specifically recognize and kill NSCLC cells. ( A ), Secretion of IFN-γ and IL-2 by R8H283 CAR T cells or CD19 CAR T cells (a control cell type targeting an irrelevant antigen) after co-culture with WT or CD98hc-deficient (CD98hc KO) A549 NSCLC cells. ( B ), 51 Cr release assay for measurement of specific lysis of WT or CD98hc KO A549 cells by R8H283 CAR T cells or CD19 CAR T cells. E/T, effector/target ratio. C, Experimental design for D and ( E . D ), Representative flow cytometry analysis of normal lung tissue. CD45 - CD31 - Ep-CAM + normal lung epithelial cells were purified and subjected to the assay. ( E ), Secretion of IFN-γ and IL-2 by R8H283 CAR T cells after co-culture with normal lung epithelial cells or A549 NSCLC cells. ( F ), Experimental design for G - I . IVIS, in vivo imaging system; i.v., intravenous. ( G ), Bioluminescence imaging of mice infused with either R8H283 or CD19 CAR T cells. ( n = 6 per group). Min, minimum. ( H ), Quantification of whole-body luminescence. Avg., average; p, photons; s, second; sr, steradian. ( I ), Survival curves of mice infused with either R8H283 or CD19 CAR T cells. The data are presented as means ± SEM. * P < 0.05 and ** P < 0.01 were calculated using two-tailed Student’s t- test ( H ) and the generalized Wilcoxon test ( I ).
In a lung cancer xenograft model established by intravenous injection of luciferase-expressing A549 cells into NOG mice 29 , infusion of R8H283 CAR T cells, but not CD19 CAR T cells, significantly decreased the tumor burden as determined by bioluminescence imaging (Fig. 4 F–H) and enhanced mouse survival (Fig. 4 I). No unexpected side effects were observed in mice injected with R8H283 CAR T cells. In a mouse that relapsed after R8H283 CAR T cell infusion, R8H283 reactivity to tumor cells was reduced compared to untreated tumor cells (Supplementary Fig. S7 ).
In this study, by thoroughly screening for NSCLC-specific mAbs from among approximately 10,000 mAbs raised against NSCLC cell lines, we found that the CD98hc-specific mAb H2A011 distinctly bound to most NSCLC cells but not to normal lung epithelial cells. Consistently, overexpression of CD98hc protein was previously demonstrated by immunohistochemistry in a subset of NSCLC samples 30 , 31 , 32 . CD98hc mRNA was not overexpressed in purified NSCLC cells compared with normal lung epithelial cells in two of three samples examined. In addition, according to the Cancer Genome Atlas, a gene expression database, CD98hc mRNA is also expressed in normal lung tissues at levels comparable with lung cancer tissues 33 . Furthermore, the analysis of publicly available single-cell RNAseq data showed that CD98hc mRNA expression did not differ between normal lung epithelial cells and NSCLC cells. These results showed that tumor-specific antigens that cannot be discovered by transcriptome analysis can be identified by thoroughly screening for tumor-specific mAbs among large numbers of mAbs raised against tumor cells, as we previously reported 23 , 24 , 34 , although the targets identified by the mAb discovery strategy are mostly conformational epitopes in proteins that are also expressed in normal tissues and on-target/off-tumor effects must be very carefully excluded.
The CD98 heterodimer is composed of CD98hc that is disulfide-linked with a light chain. The heavy chain binds to the cytoplasmic tails of integrin-β chains 35 , 36 , 37 and mediates adhesive signals that control cell spreading, survival, and growth 37 , 38 , 39 , 40 . The CD98 light chains (lcs) function in amino acid transport 41 , 42 and play important roles in the survival and growth of various cells 43 , 44 . Overexpression of CD98lc L-type amino acid transporter 1 (LAT1) in NSCLC 45 , 46 may enhance cell surface expression of CD98hc/lc heterodimers. The mechanisms of CD98hc protein overexpression on the surface of NSCLC cells should be clarified in future studies.
H2A011 CAR-transduced T cells failed to survive after 10 days of in vitro culture. While the loss of H2A011 CAR T cells could be caused by fratricide, ligand-dependent suboptimal CAR signaling could cause apoptosis of H2A011 CAR T cells as shown in a previous study 47 . In contrast, T cells transduced with the CAR derived from another anti-CD98hc mAb R8H283 could be expanded, although the in vitro expansion of R8H283 CAR T cells was not as good as that of control T cells. R8H283 reactivity in activated T cells may cause partial loss of R8H283 CAR T cells during in vitro culture.
R8H283, which has been shown to lack reactivity with CD98hc glycoforms expressed on normal hematopoietic cells and some normal tissues, reacted with NSCLC cells in a subset of patients. In the samples that we were able to analyze in pairs (tumor vs normal epithelial cells), R8H283 reacted with tumor cells but not with normal lung epithelial cells (UPN 7, 10, 11), although paired analysis of more samples should be performed in the future. In a previous report, we showed that R8H283 did not react with normal lymphocytes, monocytes, or non-hematopoietic cells such as intestinal epithelial cells or skin epidermal cells, although CD98hc protein is expressed on these cells 24 . We demonstrated that CAR T cells derived from R8H283 exerted a significant anti-tumor effect in an in vivo xenograft model. These results suggest that R8H283 CAR T cells have the potential to specifically target NSCLC cells without damaging normal cells in a subset of NSCLC patients, while the possibility of immune escape of tumor cells with low R8H283 reactivity should be carefully evaluated. Most of R8H283-reactive tumors in this study were squamous cell carcinomas, suggesting that R8H283-derived therapies will be useful in patients with squamous cell carcinoma, although a larger number of NSCLC samples should be analyzed in the future.
Although CD98hc protein expression on the surface of tumor cells was detected in most patients with NSCLC, reactivity to R8H283 was observed in only six of the 15 patients examined in this study. R8H283, which is expected to have a lower affinity for CD98hc than MEM108, may only react with NSCLC cells that express high levels of CD98hc. While the reactivity of R8H283 is certainly affected by alterations in the N-glycosylation of CD98hc, it remains unclear whether the NSCLC-specific reactivity of R8H283 is associated with alterations in the glycosylation of CD98hc in NSCLC cells.
A number of mAbs targeting CD98hc have been described in the context of cancer therapy 48 , 49 , 50 , 51 , 52 , 53 , 54 , 55 , and some have been tested in clinical trials. Since CD98hc is expressed by several normal tissues, including normal lymphocytes, on-target off-tumor toxicity in normal tissues is always a concern when CD98hc is used as a therapeutic target. Although we showed that R8H283 reactivity was not detected in the normal human tissues that were available for testing 24 , it is difficult to completely exclude the possibility that under some conditions, the epitope recognized by R8H283 is formed in normal tissues expressing CD98hc. While the low levels of cytokine production by R8H283 CAR T cells co-cultured with normal lung epithelial cells is likely to reflect the spontaneous secretion from R8H283 cells, we could not completely exclude the possibility that R8H283 CAR T cells may be weakly reactive with normal lung epithelial cells. Since R8H283 does not react with mouse CD98hc, it is impossible to analyze the toxicity of R8H283 CAR T cells against normal cells in mouse xenograft models. Therefore, we must carefully examine the off-tumor reactivity of R8H283 before initiating a clinical study. In addition, it may be beneficial to develop a logic-gated CAR 56 , 57 , 58 that recognizes only cells expressing both the R8H283 antigen and another NSCLC-specific antigen, for example mesothelin.
Lung tissue specimens from patients diagnosed with adenocarcinoma, squamous cell carcinoma, or pleomorphic carcinoma and who underwent surgical resection were used after written informed consent was obtained. This study conformed to the ethical guidelines outlined in the Declaration of Helsinki, and was approved by the institutional review boards of the Osaka University School of Medicine, Osaka Toneyama Medical Center, Osaka International Cancer Institute, Takarazuka City Hospital, Toyonaka Municipal Hospital, Suita Municipal Hospital, Minoh City Hospital, Kinki-Chuo Chest Medical Center, and Osaka Fukujuji Hospital.
