ABSTRACT
We developed a surgical support system that visualises important microanatomies using artificial intelligence (AI). This study evaluated its accuracy in recognising the thoracic nerves during lung cancer surgery. Recognition models were created with deep learning using images precisely annotated for nerves. Computational evaluation was performed using the Dice index and the Jaccard index. Four general thoracic surgeons evaluated the accuracy of nerve recognition. Further, the differences in time lag, image quality and smoothness of movement between the AI system and surgical monitor were assessed. Ratings were made using a five-point scale. The computational evaluation was relatively favourable, with a Dice index of 0.56 and a Jaccard index of 0.39. The AI system was used for 10 thoracoscopic surgeries for lung cancer. The accuracy of thoracic nerve recognition was satisfactory, with a recall score of 4.5 ± 0.4 and a precision score of 4.0 ± 0.9. Though smoothness of motion (3.2 ± 0.4) differed slightly, nearly no difference in time lag (4.9 ± 0.3) and image quality (4.6 ± 0.5) between the AI system and the surgical monitor were observed. In conclusion, the AI surgical support system has a satisfactory accuracy in recognising the thoracic nerves.
Subject(s)
Artificial Intelligence , Thoracic Nerves , Humans , Lung Neoplasms/surgery , Deep Learning , Surgery, Computer-Assisted/methods , Image Processing, Computer-Assisted/methodsABSTRACT
Background: In early-stage non-small cell lung cancer (NSCLC), the presence of a ground-glass opacity (GGO) component in the primary lesion on high-resolution computed tomography (CT) is recognized as a favorable prognostic factor. Even in NSCLC with a GGO component, lymph node metastases are occasionally detected during or after surgery. However, the prognostic impact of GGO components in these patients has not been clarified. We aimed to examine the prognostic significance of GGO components as radiological findings of primary lesions of completely resected NSCLC with pathological nodal involvement. Methods: This study included 290 patients (11%) with pathological nodal involvement among 2,546 patients who underwent complete resection of NSCLC at our institution. Patients with an unknown primary lesion (T0) or centrally located lung cancer were excluded. The 290 patients were divided into two groups [i.e., the part-solid ("PS") and "Solid" groups] according to the radiological findings of the primary lesion, and their clinicopathological characteristics and prognoses were compared. Furthermore, a multivariate analysis was performed using the Cox proportional hazards model to examine the factors affecting the overall survival (OS). Results: The OS in the PS group (n=58) was significantly longer than that in the Solid group (n=232; P=0.039). However, multivariate analysis only revealed age [hazard ratio (HR) =1.77; 95% confidence interval (CI): 1.15-2.72] and the clinical T factor (HR =1.58; 95% CI: 1.01-2.47), but not the radiological findings of primary lesions, as the independent prognostic factors. Furthermore, the OS did not differ significantly between the PS and Solid groups matched for the clinical T and N factors (n=58 patients each). Conclusions: GGO components in the primary lesion, considered a decisive prognostic factor in early-stage NSCLC, did not affect the prognosis of patients with NSCLC and pathological nodal involvement.
ABSTRACT
In this era of endoscopic surgery, feedback from recorded surgical videos is useful and efficient; therefore, effective methods of obtaining this feedback are needed. We analyzed surgical videos using motion analysis software and verified the usefulness of visualizing and objectively evaluating surgical procedures. We measured the grasping and traction angles of the vascular sheath when using forceps and the trajectory of the forceps tip for the upper pulmonary vein during right upper lobectomy during video-assisted thoracoscopic surgery performed by three trainers and trainees. Compared with the trainers, the trainees exhibited insufficient traction of the vascular sheath, performed many slow and unnecessary manipulations, and sometimes performed sudden and fast movements. By visualizing the surgical procedures, the trainee will be better able to identify dangerous or futile movements. It may also make it easier to objectively recognize improvements in one's technique. Motion analysis software could allow for efficient surgical education and self-learning.
