Segmentation of PET/CT lung cancer lesion images via a semi-supervised improved SwinUNet model

Background Using imaging techniques to diagnose malignant and inflammatory lesions in the lung can be challenging. Purpose To compare intravoxel incoherent motion (IVIM) and apparent diffusion coefficient (ADC) magnetic resonance imaging (MRI) analysis in their ability to discriminate lung cancer from focal inflammatory lung lesions. Material and Methods Thirty-eight patients with lung masses were included: 30 lung cancers and eight inflammatory lesions. Patients were imaged with 3.0T MRI diffusion weighted imaging (DWI) using 10 b values (range, 0-1000 s/mm2). Tissue diffusivity ( D), pseudo-diffusion coefficient ( D*), and perfusion fraction ( f) were calculated using segmented biexponential analysis. ADC (total) was calculated with monoexponential fitting of the DWI data. D, D*, f, and ADC were compared between lung cancer and inflammatory lung lesions. Receiver operating characteristic analysis was performed for all DWI parameters. Results The ADC was significantly higher for inflammatory lesions than for lung cancer ([1.21 ± 0.20] × 10-3 mm2/s vs. [0.97 ± 0.15] × 10-3 mm2/s; P = 0.004). By IVIM, f was found to be significantly higher in inflammatory lesions than lung cancer ([46.10 ± 12.92] % vs. [29.29 ± 10.89] %; P = 0.005). There was no difference in D and D* between lung cancer and inflammatory lesions ( P = 0.747 and 0.124, respectively). f showed comparable diagnostic performance with ADC in differentiating lung cancer from inflammatory lung lesions, with areas under the curve of 0.833 and 0.826, sensitivity 80.0% and 73.3%, and specificity 75.0% and 87.5%, respectively. Conclusion The IVIM parameter f value provides comparable diagnostic performance with ADC and could be used as a surrogate marker for differentiating lung cancer from inflammatory lesions.

From 1953 to 1985, a total of 1289 patients with primary carcinoma of the lung underwent surgical treatment. Of these 116 (8.9%) had small-sized (less than or equal to 2 cm in diameter) peripheral type lung cancer lesions. This study had three purposes: 1) to analyse how small-sized lung cancer lesions were detected; 2) to evaluate the reliability of diagnosis of small-sized cancer lesions; and 3) to evaluate pre- and post-prognostic factors of such patients compared with patients with peripheral type lung cancer lesions 2.1-3 cm in diameter. Of the 115 patients with small-sized lung cancer lesions were detected in the course of mass surveys. Cytopathological diagnosis in 75% of the patients resulted from transbronchial brushing cytology. The 5-year survival rate of patients who underwent resection of small-sized peripheral type lung cancer lesions was 70% (2.1-3 cm; 52%). Various factors such as histologic type, nodal involvement, pleural involvement, pathological stage, and success of the operation were shown to significantly affect survival. A comparison of two groups, i.e., those with lesions smaller than 2 cm in diameter and those with lesions 2.1-3 cm in diameter, showed the rate of lymph node metastasis to be significantly different. Of the patients with peripheral lung cancer lesions smaller than 2 cm who underwent surgery, 21% had peribronchial, hilar, or mediastinal lymph node metastasis. On the other hand, lymph node metastasis was seen in 43% of cases with peripheral lung cancer lesions 2.1-3 cm in diameter who underwent surgery.

Purpose: Histopathological differentiation of primary lung cancer is clinically important. We aimed to investigate whether diffusion tensor imaging (DTI) parameters of metastatic brain lesions could predict the histopathological types of the primary lung cancer.Methods: In total, 53 patients with 98 solid metastatic brain lesions of lung cancer were included. Lung tumors were subgrouped as non-small cell carcinoma (NSCLC) (n = 34) and small cell carcinoma (SCLC) (n = 19). Apparent diffusion coefficient (ADC) and Fractional anisotropy (FA) values were calculated from solid enhanced part of the brain metastases. The association between FA and ADC values and histopathological subtype of the primary tumor was investigated.Results: The mean ADC and FA values obtained from the solid part of the brain metastases of SCLC were significantly lower than the NSCLC metastases (P < 0.001 and P = 0.003, respectively). ROC curve analysis showed diagnostic performance for mean ADC values (AUC=0.889, P = < 0.001) and FA values (AUC = 0.677, P = 0.002). Cut-off value of > 0.909 × 10-3 mm2/s for mean ADC (Sensitivity = 80.3, Specificity = 83.8, PPV = 89.1, NPV = 72.1) and > 0.139 for FA values (Sensitivity = 80.3, Specificity = 54.1, PPV = 74.2, NPV= 62.5) revealed in differentiating NSCLC from NSCLC.Conclusion: DTI parameters of brain metastasis can discriminate SCLC and NSCLC. ADC and FA values of metastatic brain lesions due to the lung cancer may be an important tool to differentiate histopathological subgroups. DTI may guide clinicians for the management of intracranial metastatic lesions of lung cancer.

