Management of solitary pulmonary nodules (SPNs) of up to 3 cm was modelled on decision analysis comparing "wait and watch", transthoracic needle biopsy (TNB), exploratory surgery and full-ring dedicated positron emission tomography (PET) using fluorine-18 2-fluorodeoxyglucose (FDG). The incremental cost-effectiveness ratios (ICERs) were calculated for the main risk group, a cohort of 62-year-old men, using first "wait and watch" and second exploratory surgery as the baseline strategy. Based on published data, the sensitivity and specificity of FDG-PET were estimated at 0.95 and 0.80 for detecting malignancy in SPNs and at 0.74 and 0.96 for detecting metastasis in normal-sized mediastinal lymph nodes. The costs quoted correspond to reimbursement in 1999 by the public health provider in Germany. Decision analysis modelling indicates the potential cost-effectiveness of the FDG-PET strategy for management of SPNs. Taking watchful waiting as the low-cost baseline strategy, the ICER of PET [3218 euros (EUR) per life year saved] was more favourable than that of exploratory surgery (4210 EUR/year) or that of TNB (6120 EUR/year). Changing the baseline strategy to exploratory surgery, the use of PET led to cost savings and additional life expectancy. This constellation was described by a negative ICER of -6912 EUR/year. The PET algorithm was cost-effective for risk and non-risk patients. However, the ICER of PET as the preferred strategy was sensitive to a hypothetical deterioration of any PET parameters by more than 0.07. To transfer the diagnostic efficacy from controlled studies to the routine user and to maintain the cost-effectiveness of this technology, obligatory protocols for data acquisitions would need to be defined. If the prevalence of SPNs is estimated at the USA level (52 per 100,000 individuals) and assuming that multiple strategies without PET are the norm, the overall costs of a newly implemented PET algorithm would be limited to far less than one EUR per member of the public health provider in Germany.

Decision analysis is used here to establish the most cost-effective strategy for management of potentially operable non-small cell lung cancers (NSCLCs). The strategies compared were conventional staging (strategy A), dedicated systems of positron emission tomography (PET) using fluorine-18 fluorodeoxyglucose (FDG) in patients with normal-sized (strategy B) or in patients with enlarged mediastinal lymph nodes (part of strategy C), and FDG-PET followed by exclusion from surgical procedures when both computed tomography (CT) and PET were positive for mediastinal lymph nodes (strategy D) or when PET alone was positive (strategy E). Based on published data, the sensitivity and specificity of FDG-PET were estimated at 0.74 and 0.96 for detecting metastasis in normal-sized mediastinal lymph nodes, and at 0.95 and 0.76 when these lymph nodes were enlarged. The calculated probability of up-staging to M1 by using PET was 0.05. The costs quoted correspond to the cost reimbursed in 1999 by the public health provider in Germany. The incremental cost-effectiveness ratio (ICER) of strategy B was much more favourable (143 EUR/LYS; LYS = life year saved) than the ICER of strategy C (36,667 EUR/LYS). In strategy B, the use of PET did not raise the overall costs because the costs of PET were almost balanced by a better selection of patients for beneficial cancer resection. The exclusion from biopsy confirmation in strategies D and E led to cost savings that did not justify the expected reduction in life expectancy. In sensitivity analyses, the ICERs of strategy B were robust to the pretest likelihood of N2/N3, to penalized test parameters of PET and to reimbursement of PET. However, the ICER of strategy B would be raised to 28,000 EUR/LYS through use of thoracic PET without whole-body scanning. To conclude, the implementation of whole-body PET with a full ring of detectors in the preoperative staging of patients with NSCLC and normal-sized lymph nodes is clearly cost-effective. However, patients with nodal-positive PET results should not be excluded from biopsy.

Paragangliomas or glomus tumours of the head and neck region are rare somatostatin receptor-expressing neuroendocrine tumours. Precise preoperative diagnosis is of special importance in order to adequately weigh the potential benefit of the operation against the inherent risks of the procedure. In this study, the clinical value of somatostatin receptor imaging was assessed in 19 patients who underwent somatostatin receptor scintigraphy because of known or suspected paraganglioma of the head and neck region. The results were compared with the results of computed tomography and/or magnetic resonance imaging, histology and clinical follow-up. [(111)In-DTPA- D-Phe(1)]-octreotide scintigraphy was performed 4-6 and 24 h after i.v. injection of 140-220 MBq (111)In-octreotide. Whole-body and planar images as well as single-photon emission tomography images were acquired and lesions were graded according to qualitative tracer uptake. Somatostatin receptor imaging was positive in nine patients, identifying paragangliomas for the first time in three patients and recurrent disease in six patients. In one patient, a second, previously unknown paraganglioma site was identified. Negative results were obtained in ten patients. These patients included one suffering from chronic hyperplastic otitis externa, one with granuloma tissue and an organised haematoma, one with an acoustic neuroma, one with an asymmetric internal carotid artery, two with ectasia of the bulbus venae jugularis and one with a jugular vein thrombosis. In two patients with a strong family history of paraganglioma, individual involvement could be excluded. In only one patient did somatostatin receptor imaging and magnetic resonance imaging yield false negative results in respect of recurrent paraganglioma tissue. It is concluded that somatostatin receptor scintigraphy provides important information in patients with suspected paragangliomas of the head and neck region and has a strong impact on further therapeutic management.

Aim We aim to compare machine learning with neural network performance in predicting R0 resection (R0), length of stay > 14 days (LOS), major complication rates at 30 days postoperatively (COMP) and survival greater than 1 year (SURV) for patients having pelvic exenteration for locally advanced and recurrent rectal cancer. Method A deep learning computer was built and the programming environment was established. The PelvEx Collaborative database was used which contains anonymized data on patients who underwent pelvic exenteration for locally advanced or locally recurrent colorectal cancer between 2004 and 2014. Logistic regression, a support vector machine and an artificial neural network (ANN) were trained. Twenty per cent of the data were used as a test set for calculating prediction accuracy for R0, LOS, COMP and SURV. Model performance was measured by plotting receiver operating characteristic (ROC) curves and calculating the area under the ROC curve (AUROC). Results Machine learning models and ANNs were trained on 1147 cases. The AUROC for all outcome predictions ranged from 0.608 to 0.793 indicating modest to moderate predictive ability. The models performed best at predicting LOS > 14 days with an AUROC of 0.793 using preoperative and operative data. Visualized logistic regression model weights indicate a varying impact of variables on the outcome in question. Conclusion This paper highlights the potential for predictive modelling of large international databases. Current data allow moderate predictive ability of both complex ANNs and more classic methods.