The A549, H1792, H1975, H2228, and HCC827 cell lines were purchased from the American Type Culture Collection (ATCC). The SP2/0 mouse myeloma cell line was kindly gifted by I. Weissman (Stanford University). The Expi293 and Expi293 GnTI-deficient cell lines were purchased from Thermo Fisher Scientific. A549 cells expressing green fluorescent protein (GFP) and firefly luciferase (A549-GFP-luc) were established by retroviral transduction. Following gene transduction, GFP high cells were enriched by fluorescence-activated cell sorting (FACS) on a BD FACS Aria II (Becton Dickinson). CD98-deficient A549 cells were established using CRISPR-Cas9, as previously reported 24 .
To prepare single-cell suspensions from lung tissues, samples were dissociated using the Human Tumor Dissociation Kit (Miltenyi Biotech) and gentleMACS Octo Dissociator with Heaters (Miltenyi Biotech). After tissue dissociation, cell suspensions were filtered through a cell strainer (Corning) and red blood cells were lysed using ACK Lysing Buffer (Gibco). Cells were stained with the indicated mAbs after incubation with Human Serum AB (GeminiBio) and FcR Blocking Reagent Human (Miltenyi Biotech). The following antibodies were used: anti-human CD326 (Ep-CAM)-PE/Cyanine7 (9C4, BioLegend), anti-human CD45-APC (HI130, BioLegend), anti-human CD45-FITC (HI130, BioLegend), anti-human CD31-APC (WM-59, Invitrogen), anti-human CD3-FITC (SK7, BioLegend), anti-human CD19-APC/Cyanine7 (HIB19, BioLegend), and anti-human CD14-APC (M5E2, BioLegend), goat anti-mouse IgG, F(ab’)2 Fragment Specific-Alexa Fluor 647 (115–605-072, Jackson ImmunoResearch), Goat anti-mouse IgG-PE (405,307, BioLegend, Poly4053). H2A011 (mouse IgG1) and R8H283 (mouse IgG2a) were purified from hybridoma supernatants with Protein G Sepharose 4 Fast Flow (GE Healthcare) and used for staining at a concentration of 10 µg/ml and 50 µg/ml, respectively. Flow cytometry analysis and cell sorting were performed using a BD Canto II and Aria II (Becton Dickinson).
Phytohemagglutinin P (PHA)-activated T cells were generated by culturing human PBMC in the presence of 3ug/ml PHA (Sigma) for 72 h, stained with R8H283 or H2A011, then with goat anti-mouse IgG-PE, and analyzed on flow cytometry. Peripheral blood mononuclear cells were activated with anti-CD3 (OKT3, eBioscience) and anti-CD28 (CD28.2, eBioscience) mAbs and cultured in X-VIVO 15 (Lonza) supplemented with 5% human Sserum AB (GeminiBio) for 24 h, stained with biotinylated R8H283 or H2A011, then with streptavidin-PE (BioLegend), and analyzed on flow cytometry. R8H283 or H2A011 was biotinylated using biotin labeling kit (Dojindo).
NSCLC cells and unaffected lung epithelial cells were purified by FACS. Total RNA was extracted using TRIzol Reagent (Thermo Fisher Scientific) and the RNeasy Mini Kit (QIAGEN). Full-length cDNA was generated using the SMART-Seq HT Kit (Takara Bio). Each library was prepared using a Nextera XT DNA Library Prep Kit (Illumina). Whole-transcriptome sequencing was performed on RNA samples using an Illumina HiSeq 3000 platform (Illumina) in 100-base single-end mode. Sequenced reads were mapped to human reference genome sequences (hg19) using TopHat v2.0.13 in combination with Bowtie2 ver. 2.2.3 and SAMtools ver. 0.1.19. The number of fragments per kilobase of exon per million mapped fragments was calculated using Cufflinks ver. 2.2.1. RNA sequencing data concerning this study have been deposited in the Gene Expression Omnibus (GEO) ( https://www.ncbi.nlm.nih.gov/geo/query/acc.cgi?acc=GSE226774 ). The data sets were analyzed using Ingenuity Pathway Analysis (Ingenuity Systems Inc).
To confirm the expression of CD98hc (SLC3A2), publicly available single-cell RNA sequencing data was re-analyzed. We extracted the data of Laughney et al., which included the samples from eight lung NSCLC cells and four normal lungs from the human lung cell atlas dataset. We evaluated the gene expression changes in epithelial cells between the NSCLC tissues and normal lung tissues using the Wilcoxon rank-sum test, employing the Seurat FindMarkers function. We then checked the CD98hc data from the output. The P -value was corrected using the Bonferroni method for all genes expressed in the epithelial cells.
Six- to eight-week-old BALB/cAJcl mice (CLEA Japan) were immunized by footpad injection with human NSCLC cell lines (A549, H1792, H1975, H2228, or HCC827). Lymphocytes from popliteal lymph nodes were fused with SP2/0 mouse myeloma cells in PEG solution (Roche Applied Science). To identify hybridoma clones producing mAbs that reacted with NSCLC cells, NSCLC cells were first incubated with hybridoma supernatants, then incubated with PE-conjugated anti-mouse IgG antibody and analyzed by flow cytometry. Hybridoma clones producing mAbs that reacted with NSCLC cells were selected and stocked for further analyses.
Expression cloning was performed as previously reported 25 . A cDNA library was generated from A549 cells using the Superscript Choice System (Invitrogen) and linked with a BstXI adaptor. cDNA fragments ranging from 2.0 to 5.0 kb were selected on a CHROMA SPIN column (Takara Bio), purified by agarose gel electrophoresis, and then cloned into retrovirus vector pMX (a kind gift from T. Kitamura, Tokyo University). The A549 cDNA library was subjected to screening by transduction into Ba/F3 cells. Ba/F3 cells with which H2A011 reacted were enriched by FACS, then subjected to PCR cloning of the inserted cDNA.
cDNA of the variable region of H2A011 or R8H283 was obtained by 5’-RACE PCR with a Smarter RACE PCR Kit (Takara Bio), then sequenced. The isolated cDNAs of the κ light and heavy chain variable regions were fused to CD28 (Uniprot P10747 aa.114–220) and CD3ζ(Uniprot P20963 aa.52–164) cDNAs by overlapping PCR. Sequences of the leader peptides, linker, and variable regions of the κ light and the heavy chains of H2A011 are listed in Supplementary Table S2 . The sequences of the variable regions of the κ light and the heavy chains of R8H283 are shown in the patent (WO2017026497A1). The resultant H2A011 or R8H283 CAR constructs were inserted into pMSCV retroviral vectors. The CD19 CAR was constructed according to the reported sequences of the anti-CD19 mAb 59 , 60 . Then, 293 T cells were co-transfected with retroviral vector, gag-pol, and VSV-G envelope plasmids with Lipofectamine 2000 reagent (Thermo Fisher Scientific). Supernatants containing the retrovirus were collected 48 h and 72 h later. Activated T cells were infected with retrovirus carrying the H2A011 or R8H283 CAR. Briefly, peripheral blood mononuclear cells were activated with anti-CD3 (OKT3, eBioscience) and anti-CD28 (CD28.2, eBioscience) mAbs and cultured in X-VIVO 15 (Lonza) supplemented with 5% Human Serum AB (GeminiBio). The next day, recombinant human IL-2 (Shionogi Pharma) was added to the culture at a final concentration of 100 IU/ml. Cells were harvested 2 d after activation, then subjected to retroviral transduction with RetroNectin (Takara Bio). After transduction, the cells were cultured in the presence of 100 IU/ml IL-2 for 7 d. Dasatinib (1 µM) was added to the culture medium beginning 4 d after CAR transduction to prevent T-cell exhaustion, as previously described 61 . The transduction efficiency of each CAR was measured by staining cells with goat anti-mouse F(ab′) 2 -Alexa Fluor 647 mAb.
R8H283 CAR T cells or mock-transduced (control) T cells were tested for reactivity in cytokine release assays. Cytokine concentrations were measured using an ELISA kit (IFN-γ and IL-2; R&D Systems). Effector cells and target cells (1.0 × 10 5 cells each) were co-cultured for 16 h. Co-culture was performed in technical-triplicate wells. Cytokine secretion was measured in culture supernatants diluted to fall within the linear range of the assay.