APOBEC3 enzymes are cytosine deaminases implicated in cancer. Precisely when APOBEC3 expression is induced during cancer development remains to be defined. Here we show that specific APOBEC3 genes are upregulated in breast ductal carcinoma in situ, and in preinvasive lung cancer lesions coincident with cellular proliferation. We observe evidence of APOBEC3-mediated subclonal mutagenesis propagated from TRACERx preinvasive to invasive non-small cell lung cancer (NSCLC) lesions. We find that APOBEC3B exacerbates DNA replication stress and chromosomal instability through incomplete replication of genomic DNA, manifested by accumulation of mitotic ultrafine bridges and 53BP1 nuclear bodies in the G1 phase of the cell cycle. Analysis of TRACERx NSCLC clinical samples and mouse lung cancer models revealed APOBEC3B expression driving replication stress and chromosome missegregation. We propose that APOBEC3 is functionally implicated in the onset of chromosomal instability and somatic mutational heterogeneity in preinvasive disease, providing fuel for selection early in cancer evolution. SIGNIFICANCE: This study reveals the dynamics and drivers of APOBEC3 gene expression in preinvasive disease and the exacerbation of cellular diversity by APOBEC3B through DNA replication stress to promote chromosomal instability early in cancer evolution.This article is highlighted in the In This Issue feature, p. 2355.

Sensitive detection of small metastatic lesions, which is highly dependent on lesion visualization, is crucial for staging lung cancer. We investigated the potential benefit of HYPER Iterative for improving the visualization of small metastatic lesions of lung cancer on early 18F-labeled fibroblast activation protein inhibitor (FAPI) PET/CT. Methods: A total of 19 patients with lung cancer underwent a 60-min [18F]F-FAPI-42 dynamic total-body uEXPLORER PET/CT scan. PET images with a 5-min acquisition time were extracted at 10, 30 min, and 60 min after tracer injection. Ordered-subset expectation maximization (OSEM) and HYPER Iterative were used for image reconstruction. SUVmax, tumor-to-liver ratio, tumor-to-blood ratio, and tumor-to-adjacent-nontumor ratio were calculated and compared between the 2 reconstruction methods and at 10, 30, and 60 min after injection. Results: All HYPER and OSEM PET images were of high quality, with HYPER PET images showing superior clarity. Small positive lesions (maximum diameter, ≤1 cm) were depicted clearer on HYPER PET than on OSEM PET images at all time points, particularly at 10 min after injection, where 16.4% of lesions were poorly visualized on OSEM PET but clearly depicted on HYPER PET images. The tumor-to-liver ratio, tumor-to-blood ratio, and tumor-to-nontumor ratio at 10, 30, and 60 min after injection scan on HYPER PET images were significantly higher than those on OSEM images at corresponding time points (P ≥ 0.05 for all comparisons). SUVmax was more than 2-fold greater in large positive lesions (maximum diameter, >1.0 cm) than in small positive lesions (maximum diameter, ≤1 cm) on both OSEM and HYPER PET images at 10, 30, and 60 min after injection (P < 0.05 for all comparisons). The visualization of large positive lesions was not significantly affected by reconstruction methods or scan times. Conclusion: HYPER Iterative reconstruction enhanced the visualization of small metastatic lesions in lung cancer when compared with conventional OSEM, enabling effective early imaging using [18F]F-FAPI-42 uEXPLORER total-body PET/CT at 10 min after tracer injection.

