Search

Objective: To assess MRI derived radiomics features of liver fat collected from a cohort of individuals without prior cardiovascular events as imaging biomarkers of type 2 diabetes mellitus (T2DM) and metabolic syndrome (MetS). Material and methods: 400 participants of the KORA MRI substudy underwent comprehensive whole body MRI imaging protocols including T1-weighted dual-echo Dixon (T1-DED), T1-weighted multi-echo Dixon (T1-MED) and magnetic resonance spectroscopy (MRS) sequences. A total of 684 radiomics features were extracted on T1-DED relative fat water content (rfwc) maps of 310 artefact free manually contoured liver volumes of interest (VOI). The corresponding individuals (n = 310, T2DM 12.6 %, MetS 34.5 %) were assigned to stratified training (n = 232, 75 %) and validation (n = 78, 25 %) sets. To assess feature stability, test-retest and inter-rater variance was approximated by generating noise augmented rfwc maps and computationally deformed VOIs, respectively. Feature stability was assessed in terms of the intraclass correlation coefficient (ICC) for test-retest reliability (ICC(1,1)) and inter-rater agreement (ICC(3,k)) on training set data. Stable features (ICC ≥ 0.85) were assessed as imaging biomarkers of T2DM and MetS in random forest (RF) models. For benchmarking, RF models were trained on the participants’ hepatic proton density fat fraction (PDFF) quantified previously on the T1-MED and MRS images as well as the body mass index (BMI). All RF models were evaluated on the validation set using the area under the curve of the receiver operating characteristic (AUROC) and the balanced accuracy (AccuracyB) as performance metrics. Results: The epidemiological characteristics in training and validation sets were not statistically significantly different (p < 0.001). Furthermore, training and validation sets showed strong associations for both test-retest reliability (β 1.027 [1.015 – 1.040]) and inter-rater agreement (β 1.072 [1.040 – 1.103]). A total of 171 features (25.0 %) met the stability threshold (ICC ≥ 0.85). In subsequent RF modelling radiomics features predicted T2DM with AUROC 0.835 and AccuracyB 0.822 and MetS with AUROC 0.838 and AccuracyB 0.787, thereby outperforming all benchmark RF models in both metric categories. Conclusion: In this single-center study radiomics features of MRI-derived hepatic fat were superior biomarkers of T2DM and MetS than hepatic PDFF and the BMI. In the future, hepatic radiomics features deserve further exploration and development as potential biomarkers in metabolic disease.

European Union law overlaps and interacts with both the OECD’s Base Erosion and Profit Shifting project (BEPS) and its implementation and the member states’ tax treaties between them and with third countries, and there is also an area where all three fields meet. This intersection of EU law, BEPS and member states’ (mostly) bilateral tax treaties is the subject of this report. First, it should be noted that the Union’s competence under article 115 TFEU not only covers purely internal situations, but the Union can also use its internal competence to specify the treatment of non-EU investors or third-country investments, and it has done so, e.g., in the Anti-Tax Avoidance Directive (ATAD). This has potential impact also on tax treaties between the member states and with third countries: Given the supremacy of EU (secondary) law, domestic law implementing Directives (e.g., the ATAD) might, under certain conditions, arguably take precedence over (pre- and post-accession) tax treaties between the member states, even if that implementation is detrimental to taxpayers and irrespective of whether the specific tax treaty was concluded before or after a provision of a Directive entered into force. As for tax treaties with third countries the TFEU contains a differentiating rule, as article 351 TFEU (ex-article 307 EC) grandfathers (only) member states’ treaties with third countries, including tax treaties, that a member state concluded before 1 January 1958 or, for acceding states, before the date of their accession, so that EU law arguably takes precedence over post-accession tax treaties with third countries and, therefore, may directly affect the relevant member state’s (but of course not the third country’s) tax system. Second, the European Commission has issued various Recommendations with regard to post-BEPS tax treaties of the member states. A 2012 Recommendation “on aggressive tax planning” addressed (also) tax treaty-based double non-taxation and encouraged member states to include an appropriate subject-to-tax clause in their double taxation conventions. The Commission’s 2016 Recommendation dealt with the inclusion of a subject-to-tax clause in tax treaties, the definition of “permanent establishments” to prevent their artificial avoidance (article 5 OECD MC) and the use of an EU-compatible Principal Purposes Test (PPT), which refers to “a genuine economic activity” as a carve-out to align the clause with the case-law of the Court of Justice of the European Union as regards the abuse of law. Third, the OECD BEPS project has established a (political) minimum standard regarding measures against treaty shopping (article 7 MLI and article 29 OECD MC), and the Limitation on Benefits (LoB) clause in particular raises issues with regard to its compatibility with the EU fundamental freedoms. In particular, LoB clauses are confronted with continuing doubts regarding their compatibility with the freedom of establishment. These concerns have also found expression not only in various documents of the European Commission but also in the BEPS Action 6 Final Report, where the OECD noted that some countries may have “concerns based on EU law that prevent them from adopting the exact wording of the model provisions that are recommended in this report”, further specifying those concerns by recognizing “that the LOB rule will need to be adapted to reflect certain constraints or policy choices concerning other aspects of a bilateral tax treaty between two Contracting States” such as “concerns based on EU law”. Indeed, the “ownership clauses” in LoB provisions face scrutiny because the company’s residence state has agreed to give better conditions to companies held by shareholders resident in its own territory as compared to the ones resident elsewhere in the EU and the EEA. In such circumstances and in light of the Open Skies judgments, LoB clauses could thus be regarded as the immediate source of the discriminatory treatment. It is, however, unclear whether other – objective or subjective – tests in a typical LoB clause make them “EU compatible”, and if the source state’s perspective might require a different analysis in light of the ECJ’s decision in ACT Group Litigation. Fourth, and while the OECD BEPS project has not established a minimum standard with regard to mandatory binding arbitration, the 2017 Tax Dispute Resolution Directive (TDRD) has established a mechanism for binding arbitration with regard to tax “disputes”. While the TDRD does not address double taxation outside of a tax treaty context, it is a huge step towards the removal of double taxation caused by diverging interpretation and application of tax treaties between member states. Fifth, the OECD BEPS project has addressed situations of treaty-based non-taxation, which might also raise state aid questions under article 107 TFEU in cases where the misapplication of a tax treaty leads to “white income”. While generally “the need to avoid double taxation” would be a basis for a possible justification, it might indeed be asked if a double taxation convention must be interpreted, in light of article 107 TFEU, to not give rise to “white income” (e.g., through an unconditional exemption of untaxed income) or to “overcompensation” (e.g., through a tax sparing credit). That rather extreme path, however, was not (yet) taken by the Commission in the McDonald’s case: Indeed, to show selectivity, the Commission attempted merely to prove that Luxembourg had misapplied the applicable tax treaty. It did not rely on the alternative argument that double non-taxation resulting from the application of a tax treaty ipso facto amounts to state aid.

International audience

International audience

Le risque pris par l’agent économique n’est pas en tant que tel l’objet de réglementation fiscale. Il n’en reste pas moins qu’il est pris en considération par le droit fiscal, car celui-ci tend souvent inciter à la prise de risque par les investisseurs. Les fondements et la cohérence de la politique fiscale à cet égard méritent toutefois d’être interrogés.

International audience

International audience; Observations sous Tribunal administratif de Paris, 12 juillet 2017, n° 1505113, Société Google Ireland Limited - Qualification du jugement : Important