The cytotoxic ability of CAR T cells was evaluated by 51 Cr release assay. Briefly, target cells were labeled for 90 min at 37ºC with 25 μCi of [ 51 Cr] sodium chromate (PerkinElmer). Labeled target cells (1.0 × 10 4 ) were incubated with effector cells for 4 h at the indicated effector/target ratios. 51 Cr release in harvested supernatants was counted with a gamma counter. Total and spontaneous 51 Cr release was determined by incubation of 1.0 × 10 4 labeled target cells in either 1% Triton X-100 or culture medium. The percentage of specific lysis was calculated as ([specific 51 Cr release − spontaneous 51 Cr release] / [total 51 Cr release − spontaneous 51 Cr release]) × 100.
Total cell lysate was prepared in lysis buffer [10 mM Tris–HCl pH 7.5, 150 mM NaCl, 1 mM EDTA, 0.1% NP-40, and protease inhibitor cocktail (Nacalai Tesque)]. Cell lysates from normal lung epithelial cells or NSCLC cells were run on a 4–12% NuPAGE gel system (Invitrogen) under reducing (0.7 M 2ME) or non-reducing conditions. To remove N-glycans attached to proteins, cell lysates were incubated at 37 °C for 30 min with 1,000 units of PNGase F PRIME (N-Zyme Scientifics), and then subjected to SDS-PAGE. Western blotting was carried out with anti-CD98 polyclonal Ab (pAb) (#15,193–1-AP, ProteinTech) and subsequently with HRP-conjugated donkey anti–rabbit IgG (#NA934V, GE Healthcare). Imaging of blots was performed using the LAS system (GE Healthcare).
Female NOD/SCID/IL-2Rγcnull (NOG) mice aged 6–8 weeks (In-Vivo Science) were injected intravenously via the tail vein with 2.0 × 10 5 A549-luc/GFP tumor cells. Two days after tumor inoculation, the mice were intraperitoneally infused with VIVOGlo Luciferin (Promega, 150 mg/kg body weight), anesthetized with isoflurane, and imaged using an in vivo imaging system (IVIS) (PerkinElmer). The mice were then injected intravenously with CD19 or R8H283 CAR T cells (5.0 × 10 6 cells/mouse). Mice were reanalyzed with the IVIS every week. To minimize suffering and distress, mice were subjected to inhaled anesthesia (isoflurane) before cell injection. The health status of the mice was carefully examined three times per week by a veterinarian. Mice were euthanized when moribund or as recommended by a veterinarian. Investigators were not blinded.
All mouse experiments in this study were approved by the administrative panel on Laboratory Animal Care at Osaka University (Ethical Approval ID 03–071 (for mAb production), 28–054 and 03–045 (for xenograft models)). Mice were euthanized by CO2 asphyxia. This study conforms to the Guide for the Care and Use of Laboratory Animals published by the US National Institutes of Health and is reported in accordance with ARRIVE guidelines.
Statistical analyses for significant differences between two groups were conducted using unpaired two-tailed Student’s t -test. The generalized Wilcoxon test was used to compare survival differences between the two groups. P < 0.05 was considered to indicate a significant difference. Statistical analyses were performed in GraphPad Prism 9.
The datasets generated during the current study are available from the corresponding author on reasonable request.
Hirsch, F. R. et al. Lung cancer: Current therapies and new targeted treatments. Lancet 389 , 299–311. https://doi.org/10.1016/s0140-6736(16)30958-8 (2017).
Article CAS PubMed Google Scholar
Chen, Z., Fillmore, C. M., Hammerman, P. S., Kim, C. F. & Wong, K. K. Non-small-cell lung cancers: A heterogeneous set of diseases. Nat. Rev. Cancer 14 , 535–546. https://doi.org/10.1038/nrc3775 (2014).
Article CAS PubMed PubMed Central Google Scholar
Herbst, R. S., Morgensztern, D. & Boshoff, C. The biology and management of non-small cell lung cancer. Nature 553 , 446–454. https://doi.org/10.1038/nature25183 (2018).
Article ADS CAS PubMed Google Scholar
Thai, A. A., Solomon, B. J., Sequist, L. V., Gainor, J. F. & Heist, R. S. Lung cancer. Lancet 398 , 535–554. https://doi.org/10.1016/s0140-6736(21)00312-3 (2021).
Article PubMed Google Scholar
June, C. H. & Sadelain, M. Chimeric Antigen Receptor Therapy. N. Engl. J. Med. 379 , 64–73. https://doi.org/10.1056/NEJMra1706169 (2018).
June, C. H., O’Connor, R. S., Kawalekar, O. U., Ghassemi, S. & Milone, M. C. CAR T cell immunotherapy for human cancer. Science 359 , 1361–1365. https://doi.org/10.1126/science.aar6711 (2018).
Majzner, R. G. et al. GD2-CAR T cell therapy for H3K27M-mutated diffuse midline gliomas. Nature 603 , 934–941. https://doi.org/10.1038/s41586-022-04489-4 (2022).
Article ADS CAS PubMed PubMed Central Google Scholar
Quintarelli, C., Del Bufalo, F. & Locatelli, F. GD2-CART01 for Relapsed or Refractory High-Risk Neuroblastoma. Reply. N. Engl. J. Med. 388 , 2303–2304. https://doi.org/10.1056/NEJMc2305296 (2023).
Qi, C. et al. Claudin18.2-specific CAR T cells in gastrointestinal cancers: phase 1 trial interim results. Nat Med 28 , 1189–1198. https://doi.org/10.1038/s41591-022-01800-8 (2022).
Adusumilli, P. S. et al. A Phase I Trial of Regional Mesothelin-Targeted CAR T-cell Therapy in Patients with Malignant Pleural Disease, in Combination with the Anti-PD-1 Agent Pembrolizumab. Cancer Discov. 11 , 2748–2763. https://doi.org/10.1158/2159-8290.Cd-21-0407 (2021).
Feng, K. et al. Chimeric antigen receptor-modified T cells for the immunotherapy of patients with EGFR-expressing advanced relapsed/refractory non-small cell lung cancer. Sci. China Life Sci. 59 , 468–479. https://doi.org/10.1007/s11427-016-5023-8 (2016).
Li, H. et al. Antitumor activity of EGFR-specific CAR T cells against non-small-cell lung cancer cells in vitro and in mice. Cell. Death Dis. 9 , 177. https://doi.org/10.1038/s41419-017-0238-6 (2018).
Wei, X. et al. PSCA and MUC1 in non-small-cell lung cancer as targets of chimeric antigen receptor T cells. Oncoimmunology https://doi.org/10.1080/2162402x.2017.1284722 (2017).
Article PubMed PubMed Central Google Scholar
Gao, Q. et al. Cancer-cell-secreted CXCL11 promoted CD8(+) T cells infiltration through docetaxel-induced-release of HMGB1 in NSCLC. J. Immunother. Cancer 7 , 42. https://doi.org/10.1186/s40425-019-0511-6 (2019).
Chmielewski, M. & Abken, H. CAR T Cells Releasing IL-18 Convert to T-Bet(high) FoxO1(low) Effectors that Exhibit Augmented Activity against Advanced Solid Tumors. Cell Rep. 21 , 3205–3219. https://doi.org/10.1016/j.celrep.2017.11.063 (2017).
Thistlethwaite, F. C. et al. The clinical efficacy of first-generation carcinoembryonic antigen (CEACAM5)-specific CAR T cells is limited by poor persistence and transient pre-conditioning-dependent respiratory toxicity. Cancer Immunol. Immunother. 66 , 1425–1436. https://doi.org/10.1007/s00262-017-2034-7 (2017).
Liu, M. et al. Targeting PD-L1 in non-small cell lung cancer using CAR T cells. Oncogenesis 9 , 72. https://doi.org/10.1038/s41389-020-00257-z (2020).
Wallstabe, L. et al. ROR1-CAR T cells are effective against lung and breast cancer in advanced microphysiologic 3D tumor models. JCI Insight https://doi.org/10.1172/jci.insight.126345 (2019).
Srivastava, S. et al. Immunogenic Chemotherapy Enhances Recruitment of CAR-T Cells to Lung Tumors and Improves Antitumor Efficacy when Combined with Checkpoint Blockade. Cancer Cell. 39 , 193-208.e110. https://doi.org/10.1016/j.ccell.2020.11.005 (2021).