A randomized Phase III trial commenced in Japan in February 2016. Currently, 42 Gy in four fractions of stereotactic body radiotherapy prescribed at the D95% of the planning target volume, which is considered equal to the commonly used 48 Gy in four fractions at the isocenter using an old dose calculation algorithm, is the standard treatment in Japan for medically inoperable Stage IA non-small cell lung cancer and small lung lesions clinically diagnosed as primary lung cancer. This study aims to examine the superiority of 55 Gy in four fractions over 42 Gy in four fractions. A total of 750 patients are expected to be accrued in 5 years. The primary endpoint is overall survival and the secondary endpoints are progression-free survival, local progression-free survival, patterns of failure, local control period, adverse events and serious adverse events. This trial is registered at the UMIN Clinical Trials Registry as UMIN000021029 (http://www.umin.ac.jp/ctr/index.htm) the dose covering 95% of the volume (D95%) of the planning target volume.

The reported incidence of skin metastases from lung cancer varies from 2.8-7.5%. In the present study 8.7% of patients had skin metastases, with head and neck being the most common sites, nodular lesions the most common, and adenocarcinoma the most frequent histology. Although most patients develop these lesions during the course of a known progressive disease, they may be the presenting manifestation of a primary tumour in the lung. The occurrence of skin lesions in lung cancer announces an ominous prognosis. The response to chemotherapy is poor, possibly due to poor blood supply to the skin; monitoring response to chemotherapy, however, is easier when such lesions are present.

Rationale:Lung cancer is 1 of the most prevalent cancers globally. Definitive stereotactic ablative radiotherapy (SABR) is suggested for those who are unfit for or refuse surgical intervention. Here we present a patient with 2 lung cancer lesions who received SABR simultaneously with magnetic resonance Linear accelerator (Linac)-magnetic resonance (MR).Patient concerns:A 46-years-old man had history of left lower lung cancer post lobectomy in 2018. Two recurrent tumors were found 2 years following, then became enlarged 4 months later.Diagnoses:The recurrent tumors were found by computed tomography.Interventions:SABR was indicated due to inoperability and small size. Simulation was done both by computed tomography and MR scan with ViewRay MRIdian Linac, with the prescription dose being 50 gray in 4 fractions performed every other day within 2 weeks. The 2 lesions were irradiated at the same time with a single isocenter with mean treatment time was 78 minutes.Outcomes:No acute side effect was noted. Follow-up chest computed tomography scan 14 months after SABR showed mild consolidation and pneumonitis over the upper irradiated site favoring radiation-related reasons, while pneumonitis was resolved over the lower irradiated site. Positron emission tomography showed no definite evidence of FDG-avid recurrence. The patient has survived over 18 months following SABR and more than 4 years from the first diagnosis of lung cancer without significant adverse effects.Lessons:Simultaneous SABR for multiple lung lesions is quite challenging because tumor motion by breathing can increase the risk of missing the target. With help by MR-Linac, simultaneous SABR to multiple lung lesions can be performed safely with efficacy.

The aim of this study is to clarify the difference of F-18 FDG uptake kinetics between FDG-avid non-small-cell lung cancer (NSCLC) and benign lesions associated with various etiologies on dual-time point PET/CT scan, and to determine the optimal parameter for differentiation. The materials were 76 FDG-avid solitary NSCLC in 76 patients and 57 FDG-avid solitary benign lesions associated with various etiologies in 61 patients. FDG PET/CT scan was performed at 60 and 120 min after intravenous injection of 4.4 MBq/kg F-18 FDG. The maximum standardized uptake value (SUVmax) on early and delayed scans and the percent change of SUVmax (%DeltaSUVmax) between the two time points were measured. The optimal differential parameter was determined by receiver-operating characteristic curve analysis and evaluation of diagnostic accuracy. The mean +/- SD of early SUV max, delayed SUVmax and %DeltaSUVmax were 8.3 +/- 5.2, and 10.2 +/- 6.5, and 21.9% +/- 18.9 in FDG-avid NSCLC, and 3.8 +/- 3.2, 4.0 +/- 3.7, and 11.3% +/- 26.0 in FDG-avid benign lesions, respectively. Delayed SUVmax in NSCLC was significantly higher than early SUVmax (P < 0.0001); while not different in benign lesions. Percent change of SUVmax in NSCLC was also significantly higher than that in benign lesions (P < 0.01). The optimal parameter for the differentiation was delayed SUVmax > 5.5 and yielded sensitivity of 77.6%, specificity of 80.7% and accuracy of 78.9%, which provided better differentiation than the use of %DeltaSUVmax or the traditional parameter of early SUVmax > 2.5. However, 11 (19.2%) benign lesions were indistinguishable from NSCLC. Although delayed PET/CT scan enhances the difference of FDG uptake between FDG-avid NSCLC and benign lesions, and the use of delayed SUVmax > 5.5 appears to improve the differentiation of these hypermetabolic lesions compared with an early scan, careful interpretation and management for correct differentiation are still required.