Subject(s)
Software , Thoracic Surgery, Video-Assisted , Humans , Thoracic Surgery, Video-Assisted/methods , Video Recording , Pneumonectomy/methods , Clinical CompetenceABSTRACT
INTRODUCTION: NUT carcinoma is a rare cancer associated with a poor prognosis. Because of its rarity, its diagnosis is challenging and is usually made by excluding other diagnoses. Immunohistochemical analysis is a reliable technique that contributes to a correct diagnosis, but overestimating the expression of neuroendocrine (NE) markers may result in an incorrect diagnosis. In this study, we established the immunohistochemical phenotypes of NUT carcinoma compared with tumors that mimic its phenotype to identify potential diagnostic pitfalls. METHODS: Eight cases of NUT carcinoma were examined along with eight basaloid squamous cell carcinomas and thirteen cases of small cell carcinoma using an immunohistochemical panel consisting of various antibodies. RESULTS: Of the eight NUT carcinomas, three patients had a smoking history. All the cases examined for INSM1 were positive (6/6, 100%), although the staining was somewhat weak. Among the NE markers, synaptophysin was variably positive in two NUT carcinomas (2/6, 33%); however, all cases were negative for ASCL1, chromogranin A, and CD56. Moreover, the squamous cell markers, p40 and CK5/6, were weakly expressed in 4/6 (67%) and 3/6 (50%) of the NUT carcinomas, respectively. CONCLUSIONS: For tumors with an ambiguous morphology, applying the neuroendocrine phenotype of NUT carcinoma may be misleading; particularly, when distinguishing it from small-cell carcinoma. Similarly, null or weak expression of squamous cell markers may be observed in NUT carcinoma, but this differs from squamous cell carcinoma, which consistently demonstrates strong positivity for squamous cell markers.
Subject(s)
Carcinoma, Neuroendocrine , Carcinoma, Squamous Cell , Lung Neoplasms , Small Cell Lung Carcinoma , Humans , Biomarkers, Tumor/analysis , Synaptophysin/analysis , Carcinoma, Squamous Cell/pathology , Lung Neoplasms/pathology , Epithelial Cells/pathology , Phenotype , Carcinoma, Neuroendocrine/pathology , Repressor Proteins/analysisABSTRACT
BACKGROUND: The efficacy of local therapies for lung cancer patients with postoperative oligo-recurrence has been reported. However, whether local therapies should be chosen over molecular targeted therapies for oligo-recurrence patients with driver mutations remains controversial. Therefore, we aimed to investigate the optimal initial treatment strategy for oligo-recurrence in lung cancer patients with driver mutations. METHODS: Among 2152 patients with lung adenocarcinoma who underwent surgical resection at our institute between 2008 and 2020, 66 patients with driver mutations who experienced cancer oligo-recurrence after surgery and were treated with local or molecularly targeted therapy as an initial therapy after recurrence were evaluated. Oligo-recurrence was characterized by the presence of 1 to 3 recurrent lesions. These patients were investigated, focusing on their post-recurrence therapies and prognoses. RESULTS: The median follow-up period was 71 months. Local and molecular targeted therapies were administered to 41 and 25 patients, respectively. The number of recurrence lesions tended to be lower in the initial local therapy group than in the molecular targeted therapy group. In the initial local therapy group, 23 patients (56%) subsequently received molecular targeted therapies. The time from recurrence to the initiation of molecular targeted therapy was significantly longer in the local therapy group than in the molecular targeted therapy group (p < 0.001). There was no significant difference in post-recurrence overall survival (hazard ratio, 1.429; 95% confidence interval, 0.701-2.912; log-rank, p = 0.324) and post-recurrence progression-free survival (hazard ratio, 0.799; 95% confidence interval, 0.459-1.390; log-rank, p = 0.426) in the initial local ablative therapy group compared with the initial molecular targeted therapy group. CONCLUSIONS: Local therapies as a first-line treatment did not show statistically significant differences in post-recurrence survival or progression-free survival compared with molecular targeted therapies. However, local therapies as an initial treatment should be considered preferably, as they can cure the recurrence and can delay the start of administration of molecular targeted therapies.