Berger, C. et al. Safety of targeting ROR1 in primates with chimeric antigen receptor-modified T cells. Cancer Immunol. Res. 3 , 206–216. https://doi.org/10.1158/2326-6066.Cir-14-0163 (2015).
Vogel, C. & Marcotte, E. M. Insights into the regulation of protein abundance from proteomic and transcriptomic analyses. Nat. Rev. Genet. 13 , 227–232. https://doi.org/10.1038/nrg3185 (2012).
Posey, A. D. Jr. et al. Engineered CAR T Cells Targeting the Cancer-Associated Tn-Glycoform of the Membrane Mucin MUC1 Control Adenocarcinoma. Immunity 44 , 1444–1454. https://doi.org/10.1016/j.immuni.2016.05.014 (2016).
Hosen, N. et al. The activated conformation of integrin beta7 is a novel multiple myeloma-specific target for CAR T cell therapy. Nat. Med. 23 , 1436–1443. https://doi.org/10.1038/nm.4431 (2017).
Hasegawa, K. et al. Selective targeting of multiple myeloma cells with a monoclonal antibody recognizing the ubiquitous protein CD98 heavy chain. Sci. Transl. Med https://doi.org/10.1126/scitranslmed.aax7706 (2022).
Kitamura, T. et al. Efficient screening of retroviral cDNA expression libraries. Proc. Natl. Acad. Sci. U S A 92 , 9146–9150. https://doi.org/10.1073/pnas.92.20.9146 (1995).
Baeuerle, P. A. & Gires, O. EpCAM (CD326) finding its role in cancer. Br. J. Cancer 96 , 417–423. https://doi.org/10.1038/sj.bjc.6603494 (2007).
Laughney, A. M. et al. Regenerative lineages and immune-mediated pruning in lung cancer metastasis. Nat. Med. 26 , 259–269. https://doi.org/10.1038/s41591-019-0750-6 (2020).
Sikkema, L. et al. An integrated cell atlas of the lung in health and disease. Nat. Med. 29 , 1563–1577. https://doi.org/10.1038/s41591-023-02327-2 (2023).
Li, H. et al. Targeting brain lesions of non-small cell lung cancer by enhancing CCL2-mediated CAR-T cell migration. Nat. Commun. 13 , 2154. https://doi.org/10.1038/s41467-022-29647-0 (2022).
Kaira, K. et al. CD98 expression is associated with poor prognosis in resected non-small-cell lung cancer with lymph node metastases. Ann. Surg. Oncol. 16 , 3473–3481. https://doi.org/10.1245/s10434-009-0685-0 (2009).
Kaira, K. et al. Prognostic significance of L-type amino acid transporter 1 (LAT1) and 4F2 heavy chain (CD98) expression in stage I pulmonary adenocarcinoma. Lung. Cancer 66 , 120–126. https://doi.org/10.1016/j.lungcan.2008.12.015 (2009).
Kaira, K. et al. Prognostic significance of L-type amino acid transporter 1 (LAT1) and 4F2 heavy chain (CD98) expression in early stage squamous cell carcinoma of the lung. Cancer Sci. 100 , 248–254. https://doi.org/10.1111/j.1349-7006.2008.01029.x (2009).
Xia, P. & Dubrovska, A. CD98 heavy chain as a prognostic biomarker and target for cancer treatment. Front Oncol. 13 , 1251100. https://doi.org/10.3389/fonc.2023.1251100 (2023).
Nakagawa, T. et al. Identification of glioblastoma-specific antigens expressed in patient-derived tumor cells as candidate targets for chimeric antigen receptor T cell therapy. Neurooncol. Adv. https://doi.org/10.1093/noajnl/vdac177 (2023).
Miyamoto, Y. J., Mitchell, J. S. & McIntyre, B. W. Physical association and functional interaction between beta1 integrin and CD98 on human T lymphocytes. Mol. Immunol. 39 , 739–751. https://doi.org/10.1016/s0161-5890(02)00255-9 (2003).
Prager, G. W., Féral, C. C., Kim, C., Han, J. & Ginsberg, M. H. CD98hc (SLC3A2) interaction with the integrin beta subunit cytoplasmic domain mediates adhesive signaling. J. Biol. Chem. 282 , 24477–24484. https://doi.org/10.1074/jbc.M702877200 (2007).
Zent, R. et al. Class- and splice variant-specific association of CD98 with integrin beta cytoplasmic domains. J. Biol. Chem. 275 , 5059–5064. https://doi.org/10.1074/jbc.275.7.5059 (2000).
Fenczik, C. A., Sethi, T., Ramos, J. W., Hughes, P. E. & Ginsberg, M. H. Complementation of dominant suppression implicates CD98 in integrin activation. Nature 390 , 81–85. https://doi.org/10.1038/36349 (1997).
Feral, C. C. et al. CD98hc (SLC3A2) mediates integrin signaling. Proc. Natl. Acad. Sci. U S A 102 , 355–360. https://doi.org/10.1073/pnas.0404852102 (2005).
Rintoul, R. C. et al. Cross-linking CD98 promotes integrin-like signaling and anchorage-independent growth. Mol. Biol. Cell 13 , 2841–2852. https://doi.org/10.1091/mbc.01-11-0530 (2002).
Devés, R. & Boyd, C. A. Surface antigen CD98(4F2): Not a single membrane protein, but a family of proteins with multiple functions. J. Membr. Biol. 173 , 165–177. https://doi.org/10.1007/s002320001017 (2000).
Verrey, F. System L: Heteromeric exchangers of large, neutral amino acids involved in directional transport. Pflugers Arch. 445 , 529–533. https://doi.org/10.1007/s00424-002-0973-z (2003).
Reynolds, B., Laynes, R., Ogmundsdóttir, M. H., Boyd, C. A. & Goberdhan, D. C. Amino acid transporters and nutrient-sensing mechanisms: New targets for treating insulin-linked disorders?. Biochem. Soc. Trans. 35 , 1215–1217. https://doi.org/10.1042/bst0351215 (2007).
Cho, J. Y. et al. Cynaropicrin, a sesquiterpene lactone, as a new strong regulator of CD29 and CD98 functions. Biochem. Biophys. Res. Commun. 313 , 954–961. https://doi.org/10.1016/j.bbrc.2003.12.026 (2004).
Takeuchi, K. et al. LAT1 expression in non-small-cell lung carcinomas: analyses by semiquantitative reverse transcription-PCR (237 cases) and immunohistochemistry (295 cases). Lung. Cancer 68 , 58–65. https://doi.org/10.1016/j.lungcan.2009.05.020 (2010).
Nakanishi, K. et al. LAT1 expression in normal lung and in atypical adenomatous hyperplasia and adenocarcinoma of the lung. Virchows. Arch. 448 , 142–150. https://doi.org/10.1007/s00428-005-0063-7 (2006).
Zheng, L. et al. A Humanized Lym-1 CAR with Novel DAP10/DAP12 Signaling Domains Demonstrates Reduced Tonic Signaling and Increased Antitumor Activity in B-Cell Lymphoma Models. Clin. Cancer Res. 26 , 3694–3706. https://doi.org/10.1158/1078-0432.Ccr-19-3417 (2020).
Tian, X. et al. An anti-CD98 antibody displaying pH-dependent Fc-mediated tumour-specific activity against multiple cancers in CD98-humanized mice. Nat. Biomed. Eng. 7 , 8–23. https://doi.org/10.1038/s41551-022-00956-5 (2023).
Montero, J. C. et al. An amino acid transporter subunit as an antibody-drug conjugate target in colorectal cancer. J. Exp. Clin. Cancer Res. 42 , 200. https://doi.org/10.1186/s13046-023-02784-0 (2023).
Pellizzari, G. et al. Immunotherapy using IgE or CAR T cells for cancers expressing the tumor antigen SLC3A2. J. Immunother. Cancer https://doi.org/10.1136/jitc-2020-002140 (2021).
Bajaj, J. et al. CD98-Mediated Adhesive Signaling Enables the Establishment and Propagation of Acute Myelogenous Leukemia. Cancer Cell. 30 , 792–805. https://doi.org/10.1016/j.ccell.2016.10.003 (2016).
