Development and validation of a supervised deep learning algorithm for automated whole-slide programmed death-ligand 1 tumour proportion score assessment in non-small cell lung cancer.

AIMS: Immunohistochemical programmed death-ligand 1 (PD-L1) staining to predict responsiveness to immunotherapy in patients with advanced non-small cell lung cancer (NSCLC) has several drawbacks: a robust gold standard is lacking, and there is substantial interobserver and intraobserver variance, with up to 20% discordance around cutoff points. The aim of this study was to develop a new deep learning-based PD-L1 tumour proportion score (TPS) algorithm, trained and validated on a routine diagnostic dataset of digitised PD-L1 (22C3, laboratory-developed test)-stained samples. METHODS AND RESULTS: We designed a fully supervised deep learning algorithm for whole-slide PD-L1 assessment, consisting of four sequential convolutional neural networks (CNNs), using aiforia create software. We included 199 whole slide images (WSIs) of 'routine diagnostic' histology samples from stage IV NSCLC patients, and trained the algorithm by using a training set of 60 representative cases. We validated the algorithm by comparing the algorithm TPS with the reference score in a held-out validation set. The algorithm had similar concordance with the reference score (79%) as the pathologists had with one another (75%). The intraclass coefficient was 0.96 and Cohen's κ coefficient was 0.69 for the algorithm. Around the 1% and 50% cutoff points, concordance was also similar between pathologists and the algorithm. CONCLUSIONS: We designed a new, deep learning-based PD-L1 TPS algorithm that is similarly able to assess PD-L1 expression in daily routine diagnostic cases as pathologists. Successful validation on routine diagnostic WSIs and detailed visual feedback show that this algorithm meets the requirements for functioning as a 'scoring assistant'.

Budget impact of sequential treatment with first-line afatinib versus first-line osimertinib in non-small-cell lung cancer patients with common EGFR mutations.

BACKGROUND: The therapeutic landscape for non-small-cell lung cancer (NSCLC) patients that have common epidermal growth factor receptor (EGFR) mutations has changed radically in the last decade. The availability of these treatment options has an economic impact, therefore a budget impact analysis was performed. METHODS: A budget impact analysis was conducted from a Dutch healthcare perspective over a 5-year time horizon in EGFR-mutant NSCLC patients receiving first-line afatinib (Gilotrif®) versus first-line osimertinib (Tagrisso®), followed by subsequent treatments. A decision analysis model was constructed in Excel. Scenario analyses and one-way sensitivity analysis were used to test the models' robustness. RESULTS: Sequential treatment with afatinib versus first-line treatment with osimertinib showed mean total time on treatment (ToT) of 29.1 months versus 24.7 months, quality-adjusted life months (QALMs) of 20.2 versus 17.4 with mean cost of €108,166 per patient versus €143,251 per patient, respectively. The 5-year total budget impact was €110.4 million for the afatinib sequence versus €158.6 million for the osimertinib sequence, leading to total incremental cost savings of €48.15 million. CONCLUSIONS: First-line afatinib treatment in patients with EGFR-mutant NSCLC had a lower financial impact on the Dutch healthcare budget with a higher mean ToT and QALM compared to osimertinib sequential treatment.

Non-invasive stroke volume assessment in patients with pulmonary arterial hypertension: left-sided data mandatory.

BACKGROUND: Cardiovascular magnetic resonance (CMR) is an emerging modality in the diagnosis and follow-up of patients with pulmonary arterial hypertension (PAH). Derivation of stroke volume (SV) from the pulmonary flow curves is considered as a standard in this respect. Our aim was to investigate the accuracy of pulmonary artery (PA) flow for measuring SV. METHODS: Thirty-four PAH patients underwent both CMR and right-sided heart catheterisation. CMR-derived SV was measured by PA flow, left (LV) and right ventricular (RV) volumes, and, in a subset of nine patients also by aortic flow. These SV values were compared to the SV obtained by invasive Fick method. RESULTS: For SV by PA flow versus Fick, r = 0.71, mean difference was -4.2 ml with limits of agreement 26.8 and -18.3 ml. For SV by LV volumes versus Fick, r = 0.95, mean difference was -0.8 ml with limits of agreement of 8.7 and -10.4 ml. For SV by RV volumes versus Fick, r = 0.73, mean difference -0.75 ml with limits of agreement 21.8 and -23.3 ml. In the subset of nine patients, SV by aorta flow versus Fick yielded r = 0.95, while in this subset SV by pulmonary flow versus Fick yielded r = 0.76. For all regression analyses, p < 0.0001. CONCLUSION: In conclusion, SV from PA flow has limited accuracy in PAH patients. LV volumes and aorta flow are to be preferred for the measurement of SV.