ABSTRACT
OBJECTIVE: We investigated if PD-L1 expression can be predicted by machine learning using clinical and imaging features. METHODS: We included 117 patients with c-stage I/II non-small cell lung cancer who underwent radical resection. A total of 3951 radiomic features were extracted by defining the tumor (within tumor contour), rim (contour ±3 mm) and exterior (contour +10 mm) on preoperative contrast computed tomography. After feature selection by Boruta algorithm, prediction models of tumor PD-L1 expression (22C3: ≥1%, <1%) of resected specimens were constructed using Random Forest: radiomics, clinical, and combined models. Their performance was evaluated by 5-fold cross-validation, and AUCs were compared using Delong test. Next, study groups were categorized as patients without biopsy (training set), and those with biopsy (test set). Predictive ability of biopsy was compared to each prediction model. RESULTS: Of 117 patients (66 ± 10 years old, 48% male), 33 (28.2%) had PD-L1≥1%. Mean AUC of PD-L1≥1% for the validation set in radiomics, clinical, and combined models were 0.80, 0.80, and 0.83 (P = .32 vs. clinical model), respectively. The diagnosis of malignancy was made in 22 of 38 (58%) patients with attempted biopsies, and PD-L1 was measurable in 19 of 38 (50%) patients. Diagnostic accuracies of PD-L1≥1% from 19 determinable biopsies and 38 all attempted biopsies were 0.68 and 0.34, respectively. These were out performed by machine learning: 0.71, 0.71, and 0.74 for radiomics, clinical, and combined models, respectively. CONCLUSIONS: Our machine learning could be an adjunctive tool in estimating PD-L1 expression prior to neoadjuvant treatment, particularly when PD-L1 is indeterminable with biopsy.
Subject(s)
Carcinoma, Non-Small-Cell Lung , Lung Neoplasms , Aged , Female , Humans , Male , Middle Aged , B7-H1 Antigen/metabolism , Biopsy , Carcinoma, Non-Small-Cell Lung/diagnosis , Carcinoma, Non-Small-Cell Lung/surgery , Carcinoma, Non-Small-Cell Lung/drug therapy , Lung Neoplasms/diagnosis , Lung Neoplasms/surgery , Lung Neoplasms/drug therapy , Retrospective Studies , Tomography, X-Ray ComputedABSTRACT
OBJECTIVE: Pancreatic cancer with lung oligometastasis may have favourable overall survival. The aim of this study was to evaluate outcomes of pancreatic cancer with lung oligometastases including both synchronous and metachronous metastases. METHODS: Consecutive pancreatic cancer patients with lung metastasis treated at our institution between February 2015 and December 2021 were identified from our prospectively maintained database. Clinical characteristics and outcomes were compared and analysed according to the extent of lung metastases. Predictors for overall survival were analysed using the Cox proportional hazards model. RESULTS: A totoal of 171 patients were included (oligometastasis/polymetastasis/multi-organ metastasis: 34/50/87). Patients with oligometastases were more likely to undergo surgical resection (41% vs. 0% vs. 2%) and showed a longer median overall survival (41.3 vs. 17.6 vs. 13.1 months) compared with those with other types of metastases. Oligometastasis (hazard ratio, 0.43; 95% confidence interval, 0.24-0.76; P = 0.004) was identified as an independent factor predicting favourable overall survival in patients with lung-only metastasis. Disease status (synchronous vs. metachronous) was not associated with survival in patients with oligometastasis (29.4 vs. 41.3 months, P = 0.527) and polymetastasis (17.9 vs. 16.7 months, P = 0.545). Selected patients who underwent surgical resection showed a median overall survival of 52.7 months. CONCLUSIONS: Patients with lung oligometastases presented a favourable prognosis. Surgical resection in selected patients was associated with a long median overall survival.