Hayes, G. M. et al. Antitumor activity of an anti-CD98 antibody. Int. J. Cancer 137 , 710–720. https://doi.org/10.1002/ijc.29415 (2015).
Cantor, J. M. & Ginsberg, M. H. CD98 at the crossroads of adaptive immunity and cancer. J. Cell. Sci. 125 , 1373–1382. https://doi.org/10.1242/jcs.096040 (2012).
Papetti, M. & Herman, I. M. Controlling tumor-derived and vascular endothelial cell growth: role of the 4Ff2 cell surface antigen. Am. J. Pathol. 159 , 165–178. https://doi.org/10.1016/s0002-9440(10)61683-5 (2001).
Yagita, H., Masuko, T. & Hashimoto, Y. Inhibition of tumor cell growth in vitro by murine monoclonal antibodies that recognize a proliferation-associated cell surface antigen system in rats and humans. Cancer Res. 46 , 1478–1484 (1986).
CAS PubMed Google Scholar
Roybal, K. T. et al. Precision Tumor Recognition by T Cells With Combinatorial Antigen-Sensing Circuits. Cell 164 , 770–779. https://doi.org/10.1016/j.cell.2016.01.011 (2016).
Tousley, A. M. et al. Co-opting signalling molecules enables logic-gated control of CAR T cells. Nature 615 , 507–516. https://doi.org/10.1038/s41586-023-05778-2 (2023).
Wilkie, S. et al. Dual targeting of ErbB2 and MUC1 in breast cancer using chimeric antigen receptors engineered to provide complementary signaling. J. Clin. Immunol. 32 , 1059–1070. https://doi.org/10.1007/s10875-012-9689-9 (2012).
Kowolik, C. M. et al. CD28 costimulation provided through a CD19-specific chimeric antigen receptor enhances in vivo persistence and antitumor efficacy of adoptively transferred T cells. Cancer Res. 66 , 10995–11004. https://doi.org/10.1158/0008-5472.Can-06-0160 (2006).
Terakura, S. et al. Generation of CD19-chimeric antigen receptor modified CD8+ T cells derived from virus-specific central memory T cells. Blood 119 , 72–82. https://doi.org/10.1182/blood-2011-07-366419 (2012).
Weber, E. W. et al. Transient rest restores functionality in exhausted CAR-T cells through epigenetic remodeling. Science https://doi.org/10.1126/science.aba1786 (2021).
Download references
The authors thank Y. Kadota (Osaka Habikino Medical Center) and Y. Susaki (Ikeda City Hospital) for clinical samples. They also thank K. Terasaki for technical assistance, and I. Weissman (Stanford University) and T. Kitamura (Tokyo University) for providing materials.
This work was supported in part by the Japan Agency for Medical Research and Development (AMED) (23ama221318h0001 to N.H), the Japan Society for the Promotion of Science (JSPS) KAKENHI (JP19K16799 and JP21K15487 to K.H., and JP19H04810 and JP20H03710 to N.H.), and by grants from the Yasuda Kinen Medical Foundation (to N.H.), the SENSHIN Medical Research Foundation (to N.H.), KAKETSUKEN (to N.H.), the Uehara Memorial Foundation (to N.H.), the Astellas Foundation for Research on Metabolic Disorders (to N.H.), and the Takeda Science Foundation (to N.H.).
Authors and affiliations.
Department of Respiratory Medicine and Clinical Immunology, Graduate School of Medicine, Osaka University, Suita, Osaka, Japan
Moto Yaga, Yuya Shirai, Wibowo Tansri, Shinji Futami, Yujiro Naito, Takayuki Shiroyama, Kotaro Miyake, Shohei Koyama, Haruhiko Hirata, Yoshito Takeda & Atsushi Kumanogoh
Laboratory of Immunopathology, World Premier International Research Center Initiative (WPI), Immunology Frontier Research Center (IFReC), Osaka University, Suita, Osaka, Japan
Moto Yaga & Atsushi Kumanogoh
Laboratory of Cellular Immunotherapy, World Premier International Research Center Initiative (WPI), Immunology Frontier Research Center (IFReC), Osaka University, Suita, Osaka, Japan
Kana Hasegawa & Naoki Hosen
Department of Clinical Laboratory and Biomedical Sciences, Graduate School of Medicine, Osaka University, Suita, Osaka, Japan
Shunya Ikeda, Miwa Matsubara, Hirofumi Uehara & Mana Tachikawa
Department of General Thoracic Surgery, Graduate School of Medicine, Osaka University, Suita, Osaka, Japan
Takashi Hiroshima, Toru Kimura, Soichiro Funaki & Yasushi Shintani
Department of Statistical Genetics, Graduate School of Medicine, Osaka University, Suita, Osaka, Japan
Yuya Shirai
Osaka Research Center for Drug Discovery, Otsuka Pharmaceutical Co., Ltd, Osaka, Japan
Yuzuru Okairi, Masayuki Sone, Hiromi Mori, Yosuke Kogue & Hiroki Akamine
Genome Information Research Center, Research Institute for Microbial Diseases (RIMD), Osaka University, Suita, Osaka, Japan
Daisuke Okuzaki
Laboratory of Human Immunology (Single Cell Genomics), World Premier International Research Center Initiative (WPI), Immunology Frontier Research Center (IFReC), Osaka University, Suita, Osaka, Japan
Department of General Thoracic Surgery, National Hospital Organization Osaka Toneyama Medical Center, Toyonaka, Osaka, Japan
Kotaro Kawagishi, Satoshi Kawanaka, Hiroyuki Yamato & Yukiyasu Takeuchi
Department of Surgery, Takarazuka City Hospital, Takarazuka, Hyogo, Japan
Department of General Thoracic Surgery, Osaka International Cancer Institute, Osaka, Osaka, Japan
Ryu Kanzaki & Jiro Okami
Department of Pathology, Minoh City Hospital, Minoh, Osaka, Japan
Itsuko Nakamichi
Department of Surgery, Minoh City Hospital, Minoh, Osaka, Japan
Shigeru Nakane
Department of Surgery, Toyonaka Municipal Hospital, Toyonaka, Osaka, Japan
Aki Kobayashi & Takashi Iwazawa
Department of General Thoracic Surgery, National Hospital Organization Kinki-Chuo Chest Medical Center, Sakai, Osaka, Japan
Toshiteru Tokunaga
Department of Surgery, Suita Municipal Hospital, Suita, Osaka, Japan
Hideoki Yokouchi
Department of Thoracic Oncology, National Hospital Organization Osaka Toneyama Medical Center, Toyonaka, Osaka, Japan
Yukihiro Yano, Junji Uchida & Masahide Mori
Department of Internal Medicine, Osaka Anti-Tuberculosis Association Osaka Fukujuji Hospital, Neyagawa, Osaka, Japan
Kiyoshi Komuta
Department of Neurosurgery, Graduate School of Medicine, Osaka University, Suita, Osaka, Japan
Tetsuro Tachi, Hideki Kuroda, Noriyuki Kijima & Haruhiko Kishima
Department of Hematology and Oncology, Osaka University Graduate School of Medicine, 2-2, Yamada-Oka, Suita, Osaka, 565-0871, Japan
Michiko Ichii & Naoki Hosen
Department of Immunology and Molecular Medicine, Graduate School of Medicine, Osaka University, Suita, Osaka, Japan
Shohei Koyama
Division of Cancer Immunology, Research Institute/Exploratory Oncology Research and Clinical Trial Center (EPOC), National Cancer Center, Tokyo/Chiba, Japan
Integrated Frontier Research for Medical Science Division, Institute for Open and Transdisciplinary Research Initiatives (OTRI), Osaka University, Suita, Osaka, Japan
Atsushi Kumanogoh
Center for Infectious Diseases for Education and Research (CiDER), Osaka University, Suita, Osaka, Japan
Japan Agency for Medical Research and Development – Core Research for Evolutional Science and Technology (AMED–CREST), Osaka University, Suita, Osaka, Japan
Center for Advanced Modalities and DDS (CAMaD), Osaka University, Suita, Osaka, Japan
Integrated Frontier Research for Medical Science Division, Institute for Open and Transdisciplinary Research Initiatives, Osaka University, Suita, Osaka, Japan
Naoki Hosen
You can also search for this author in PubMed Google Scholar
M.Y., K.H., and N.H. designed the experiments; M.Y., K.H., S.I., W.T., M. Matsubara, T.H., H.U., M.T., Y.O., M.S., H.M., Y.K., H.A. and D.O. performed the experiments; M.Y., W.T., M. Matsubara, H.U., M.T., T.H., T. Kimura, K. Kawagishi, S. Kawanaka, H. Yamato, Y. Takeuchi, E.O., T. Kanzaki, J.O., I.N, S.N., A. Kobayashi, T.I., T. Tokunaga, H. Yokouchi, Y.Y, J.U., M. Mori, K. Komuta, T. Tachi, H. Kuroda, N.K., H. Kishima, S.F., Y.N., T.S., K.M., S. Koyama, H.H., Y. Takeda, and Y.S. collected and analyzed clinical samples; M.Y., K.H., M. Matsubara, Y.S., D.O., and N.H. analyzed the data; M.Y., D.O., A. Kumanogoh, and N.H. wrote the manuscript; and all authors reviewed and approved the final version of the manuscript.