Traditional versus up-front [18F] fluorodeoxyglucose-positron emission tomography staging of non-small-cell lung cancer: a Dutch cooperative randomized study.

PURPOSE: We investigated whether application of positron emission tomography (PET) immediately after first presentation might simplify staging while maintaining accuracy, as compared with traditional strategy in routine clinical setting. METHODS: At first presentation, patients with a provisional diagnosis of lung cancer without overt dissemination were randomly assigned to traditional work-up (TWU) according to international guidelines or early PET followed by histologic/cytologic verification of lesions, or imaging and follow-up. Patients with [18F] fluorodeoxyglucose (18FDG) -avid, noncentral tumors without suspicion of mediastinal or distant metastases on PET proceeded directly to thoracotomy. Follow-up in presumed benign lesions was at least 12 months. In patients treated with surgery or neoadjuvant therapy, the quality of staging was measured by comparing the clinical stage to the final stage (combination of peroperative staging and 6 months of follow-up). To investigate test substitution, we analyzed the number of (non)invasive tests to achieve clinical TNM staging, and its associated costs. RESULTS: Between 1999 and 2001, 465 patients (233 TWU, 232 PET) were enrolled at 22 hospitals. The mean (standard deviation) number of procedures to finalize staging was equal in the TWU arm and the PET arm: 7.9 (2.0) v 7.9 (1.9), P = .90, respectively. Mediastinoscopies occurred significantly less often in the PET arm. Agreement between clinical and final stage was good in both arms (kappa = .85 v .78; P = .07). Costs did not differ significantly. CONCLUSION: Up-front 18FDG-PET in patients with (suspected) lung cancer does not reduce the overall number of diagnostic test, but it maintains quality of TNM staging with the use of less invasive surgery.

Assessment of the pulmonary volume pulse in idiopathic pulmonary arterial hypertension by means of electrical impedance tomography.

BACKGROUND: Electrical impedance tomography (EIT) is a non-invasive imaging technique which can be used to measure the blood volume changes in the pulmonary vascular bed during the cardiac cycle. STUDY OBJECTIVES: This study was performed to evaluate the differences in the EIT signal of the pulmonary vascular bed between healthy subjects and patients with idiopathic pulmonary arterial hypertension (IPAH), who are known to have a remodelled pulmonary vascular bed. PATIENTS AND METHODS: Twenty-one patients (17 females, 4 males) with IPAH and 30 healthy controls (5 females, 25 males) were measured. EIT measurements were performed in duplicate, on the same day as right heart catheterization to obtain haemodynamic data. The maximal impedance change during systole (Delta Z(sys)) was used as a measure of the pulmonary volume pulse and expressed in arbitrary units (AU). Total lung capacity, spirometric values and diffusion capacity for carbon monoxide were measured as well. RESULTS: Mean Delta Z(sys) was 215 +/- 58 x 10(-2) AU (95% CI 193 x 10(-2) to 236 x 10(-2)) in the healthy subjects and 78 +/- 27 x 10(-2) AU (95% CI 66 x 10(-2) to 91 x 10(-2)) in the IPAH patient group (p < 0.0001). No significant correlation was found between Delta Z(sys) and any of the haemodynamic or lung function data. CONCLUSION: The impedance pulsation of the pulmonary vascular bed is reduced in IPAH in comparison with controls, indicating a reduced volume pulse. This might represent the reduced cross section area, as well as the reduced compliance and number of the pulmonary vessels in these patients.

Cost-effectiveness of early intervention: comparison between intraluminal bronchoscopic treatment and surgical resection for T1N0 lung cancer patients.