Subject(s)
Lung Neoplasms , Pancreatic Neoplasms , Humans , Pancreatic Neoplasms/pathology , Lung Neoplasms/pathology , Prognosis , Proportional Hazards Models , Lung/pathology , Retrospective StudiesABSTRACT
INTRODUCTION: Various calculation models to predict surgical risk have been developed globally. These have been reported to be helpful for estimating the long-term prognosis. In Japan, a similar model for lung cancer surgery was developed in 2017; however, there have been no reports investigating its association with the long-term prognosis. The objective of this study was to assess the association of the model's predictions with the long-term prognosis. PATIENTS AND METHODS: In this retrospective single-institutional study, we analyzed lung cancer patients who underwent radical lobectomy between 2010 and 2016. We calculated the predicted rates of mortality (PRM) and composite outcomes of mortality with major morbidity (PRMM) in eligible patients (N = 1054) using this model and classified them into 2 classes (class A, PRM ≥0.8% and PRMM ≥5.9%; class B, others) based on their models' predictions. We assessed the prognostic impact and clinical utility of the model's predictions. RESULTS: Class A included patients with significantly poorer postoperative overall survival than class B (log-rank, P < .001; hazard ratio, 3.160; 95% confidence interval, 2.390-4.178). Time-dependent receiver operating characteristic curve analyses revealed that the model's predictions correlated strongly with 1- and 2-year overall survival and decision curve analysis showed that they had high net benefits for prediction of those. CONCLUSION: The Japanese risk calculator could stratify the long-term prognosis for lung cancer patients after surgery. This model may be a valuable tool not only for multidisciplinary thoracic oncology teams to discuss treatment strategies for high-risk cases but also for them to share the decision-making process with patients.
Subject(s)
Lung Neoplasms , Humans , Lung Neoplasms/surgery , Retrospective Studies , Risk Assessment , East Asian People , PrognosisABSTRACT
Background: Bronchopleural fistula (BPF) after lobectomy for lung cancer is a rare but serious complication. This study aimed to stratify the risk factors of BPF. Methods: Patients who underwent lobectomy without bronchoplasty and preoperative treatment for lung cancer between 2005 and 2020 were retrospectively reviewed. We examined the association between the incidence of BPF and background factors, including comorbidities, preoperative blood test results, respiratory function, surgical procedure, and extent of lymphadenectomy. Results: Among the 3,180 patients who underwent lobectomy, 14 (0.44%) developed BPF. The median interval from surgery to BPF onset was 21 days (range, 10-287). Two of the 14 patients died of BPF (mortality rate, 14%). All 14 patients who developed BPF were men and had undergone right lower lobectomy. Other factors significantly associated with the development of BPF were older age, heavy smoking, obstructive ventilatory failure, interstitial pneumonia, history of malignancy, history of gastric cancer surgery, low serum albumin levels, and histology. Multivariable analysis in the subgroup of men who underwent right lower lobectomy revealed that high level of serum C-reactive protein and a history of gastric cancer surgery were significantly associated with BPF, whereas bronchial stump coverage was inversely associated with BPF. Conclusions: Men who underwent right lower lobectomy were at increased risk of BPF. The risk was higher when the patient had high serum C-reactive protein or a history of gastric cancer surgery. Bronchial stump coverage might be effective in patients at high risk of BPF.
ABSTRACT
Background: We hypothesized that epidermal growth factor receptor (EGFR) mutations could be detected in early-stage lung adenocarcinoma using radiomics. Methods: This retrospective study included consecutive patients with clinical stage I/II lung adenocarcinoma who underwent curative-intent pulmonary resection from March-December 2016. Using preoperative enhanced chest computed tomography, 3,951 radiomic features were extracted in total from the tumor (area within the tumor boundary), tumor rim (area within ±3 mm of the tumor boundary), and tumor exterior (area between +10 mm outside the tumor and tumor boundary). A machine learning-based radiomics model was constructed to detect EGFR mutations. The combined model incorporated both radiomic and clinical features (gender and smoking history). The performance was validated with five-fold cross-validation and evaluated using the mean area under the curve (AUC). Results: Of 99 patients (mean age, 66±11 years; female, 66.6%; clinical stage I/II, 89.9%/10.1%), EGFR mutations in the surgical specimen were detected in 46 (46.5%). A median of 4 (range, 2 to 8) radiomic features was selected for each validation session. The mean AUCs in the radiomics and combined models were 0.75 and 0.83, respectively. The two top-ranked features in the combined model were the radiomic features extracted from the tumor exterior and the tumor, indicating a higher impact of radiomic features over relevant clinical features. Conclusions: Radiomic features, including those in the peri-tumoral area, may help detect EGFR mutations in lung adenocarcinomas in preoperative settings. This non-invasive image-based technology could help guide future precision neoadjuvant therapy.