Correspondence to Naoki Hosen .
Competing interests.
Naoki Hosen and Atsushi Kumanogoh: a patent application has been filed on R8H283. All other authors don't have any competing interest.
Publisher's note.
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Supplementary information., rights and permissions.
Open Access This article is licensed under a Creative Commons Attribution-NonCommercial-NoDerivatives 4.0 International License, which permits any non-commercial use, sharing, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if you modified the licensed material. You do not have permission under this licence to share adapted material derived from this article or parts of it. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by-nc-nd/4.0/ .
Reprints and permissions
Cite this article.
Yaga, M., Hasegawa, K., Ikeda, S. et al. CD98 heavy chain protein is overexpressed in non-small cell lung cancer and is a potential target for CAR T-cell therapy. Sci Rep 14 , 17917 (2024). https://doi.org/10.1038/s41598-024-68779-9
Download citation
Received : 08 February 2024
Accepted : 29 July 2024
Published : 02 August 2024
DOI : https://doi.org/10.1038/s41598-024-68779-9
Anyone you share the following link with will be able to read this content:
Sorry, a shareable link is not currently available for this article.
Provided by the Springer Nature SharedIt content-sharing initiative
By submitting a comment you agree to abide by our Terms and Community Guidelines . If you find something abusive or that does not comply with our terms or guidelines please flag it as inappropriate.
Sign up for the Nature Briefing: Cancer newsletter — what matters in cancer research, free to your inbox weekly.
You are accessing a machine-readable page. In order to be human-readable, please install an RSS reader.
All articles published by MDPI are made immediately available worldwide under an open access license. No special permission is required to reuse all or part of the article published by MDPI, including figures and tables. For articles published under an open access Creative Common CC BY license, any part of the article may be reused without permission provided that the original article is clearly cited. For more information, please refer to https://www.mdpi.com/openaccess .
Feature papers represent the most advanced research with significant potential for high impact in the field. A Feature Paper should be a substantial original Article that involves several techniques or approaches, provides an outlook for future research directions and describes possible research applications.
Feature papers are submitted upon individual invitation or recommendation by the scientific editors and must receive positive feedback from the reviewers.
Editor’s Choice articles are based on recommendations by the scientific editors of MDPI journals from around the world. Editors select a small number of articles recently published in the journal that they believe will be particularly interesting to readers, or important in the respective research area. The aim is to provide a snapshot of some of the most exciting work published in the various research areas of the journal.
Original Submission Date Received: .
Find support for a specific problem in the support section of our website.
Please let us know what you think of our products and services.
Visit our dedicated information section to learn more about MDPI.
Is there a link between chronic obstructive pulmonary disease and lung adenocarcinoma a clinico-pathological and molecular study.
2. materials and methods, 2.1. study design and population, 2.2. morphological analyses, 2.3. dna extraction, 2.4. next generation sequencing, 2.5. bioinformatics analysis.
3.1. study population, 3.2. distribution of mutations in copd, smokers and non smokers tumor samples, 3.3. analysis of matched pathological/healthy tissues, 4. discussion, supplementary materials, author contributions, institutional review board statement, informed consent statement, data availability statement, acknowledgments, conflicts of interest.
Click here to enlarge figure
Characteristics | COPD/Smoker LUAD N = 38 | Non-COPD/Smoker LUAD N = 54 | Non-COPD/Non-Smoker LUAD N = 18 | p-Value | q-Value |
---|---|---|---|---|---|
Sex | 0.001 | 0.015 | |||
males | 30 (79%) | 34 (63%) | 5 (28%) | ||
females | 8 (21%) | 20 (37%) | 13 (72%) | ||
Age | 72 (68, 75) | 69 (62, 73) | 65 (59, 72) | 0.11 | 0.3 |
GOLD stage | >0.9 | >0.9 | |||
I | 18 (47.5%) | 0 (NA%) | 0 (NA%) | ||
II | 18 (47.5%) | 0 (NA%) | 0 (NA%) | ||
III | 2 (5%) | 0 (NA%) | 0 (NA%) | ||
Smoking history (pack years) | 40 (27.5–50) | 36 (22–50) | - | 0.245 | 0.67 |
FEV1 (% of predict) | 76 (64, 87) | 93 (88, 113) | 117 (104, 130) | <0.001 | <0.001 |
FEV1/FV (% of predict) | 67 (58, 70) | 83 (78, 85) | 82 (78, 86) | <0.001 | <0.001 |
FVC (% of predict) | 92 (74, 101) | 92 (82, 105) | 119 (108, 128) | <0.001 | 0.002 |
DLCO/VA (%) | 70 (48, 89) | 80 (65, 91) | 84 (81, 99) | 0.085 | 0.2 |
SUV | 6 (2, 11) | 6 (2, 12) | 2 (1, 5) | 0.061 | 0.2 |
WBC (n × 10 /L) | 7.35 (5.78, 8.68) | 6.92 (5.45, 9.30) | 5.52 (4.70, 7.01) | 0.012 | 0.072 |
RBC (n × 10 /L) | 4.50 (4.18, 5.00) | 4.42 (4.14, 4.77) | 4.62 (4.21, 4.96) | 0.7 | 0.8 |
HgB (g/dL) | 13.70 (12.35, 14.65) | 13.70 (12.50, 14.70) | 13.80 (13.03, 14.80) | 0.7 | 0.8 |
Neutrophils (n × 10 /L) | 4.40 (3.54, 5.74) | 4.02 (3.06, 6.02) | 2.87 (2.81, 4.52) | 0.046 | 0.2 |
Neutrophils (%) | 64 (56, 69) | 62 (56, 69) | 56 (52, 66) | 0.3 | 0.4 |
Lymphocytes (n × 10 /L) | 1.77 (1.48, 2.12) | 1.75 (1.55, 2.06) | 1.62 (1.39, 1.93) | 0.7 | 0.8 |
Lymphocytes (%) | 24 (20, 32) | 26 (20, 32) | 31 (24, 33) | 0.3 | 0.4 |