BACKGROUND: For patients with early-stage lung cancer (ESLC) and severe comorbidities, the cost-effectiveness of early intervention may be reduced by screening and treatment-related morbidity and mortality in addition to the risk for non-cancer-related deaths. OBJECTIVES: The use of bronchoscopic treatment (BT) for centrally located ESLC as minimally invasive technique has raised questions whether this approach will be more cost-effective than standard surgical resection in the above-mentioned cohort of patients. METHODS: The cost-effectiveness of BT of 32 medically inoperable patients with intraluminal tumor has been compared to a matched control group of surgically treated stage IA cancer patients. RESULTS: Median follow-up after BT for ESLC has been 5 years (range 2-10) versus 6.7 years (range 2-10) for the surgical group. Five patients (16%) developed subsequent primaries/local recurrences after BT versus 4 (12.5%) in the surgical group. The respective percentages of actual survival during follow-up have been 50 and 41%, non-lung-cancer-related death 22 and 31% and lung-cancer-related death 28% in both groups, respectively. So far, the average costs per individual for early management by BT have been Euro 22,638 by surgery, and total expenses have been Euro 209,492 and Euro 724,403, respectively. CONCLUSIONS: Despite the worse initial health status of patients treated with BT, actual survival rates and costs for early intervention underscore the superior cost-effectiveness of BT as early intervention in properly selected individuals with ESLC in the central airways.

Cost-effectiveness of FDG-PET in staging non-small cell lung cancer: the PLUS study.

Currently, up to 50% of the operations in early-stage non-small cell lung cancer (NSCLC) are futile owing to the presence of locally advanced tumour or distant metastases. More accurate pre-operative staging is required in order to reduce the number of futile operations. The cost-effectiveness of fluorine-18 fluorodeoxyglucose positron emission tomography ((18)FDG-PET) added to the conventional diagnostic work-up was studied in the PLUS study. Prior to invasive staging and/or thoracotomy, 188 patients with (suspected) NSCLC were randomly assigned to conventional work-up (CWU) and whole-body PET or to CWU alone. CWU was based on prevailing guidelines. Pre-operative staging was followed by 1 year of follow-up. Outcomes are expressed in the percentage of correctly staged patients and the associated costs. The cost price of PET varied between 736 and 1,588 depending on the (hospital) setting and the procurement of (18)FDG commercially or from on-site production. In the CWU group, 41% of the patients underwent a futile thoracotomy, whereas in the PET group 21% of the thoracotomies were considered futile ( P=0.003). The average costs per patient in the CWU group were 9,573 and in the PET group, 8,284. The major cost driver was the number of hospital days related to recovery from surgery. Sensitivity analysis on the cost and accuracy of PET showed that the results were robust, i.e. in favour of the PET group. The addition of PET to CWU prevented futile surgery in one out of five patients with suspected NSCLC. Despite the additional PET costs, the total costs were lower in the PET group, mainly due to a reduction in the number of futile operations. The additional use of PET in the staging of patients with NSCLC is feasible, safe and cost saving from a clinical and from an economic perspective.

Effectiveness of positron emission tomography in the preoperative assessment of patients with suspected non-small-cell lung cancer: the PLUS multicentre randomised trial.

BACKGROUND: Up to 50% of curative surgery for suspected non-small-cell lung cancer is unsuccessful. Accuracy of positron emission tomography (PET) with 18-fluorodeoxyglucose (18FDG) is thought to be better than conventional staging for diagnosis of this malignancy. Up to now however, there has been no evidence that PET leads to improved management of patients in routine clinical practice. We did a randomised controlled trial in patients with suspected non-small-cell lung cancer, who were scheduled for surgery after conventional workup, to test whether PET with 18FDG reduces number of futile thoracotomies. METHODS: Before surgery (mediastinoscopy or thoracotomy), 188 patients from nine hospitals were randomly assigned to either conventional workup (CWU) or conventional workup and PET (CWU+PET). Patients were followed up for 1 year. Thoracotomy was regarded as futile if the patient had benign disease, explorative thoracotomy, pathological stage IIIA-N2/IIIB, or postoperative relapse or death within 12 months of randomisation. The primary outcome measure was futile thoracotomy. Analysis was by intention to treat. FINDINGS: 96 patients were randomly assigned CWU and 92 CWU+PET. Two patients in the CWU+PET group did not undergo PET. 18 patients in the CWU group and 32 in the CWU+PET group did not have thoracotomy. In the CWU group, 39 (41%) patients had a futile thoracotomy, compared with 19 (21%) in the CWU+PET group (relative reduction 51%, 95% CI 32-80%; p=0.003). INTERPRETATION: Addition of PET to conventional workup prevented unnecessary surgery in one out of five patients with suspected non-small-cell lung cancer.