ABSTRACT
Objective: There is little evidence of the outcome of pulmonary metastasectomy for uterine tumors when comparing different histologies. This study aimed to delineate the primary histology that leads to more favorable outcomes after pulmonary metastasectomy. Methods: The database of the Metastatic Lung Tumor Study Group of Japan for 1984 to 2016 was used to analyze the outcomes of patients with gynecologic malignancies who underwent pulmonary metastasectomy. Prognostic factors and long-term outcomes were compared according to the histology of the primary uterine tumors, specifically adenocarcinoma, squamous cell carcinoma, and sarcoma. The adjusted hazard risks according to disease-free intervals (DFIs) and the number and maximum size of resected tumors were also analyzed to delineate the pattern of risk trends. Results: A total of 319 patients were included in the analysis (122 with adenocarcinomas, 113 with squamous cell carcinomas, 46 with sarcomas, and 38 with other types). The 5-year survival rate was 66.5% for the entire cohort, 71.6% for the patients with adenocarcinoma, 61.3% for those with squamous cell carcinoma, and 55.4% for those with sarcoma. Multivariate analyses identified the positive prognostic factors as DFI ≥12 months in adenocarcinoma and sarcoma and the primary site (corpus) of uterine tumors in adenocarcinoma. The nonlinear adjusted hazard risks indicated that a shorter DFI was associated with an elevated risk of death in patients with adenocarcinoma and sarcoma. Conclusions: The survival outcome after pulmonary metastasectomy varies according to primary tumor histology, and the prognostic factors differ among histologic subtypes. Surgical indications should be determined based on the prognostic factors for each histology.
ABSTRACT
To clarify the clinical impact and to identify prognostic predictors of surgical intervention for pulmonary metastasis from esophageal cancer, a registry database analysis was performed. From January 2000 to March 2020, patients who underwent resection of pulmonary metastases from primary esophageal cancer at 18 institutions were registered in a database developed by the Metastatic Lung Tumor Study Group of Japan. An amount of 109 cases were reviewed and examined for the prognostic factors for pulmonary metastasectomy of metastases from esophageal cancer. As a result, five-year overall survival after pulmonary metastasectomy was 34.4% and five-year disease-free survival was 22.1%. The multivariate analysis for overall survival revealed that initial recurrence site, maximum tumor size, and duration from primary tumor treatment to lung surgery were selected as the significant prognostic factors (p = 0.043, p = 0.048, and p = 0.037, respectively). In addition, from the results of the multivariate analysis for disease free survival, number of lung metastases, initial recurrence site, duration from primary tumor treatment to lung surgery, and preoperative chemotherapy for lung metastasis were selected as the significant prognostic factors (p = 0.037, p = 0.008, p = 0.010, and p = 0.020, respectively). In conclusion, eligible patients with pulmonary metastasis from esophageal cancer selected based on the identified prognostic predictors would be good candidates for pulmonary metastasectomy.
ABSTRACT
PURPOSE: Most robot-assisted thoracoscopic surgery (RATS) is performed from the vertical view. This study evaluates the initial outcomes of our novel confronting RATS technique, in which the patient was viewed horizontally, as in open thoracotomy. METHODS: We reviewed data on patients who underwent thoracoscopic lobectomy between January, 2019 and April, 2022. Perioperative outcomes were compared between RATS and video-assisted thoracoscopic surgery (VATS), using propensity-score matching. RESULTS: RATS and VATS were performed for 83 and 571 patients, respectively. After propensity-score matching, data on 81 patients from each of the two groups were retrieved. The operative time was significantly longer for RATS than for VATS (199 ± 44 min vs. 173 ± 37 min, p < 0.001). There was no mortality or conversion to thoracotomy in either of the groups. The rates of overall complications and prolonged air leak did not differ significantly between the groups. The serum creatine phosphokinase level on postoperative day 4 was higher after RATS than after VATS. The number of resected lymph nodes and the rates of nodal upstaging did not differ significantly between the groups. CONCLUSION: The initial perioperative outcomes of RATS using the confronting settings were comparable to those of VATS.