Monocytes (n × 10 /L) | 0.60 (0.49, 0.70) | 0.57 (0.42, 0.70) | 0.49 (0.40, 0.58) | 0.082 | 0.2 |
Monocytes (%) | 8.55 (7.03, 10.05) | 8.00 (7.10, 9.80) | 9.30 (6.90, 10.50) | 0.7 | 0.8 |
Eosinophils (n × 10 /L) | 0.16 (0.06, 0.26) | 0.08 (0.05, 0.16) | 0.12 (0.06, 0.26) | 0.079 | 0.2 |
Eosinophils (%) | 2.05 (1.33, 3.63) | 1.30 (0.80, 2.30) | 2.20 (1.30, 3.80) | 0.029 | 0.13 |
Basophils (n × 10 /L) | 0.030 (0.020, 0.040) | 0.020 (0.010, 0.030) | 0.020 (0.020, 0.030) | 0.13 | 0.3 |
Basophils (%) | 0.40 (0.30, 0.60) | 0.30 (0.20, 0.50) | 0.40 (0.30, 0.60) | 0.2 | 0.3 |
ESR (mm/h) | 21 (12, 32) | 20 (10, 28) | 14 (9, 20) | 0.090 | 0.2 |
CRP (mg/L) | 2.0 (1.4, 6.9) | 2.9 (1.3, 4.7) | 0.8 (0.3, 2.9) | 0.029 | 0.13 |
Clinical stage | 0.2 | 0.3 | |||
IA | 7 (18.5%) | 17 (31.5%) | 6 (33%) | ||
IB | 16 (42%) | 11 (20.5%) | 9 (50%) | ||
IIA | 4 (10.5%) | 3 (6%) | 0 (0%) | ||
IIB | 7 (18.5%) | 10 (18%) | 2 (11%) | ||
IIIA | 4 (10.5%) | 10 (18%) | 1 (6%) | ||
IIIB | 0 (0%) | 3 (6%) | 0 (0%) |
Characteristic | COPD/Smoker LUAD N = 38 | Non-COPD/Smoker LUAD N = 54 | Non-COPD/Non-Smoker LUAD N = 18 | p-Value | q-Value |
---|---|---|---|---|---|
Tumor cells (%) | 70 (50, 80) | 65 (50, 80) | 70 (70, 80) | 0.5 | 0.7 |
Prevalent pattern | 0.3 | 0.4 | |||
Lepidic pattern | 1 (2.5%) | 8 (15%) | 1 (5.6%) | ||
Acinar pattern | 28 (74%) | 32 (59%) | 14 (78%) | ||
Papillary pattern | 1 (2.5%) | 4 (7.5%) | 0 (0%) | ||
Solid pattern | 8 (21%) | 10 (18.5%) | 3 (17%) | ||
Lepidic pattern (%) | 0 (0, 5) | 2 (0, 24) | 18 (5, 30) | 0.008 | 0.057 |
Acinar pattern (%) | 60 (42, 85) | 50 (20, 75) | 60 (45, 84) | 0.14 | 0.3 |
Papillary pattern (%) | 0 (0, 0) | 0 (0, 8) | 0 (0, 8) | 0.7 | 0.8 |
Micropapillary pattern (%) | 0 (0, 2) | 0 (0, 1) | 0 (0, 4) | >0.9 | >0.9 |
Solid pattern (%) | 10 (0, 35) | 0 (0, 32) | 0 (0, 0) | 0.2 | 0.3 |
WHO Grading | |||||
1 | 1 (2%) | 5 (9%) | 1 (5%) | ||
2 | 17 (45%) | 21 (39%) | 14 (78%) | ||
3 | 20 (53%) | 28 (52%) | 3 (17%) | ||
MIB1 (%) | 20 (10, 58) | 40 (20, 70) | 10 (9, 21) | 0.001 | 0.011 |
Necrosis (%) | 0.2 | 0.3 | |||
0 | 6 (16%) | 12 (22%) | 7 (39%) | ||
≤10% | 19 (50%) | 25 (46%) | 7 (39%) | ||
11–30% | 4 (10%) | 7 (13%) | 4 (22%) | ||
>30% | 9 (24%) | 10 (19%) | 0 (0%) | ||
Inflammation (%) | 0.2 | 0.3 | |||
0 | 0 (0%) | 0 (0%) | 1 (6%) | ||
≤10% | 13 (34%) | 24 (44%) | 4 (22%) | ||
11–30% | 18 (47.5%) | 23 (43%) | 11 (61%) | ||
>30% | 7 (18.5%) | 7 (13%) | 2 (11%) | ||
Fibrosis (%) | 0.2 | 0.3 | |||
0 | 3 (8%) | 4 (7%) | 0 (0%) | ||
≤10% | 16 (42%) | 22 (41%) | 3 (17%) | ||
11–30% | 12 (32%) | 18 (33%) | 8 (44%) | ||
>30% | 7 (18%) | 10 (19%) | 7 (39%) | ||
Vascular invasion | 0.5 | 0.7 | |||
No | 18 (47%) | 23 (43%) | 9 (50%) | ||
Yes | 20 (53%) | 31 (57%) | 9 (50%) | ||
Pleural invasion | 0.6 | 0.8 | |||
No | 20 (53%) | 27 (50%) | 11 (61%) | ||
Yes | 18 (47%) | 27 (50%) | 7 (39%) | ||
Type of visceral pleura invasion | >0.9 | >0.9 | |||
PL0 | 20 (53%) | 27 (50%) | 11 (61%) | ||
PL1 | 15 (39%) | 23 (43%) | 6 (33%) | ||
PL2 | 3 (8%) | 4 (7%) | 1 (6%) | ||
Perineural invasion | 0.5 | 0.7 | |||
No | 34 (89%) | 50 (93%) | 18 (100%) | ||
Yes | 4 (11%) | 4 (7%) | 0 (0%) | ||
Lymph node invasion | 0.089 | 0.2 | |||
No | 30 (79%) | 38 (70%) | 17 (94%) | ||
Yes | 8 (21%) | 16 (30%) | 1 (6%) | ||
Type of lymph node invasion | 0.14 | 0.3 | |||
0 | 30 (79%) | 38 (70%) | 17 (94%) | ||
1 | 6 (16%) | 7 (13%) | 1 (6%) | ||
2 | 2 (5%) | 9 (17%) | 0 (0%) |
Gene | COPD/Smoker LUAD N = 38 | Non-COPD/Smoker LUAD N = 54 | Non-COPD/Non-Smoker LUAD N = 18 |
---|---|---|---|
KRAS | 21 (55%) | 27 (50%) | 3 (17%) |
EGFR | 9 (24%) | 12 (22%) | 10 (56%) |
NTRK3 | 8 (21%) | 14 (26%) | 5 (28%) |
TP53 | 10 (26%) | 15 (28%) | 4 (22%) |
NTRK2 | 8 (21%) | 10 (19%) | 1 (6%) |
PIK3CA | 7 (18%) | 9 (17%) | 0 (0%) |
STK11 | 9 (24%) | 13 (24%) | 1 (6%) |
MET | 5 (13%) | 10 (19%) | 0 (0%) |
NOTCH1 | 6 (16%) | 11 (20%) | 2 (11%) |
FBXW7 | 5 (13%) | 5 (9%) | 0 (0%) |
The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |
Lunardi, F.; Nardo, G.; Lazzarini, E.; Tzorakoleftheraki, S.-E.; Comacchio, G.M.; Fonzi, E.; Tebaldi, M.; Vedovelli, L.; Pezzuto, F.; Fortarezza, F.; et al. Is There a Link between Chronic Obstructive Pulmonary Disease and Lung Adenocarcinoma? A Clinico-Pathological and Molecular Study. J. Pers. Med. 2024 , 14 , 839. https://doi.org/10.3390/jpm14080839
Lunardi F, Nardo G, Lazzarini E, Tzorakoleftheraki S-E, Comacchio GM, Fonzi E, Tebaldi M, Vedovelli L, Pezzuto F, Fortarezza F, et al. Is There a Link between Chronic Obstructive Pulmonary Disease and Lung Adenocarcinoma? A Clinico-Pathological and Molecular Study. Journal of Personalized Medicine . 2024; 14(8):839. https://doi.org/10.3390/jpm14080839
Lunardi, Francesca, Giorgia Nardo, Elisabetta Lazzarini, Sofia-Eleni Tzorakoleftheraki, Giovanni Maria Comacchio, Eugenio Fonzi, Michela Tebaldi, Luca Vedovelli, Federica Pezzuto, Francesco Fortarezza, and et al. 2024. "Is There a Link between Chronic Obstructive Pulmonary Disease and Lung Adenocarcinoma? A Clinico-Pathological and Molecular Study" Journal of Personalized Medicine 14, no. 8: 839. https://doi.org/10.3390/jpm14080839
Article access statistics, supplementary material.
ZIP-Document (ZIP, 117 KiB)
Mdpi initiatives, follow mdpi.
Subscribe to receive issue release notifications and newsletters from MDPI journals
IMAGES
COMMENTS
Introduction. Lung cancer, a highly invasive, rapidly metastasizing and prevalent cancer, is the top killer cancer in both men and women in the United States of America (USA). ... Medical research council lung cancer working party. Br J Cancer. 1993; 68 (6):1157-1166. [PMC free article] [Google Scholar] 125. Scagliotti G, Hanna N, Fossella F ...