Background: Previous familial linkage studies have identified a tumor suppressor locus at 6q23-25 in lung cancer (LC). However, no single gene has yet been implicated within this 30 Mb region. The identification of two-hit targets — genes disrupted by two mechanisms at the DNA level, such as copy number loss and hypermethylation — involved in LC risk and development will lead to better methods for risk assessment, early detection and treatment. Methods: EYA4 was identified as a two-hit target using multidimensional genome-wide analysis combining gene dosage, DNA methylation, and gene expression data in non-small cell lung cancer (NSCLC) and early pre-malignant lesions with matched adjacent non-malignant tissue. The association of allelic variants with LC risk was assessed in data collected by the Genetic Epidemiology of Lung Cancer Consortium (GELCC), and the association of gene expression with prognosis was investigated in public data using statistical analyses. Bronchial epithelial cells of LC patients and high risk individuals were examined for early hypermethylation of EYA4. The role of EYA4 in apoptosis, and the association of EYA4 with the DNA repair genes GADD45a and γ-H2AX was assayed using stable EYA4 mRNA knock-downs in lymphoblast cell lines. Cells were either serum-starved for 12 and 24 hours followed by Annexin5/propidium iodide staining and GADD45a qRT-PCR, or irradiated, and levels of γ-H2AX protein were measured over time. Results: Integration of NSCLC gene dosage, DNA methylation and mRNA expression showed EYA4 to be frequently affected by two-hits and significantly down-regulated. Numerous EYA4 variants were associated with familial LC, and low EYA4 expression was significantly associated with poor prognosis in multiple independent data sets. Congruent with EYA family member function, in vitro assays revealed that EYA4 knock-down cells displayed a decrease in the number of apoptotic cells following serum starvation. Further investigations led to the discovery of EYA4 hypermethylation not only in the bronchial epithelial brushings of NSCLC patients, but also in the brushings of individuals at a high risk of LC development. EYA4 knock-down led to attenuation of the GADD45a DNA damage repair response, as well as increased levels of γ-H2AX following DNA damage repair. Conclusions: EYA4 is a gene disrupted early and frequently that maps to a locus previously associated with cancer risk. It is implicated in somatic as well as familial cancers, and is likely a tumor suppressor gene with apoptotic and DNA repair functions. The direct association of EYA4 with risk and survival as well as its early disruption underscores its relevance on a clinical level. Citation Information: Cancer Prev Res 2010;3(12 Suppl):A29.

2543 Background: Although docetaxel is an effective treatment of lung cancer, a number of patients do not benefit from this therapy due to tumor resistance. Positron emission tomography (PET) is a noninvasive imaging technique that allows for quantification of radiolabeled docetaxel ([11C]docetaxel) kinetics and might be useful for predicting response to treatment. The aim of the present study was to determine the feasibility and reproducibility of [11C]docetaxel PET scans in lung cancer and to investigate whether [11C]docetaxel uptake was related to tumor perfusion. Methods: Fourteen patients with advanced lung cancer underwent a dynamic PET-CT scan with [11C]docetaxel (60 min) and H215O (10 min). In addition, patients underwent a second [11C]docetaxel PET scan to assess test-retest reproducibility. Lesions were delineated on the CT scan and projected onto the dynamic PET frames. [11C]docetaxel uptake in tumors was quantified using the net influx rate (Ki). Tumor perfusion was quantified by applying the standard single tissue compartment model to the H215O data. Results: In total, 24 lesions were defined, including both primary tumors and metastases. Clearance of [11C]docetaxel from plasma was rapid and later PET frames suffered from high liver uptake. Therefore, only the first 10 min of data were used for further analysis. The median net influx rate of [11C]docetaxel was 0.009 min-1 (range, 0.000-0.022). [11C]docetaxel PET scans showed good reproducibility (intraclass correlation coefficient, 0.89). The inter- and intra-individual variability of [11C]docetaxel uptake in lesions was high. [11C]docetaxel uptake was correlated with tumor perfusion (r=0.853, p<0.001). Conclusions: Measurement of [11C]docetaxel uptake in lung cancer is feasible with good reproducibility. [11C]docetaxel uptake depends on tumor perfusion. The variability of [11C]docetaxel uptake in lung cancer lesions may reflect differential sensitivity to docetaxel treatment, suggesting that PET scans with [11C]docetaxel may be useful for personalized treatment planning. Author Disclosure Employment or Leadership Position Consultant or Advisory Role Stock Ownership Honoraria Research Funding Expert Testimony Other Remuneration Philips Healthcare