Subject(s)
Lung Neoplasms , Robotics , Humans , Lung Neoplasms/pathology , Lymph Node Excision , Pneumonectomy/methods , Retrospective Studies , Thoracic Surgery, Video-Assisted/methods , Thoracotomy/methodsABSTRACT
BACKGROUND. Differentiation of primary lung cancers and pulmonary metastases may present a diagnostic dilemma given overlapping CT findings. OBJECTIVE. The purpose of this study was to compare the utility of ringlike peripheral increased iodine concentration and conventional findings for differentiating primary lung cancers from pulmonary metastases on dual-energy CT (DECT). METHODS. This retrospective study included 93 patients (64 men, 29 women; median age, 70 years) who underwent resection of a primary lung cancer (n = 68) or pulmonary metastasis (n = 25) corresponding to a solid lesion on preoperative contrast-enhanced DECT performed between April 2020 and March 2021. Venous phase 120-keV single-energy images, equilibrium phase 66-keV virtual monoenergetic images, and iodine concentration maps were reconstructed. Two radiologists independently assessed lesions for spiculated margins, air bronchograms, rim enhancement, and thin ringlike peripheral high iodine concentration; differences were resolved by consensus. Inter-reader agreement and diagnostic performance were assessed. Multivariable logistic regression analysis incorporated additional patient and lesion characteristics. RESULTS. Interobserver agreement, expressed as kappa, was 0.26 for spiculated margins, 0.60 for air bronchograms, 0.56 for rim enhancement, and 0.80 for ringlike peripheral high iodine concentration. Pulmonary metastases, compared with primary lung cancers, exhibited significantly higher frequency of ringlike peripheral high iodine concentration (52% vs 19%; p = .004) but no significant difference in frequency of spiculated margins (49% vs 32%; p = .17), air bronchograms (36% vs 51%; p = .24), or rim enhancement (4% vs 4%; p > .99). Sensitivity and specificity for diagnosing pulmonary metastasis were 68% and 49% for absence of spiculated margins, 64% and 51% for absence of air bronchograms, 4% and 96% for presence of rim enhancement, and 52% and 81% for presence of ringlike peripheral high iodine concentration. In multivariable analysis including smoking history, lesion diameter, multiple resected lesions, and ringlike peripheral high iodine concentration, the only independent significant predictor of pulmonary metastasis was ringlike peripheral high iodine concentration (OR, 7.81 [95% CI, 2.28-29.60); p = .001). CONCLUSION. Ringlike peripheral high iodine concentration had excellent interobserver agreement and high specificity (albeit poor sensitivity) for differentiating pulmonary metastasis from primary lung cancer and was independently predictive of pulmonary metastasis. CLINICAL IMPACT. Ringlike peripheral high iodine concentration could help guide management of patients with known cancer and an indeterminate solitary nodule.
Subject(s)
Iodine , Lung Neoplasms , Radiography, Dual-Energy Scanned Projection , Male , Humans , Female , Aged , Tomography, X-Ray Computed/methods , Contrast Media , Retrospective Studies , Lung Neoplasms/diagnostic imaging , Radiography, Dual-Energy Scanned Projection/methodsABSTRACT
INTRODUCTION: Although solid appearance on computed tomography and positive findings on positron emission tomography (PET) have been both associated with poor outcome in lung adenocarcinoma, the extent to which these findings overlap is unknown. This study aimed to determine the differences in prognostic significance of PET findings in part-solid nodules (PSNs) and solid nodules. MATERIALS AND METHODS: We retrospectively investigated 417 patients with clinical stage IA adenocarcinoma who underwent curative resection between 2010 and 2017. We compared disease-free survival (DFS), cumulative incidence of disease recurrence (CIR) and clinicopathological characters between PET-positive and negative groups among PSNs and solid nodules, respectively. We used 2.5 as a cut-off value of maximum standardized uptake value (SUV max). RESULTS: In PSNs (n = 235), PET-positive group (n = 59) showed more aggressive features in several clinicopathological variables, poorer DFS (P < .001) and higher CIR (P < .001) than PET-negative group (n = 176). In contrast, in solid nodules (n = 182), DFS (P = .521) and CIR (P = .311) were not significantly different between PET-positive (n = 128) and negative groups (n = 54). SUV max was proved to be the independent prognostic factor of DFS by multivariate analysis (HR, 1.155; 95% CI, 1.036-1.287) only in PSNs. CONCLUSION: These findings showed distinct impact on prognosis of PET findings between PSNs and solid nodules. PET-positive finding was more important prognostic factor in PSNs than in solid nodules among clinical stage IA lung adenocarcinoma.