Lung cancer is one of the most frequently diagnosed cancers and the leading cause of cancer-related deaths worldwide with an estimated 2 million new cases and 1·76 million deaths per year. Substantial improvements in our understanding of disease biology, application of predictive biomarkers, and refinements in treatment have led to remarkable progress in the past two decades and transformed ...
The review paper accounts for the novel and current immunotherapy and targeted therapy available for lung cancer in clinical trials and in the research phases in depth. Special attention is being paid to mark out single or multiple genes that are required for malignancy and survival while developing targeted therapies for lung cancer treatment.
The Lung Cancer Master Protocol, or Lung-MAP, is a precision medicine research study for people with advanced non-small cell lung cancer that has continued to grow after treatment. Patients are assigned to different study drug combinations based on the results of genomic profiling of their tumors.
Introduction. The diagnosis and treatment of lung cancer became a hot spot of research due to the disease's increased morbidity and mortality in recent years. Therefore, more and more novel diagnosis methods for early-stage lung cancer and plenty of clinical trials have been conducted. ... This paper was included in the recommended literature ...
1. Introduction. Lung cancer had been diagnosed in 2.09 million people and resulted in 1.76 million deaths worldwide in 2018, and it's the most common cause of cancer-related death in men and the second most common in women (Bray et al., 2018).It was estimated that 228,820 new cases were diagnosed and 135,720 deaths were resulted from lung cancer in 2020 in the United States (Howlader NKrapcho ...
1.1 Introduction. Worldwide, there are many different cancers that affect people, most of which are characterised by uncontrolled cell growth and the spread of cancerous cells into neighbouring local tissues. Lung cancer has been one of the most frequently diagnosed cancer types in the globe for many years.
However, developments over the past 10 years have dramatically improved the outlook for lung cancer through advances in prevention, screening, early diagnosis, and therapy. The advent of screening has underscored the importance of smoking cessation for improving lung cancer outcomes. Analysis of National Lung Screening Trial data showed that 7 ...
Lung cancer. Alesha A Thai, Benjamin J Solomon, Lecia V Sequist, Justin F Gainor, Rebecca S Heist. Lung cancer is one of the most frequently diagnosed cancers and the leading cause of cancer-related deaths worldwide with an estimated 2 million new cases and 1·76 million deaths per year. Substantial improvements in our understanding of disease ...
Lung cancer, a multifaceted disease, demands tailored therapeutic approaches due to its diverse subtypes and stages. This comprehensive review explores the intricate landscape of lung cancer research, delving into recent breakthroughs and their implications for diagnosis, therapy, and prevention. Genomic profiling and biomarker identification have ushered in the era of personalised medicine ...
MeSH terms. Lung cancer is one of the most frequently diagnosed cancers and the leading cause of cancer-related deaths worldwide with an estimated 2 million new cases and 1·76 million deaths per year. Substantial improvements in our understanding of disease biology, application of predictive biomarkers, and ref ….
An outline for future research in lung cancer survival prediction. ... Introduction. Lung cancer is the leading cause of cancer-related deaths, ... In this paper, we review and explore studies using machine learning to predict lung cancer survival based on clinical data. It provides an overview of data sources, patient cohort, data pre ...
Introduction. In 2013 it was estimated that there were 14.9 million incident cancer cases and 8.2 million deaths. 1 Of these cases, lung cancer remains the leading cause of cancer death globally in both high-income and low- and middle-income countries, accounting for 1.6 million cancer deaths annually (approximately 20% of total cancer deaths), with an estimated 1.8 million new cases annually ...
Lung cancer is broadly categorized into small-cell and non-small-cell histologies; non-small-cell lung cancer (NSCLC) comprises >85% of all cases and can be further classified by histological subtype.
Lung cancer is the primary cause of cancer death worldwide, with 2.09 million new cases and 1.76 million people dying from lung cancer in 2018 1.Four case-controlled studies from Japan reported in ...
Lung cancer is thought to be a genetic disease with a variety of unknown origins. Globocan2020 report tells in 2020 new cancer cases identified was 19.3 million and nearly 10.0 million died owed ...
Introduction. During the last one hundred years, our understanding of the biology of cancer increased in an extraordinary way. 1-4 Such a progress has been particularly prompted during the last few decades because of technological and conceptual progress in a variety of fields, including massive next-generation sequencing, inclusion of "omic" sciences, high-resolution microscopy, molecular ...
Lung cancer, a fierce adversary in the realm of oncology, stands as a leading cause of cancer-related mortality worldwide. Despite significant strides in medical research and treatment modalities, the complexity of lung cancer poses a continuous challenge, demanding innovative approaches for early and accurate diagnosis.
Worldwide in 2012 lung cancer resulted in 1.6 million. deaths. Risk factors include smoking, exposure to radon gas, asbestos, second-hand smoke, air pollution, and. geneticfactors. Pathogenesis is ...
Joseph A. Greer, MD, associate professor of psychology at Harvard Medical School and codirector of the Cancer Outcomes Research and Education Program at the Massachusetts General Hospital Cancer Center, presented the results of the REACH PC trial, in which 1250 patients with advanced NSCLC were randomized 1:1 to monthly palliative care visits ...
INTRODUCTION. Advances in lung cancer detection by computed tomography (CT) scan have increased the frequency of detecting multiple pulmonary tumor nodules in clinical practice, accounting for approximately 10% of all surgically removed lung cancers. 1, 2 When patients are found to have multiple primary lung lesions, determining the lineage relationship is necessary for staging disease and ...
Lung cancer is a significant public health concern, causing a considerable number of deaths globally. GLOBOCAN 2020 estimates of cancer incidence and mortality produced by the International Agency for Research on Cancer (IARC) show as lung cancer remains the leading cause of cancer death, with an estimated 1.8 million deaths (18%) in 2020.
Introduction. Lung cancer is one of the most frequently diagnosed cancers and the leading cause of cancer deaths worldwide. About 2.20 million new patients are diagnosed with lung cancer each year [1], and 75% of them die within five years of diagnosis [2].High intra-tumor heterogeneity (ITH) and complexity of cancer cells giving rise to drug resistance make cancer treatment more challenging [3].
Abstract. Comprehensive N6-methyladenosine (m6A) epitranscriptomic profiling of primary tumors remains largely uncharted. Here, we profiled the m6A epitranscriptome of 10 nonneoplastic lung tissues and 51 lung adenocarcinoma (LUAD) tumors, integrating the corresponding transcriptomic, proteomic, and extensive clinical annotations. We identified distinct clusters and genes that were exclusively ...
Funding: A.Q.-V. was supported by NIH grant # T32 CA1600001, by the American Lung Association, and by the Druckenmiller Center for Lung Cancer Research. These studies were supported by NCI R01 CA264078, NCI R35 CA263816, NCI U24 CA213274, and by the Druckenmiller Center for Lung Cancer Research.
The pooled sensitivity and specificity of deep learning approaches for lung cancer detection were 93% and 68%, respectively. The results showed that AI plays an important role in medical imaging, but there are still many research challenges. Go to: This study extensively surveys papers published between 2014 and 2022.
The research team used advanced technology and worked together with other experts to make these breakthroughs. The breath test for lung cancer is especially exciting because it could be a simple way to check for cancer without invasive procedures. It might even help find other types of cancer in the future.
Non-small cell lung cancer (NSCLC) is one of the most common causes of cancer deaths worldwide 1,2. Although immune checkpoint blockade therapy has largely improved the prognosis of NSCLC 3 , 4 ...
Chronic Obstructive Pulmonary Disease (COPD) and lung cancer are strictly related. To date, it is unknown if COPD-associated cancers are different from the tumors of non-COPD patients. The main goal of the study was to compare the morphological/molecular profiles of lung adenocarcinoma (LUAD) samples of COPD, non-COPD/smokers and non-COPD/non-smokers, and to investigate if a genetic ...
Keywords. 1.1. Introduction. Cancer is the uninhibited growth and development of abnormal cells in the body, and is one of the foremost reasons of deaths throughout the world ( Paul and Jindal, 2017 ). These abnormal cells are commonly designated as cancerous cells, tumorous cells, or malignant cells. In 2018, cancer accounted for an estimated ...