Programmed Death-Ligand 1 Expression and EGFR Mutations in Multifocal Lung Cancer

Whole-lesion histogram analysis can provide comprehensive assessment of tissues by calculating additional quantitative metrics such as skewness and kurtosis; however, few studies have evaluated its value in the differential diagnosis of lung lesions. To compare the diagnostic performance of conventional diffusion-weighted imaging (DWI), intravoxel incoherent motion (IVIM) magnetic resonance imaging (MRI) and diffusion kurtosis imaging (DKI) in differentiating lung cancer from focal inflammatory lesions, based on whole-lesion volume histogram analysis. Fifty-nine patients with solitary pulmonary lesions underwent multiple b-values DWIs, which were then postprocessed using mono-exponential, bi-exponential and DKI models. Histogram parameters of the apparent diffusion coefficient (ADC), true diffusivity (D), pseudo-diffusion coefficient (D*), and perfusion fraction (f), apparent diffusional kurtosis (Kapp) and kurtosis-corrected diffusion coefficient (Dapp) were calculated and compared between the lung cancer and inflammatory lesion groups. Receiver operating characteristic (ROC) curves were constructed to evaluate the diagnostic performance. The ADCmean, ADCmedian, D mean and D median values of lung cancer were significantly lower than those of inflammatory lesions, while the ADCskewness, Kapp mean, Kapp median, Kapp SD, Kapp kurtosis and Dapp skewness values of lung cancer were significantly higher than those of inflammatory lesions (all p < 0.05). ADCskewness (p = 0.019) and D median (p = 0.031) were identified as independent predictors of lung cancer. D median showed the best performance for differentiating lung cancer from inflammatory lesions, with an area under the ROC curve of 0.777. Using a D median of 1.091 × 10-3 mm2/s as the optimal cut-off value, the sensitivity, specificity, positive predictive value and negative predictive value were 69.23%, 85.00%, 90.00% and 58.62%, respectively. Whole-lesion histogram analysis of DWI, IVIM and DKI parameters is a promising approach for differentiating lung cancer from inflammatory lesions, and D median shows the best performance in the differential diagnosis of solitary pulmonary lesions.

Purpose: To evaluate the CT findings of small bowel metastases from primary lung cancer. Materials and Methods: Of the 1468 patients with primary lung cancer between 1990 and 2000, 13 patients who had metastasis to the small intestine were collected. Of these 13 patients, nine who underwent CT scan were included for analysis. The pathologic diagnoses of primary lung cancer in these nine patients were squamous cell carcinoma in six, adenocarcinoma in two, and large cell carcinoma in one. CT scans were analyzed with regard to the site and patterns (intraluminal mass/bowel wall thickening/bowel implants) of metastatic masses, and the presence or absence of complication such as intussusception, obstruction, or perforation of the small bowel. The medical records of the patients were also reviewed retrospectively for evaluation of presenting abdominal symptom and time interval of metastases from initial diagnosis of lung cancer. Results: Metastatic lesions were distributed throughout the small intestine: the duodenum in five, the jejunum in four, the ileum in six, and both jejunum and ileum in one patient. The size of metastatic masses of small bowel ranged from 1.3 cm to 5.0 cm (mean size, 2.6 cm) On CT, the small bowel was involved with intraluminal masses (mean size, 3.4 cm) in eight patients, diffuse wall thickening (mean thickness, 1.6 cm) in five, and bowel implants (mean size, 2.2 cm) in two. Complications occurred in seven patients, including intussusceptions without obstruction in two patients and with obstruction in two, obstruction without intussusceptions in two, and bowel perforation in one. Of 9 patients, 6 had at least one symptom referable to the small bowel including abdominal pain in 4, anemia in 3, vomiting in 1, and jaundice in 1. Lung cancer and small bowel lesions were detected simultaneously in four patients and the time interval of metastases from initial diagnosis of lung cancer ranged from 10 days to 30 months (median interval, 54 days) in 5 patients. Conclusion: CT helps in defining the extent and pattern of small bowel metastases as well as in demonstrating their complication.

Intractable pleural effusion associated with superior vena cava and upper extremity deep vein thrombosis in a patient with advanced lung cancer