Anatomical subgroup analysis of the MERIDIAN cohort :

Objective To assess the contribution of in utero magnetic resonance (iuMR) imaging in fetuses diagnosed with ventriculomegaly as the only abnormal intracranial finding on antenatal ultrasonography (USS) in the MERIDIAN cohort. Methods We report a sub-group analysis from the MERIDIAN study, of cases of fetal ventriculomegaly diagnosed on USS who then had iuMR imaging within 2 weeks of USS and for whom outcome reference data were available. The diagnostic accuracy of USS and iuMR are reported in relation to the severity of ventriculomegaly. We also study the This article is protected by copyright. All rights reserved. A cc ep te d A rti cl e difference in measurements of trigone size on the two imaging methods and present the clinical impact of adding iuMR to the diagnostic pathway. Results USS failed to detect 31 additional brain abnormalities other than ‘ventriculomegaly’ in the subgroup of 306 fetuses, giving an overall diagnostic accuracy of 89.9% for USS, whilst iuMR correctly detected 27 giving a diagnostic accuracy of 98.4% (statistically significant p<0.0001). There were other brain abnormalities in 14/244 of fetuses with mild ventriculomegaly on USS (diagnostic accuracy 94.3%), and iuMR correctly diagnosed 12 of those (diagnostic accuracy 98.8%). All of those results reached statistical significance in favour of iuMR. There was a close agreement between the size of trigones measured on USS and on iuMR, with categorical differences in only 16% of cases. iuMR did not systematically overestimate trigone size, as suspected before the study commenced. Complete prognosis data were available in 295/306 fetuses and the prognosis category changed after iuMR in 69/295 (23.4%) cases. The overall effect of iuMR on clinical management was considered to be either ‘significant’, ‘major’ or ‘decisive’ in 76/295 (25.8%) cases. Conclusion Our data suggest that any woman whose fetus has ventriculomegaly as the only intracranial finding on USS should be offered an adjuvant investigation of iuMR for further evaluation. This article is protected by copyright. All rights reserved. A cc ep te d A rti cl e INTRODUCTION Women in the UK are offered ‘mid-trimester anomaly scanning’ using ultrasonography (USS) at approximately 20 weeks gestational age with the aims set out by the national fetal anomaly screening program. These include the identification of serious abnormalities incompatible with life or associated with morbidity. Some abnormalities, however, are recognised for the first time later in pregnancy when further USS might be performed for reasons such as monitoring fetal growth or investigating reduced fetal movements. Fetal brain abnormalities are amongst the commonest lesions recognised on antenatal USS and are important because of the high risk of detrimental long-term neurodevelopmental outcomes. Enlargement of the lateral cerebral ventricles (or ventriculomegaly VM) is by far the commonest intracranial abnormality recognised on antenatal USS (approximately 2.5/1000 pregnancies) but the clinical significance of this finding is complex. VM can be associated with other structural abnormalities (in which case the prognosis is often poor) but most often it is the only abnormal intracranial finding. In this situation, prognosis is related to the degree of VM, fetuses with larger ventricles generally have a higher risk of poor neurodevelopmental outcome. The ‘Magnetic Resonance imaging to enhance the diagnosis of fetal developmental brain abnormalities in utero’ (MERIDIAN) study is a multicentre prospective cohort study based in the UK designed to evaluate the diagnostic and clinical benefits of in utero Magnetic Resonance (iuMR) imaging in diagnosing fetal brain pathology antenatally. A description of the overall findings from the cohort has been reported previously and has confirmed major diagnostic and management impact when iuMR is included in the diagnostic pathway. At the planning stage of the study, focus groups of This article is protected by copyright. All rights reserved. A cc ep te d A rti cl e fetal medicine consultants were asked what information MERIDIAN should attempt to provide in order to improve clinical practice and the most frequent response called for a sub-group analysis of apparently isolated VM on USS. In this paper, we detail the diagnostic and clinical impact brought about by including iuMR in the diagnostic pathway in 306 fetuses with VM as the only abnormality shown on antenatal USS. This article is protected by copyright. All rights reserved. A cc ep te d A rti cl e METHODS The fetuses reported in this paper came from the MERIDIAN cohort and the overarching methodology of that study is reported elsewhere but the parts relevant to this paper are summarised here. MERIDIAN was undertaken in accordance with the Medicines for Human Use (Clinical Trials) Regulations 2004 and ethics approval was obtained for a multi-centre study through the Integrated Research Application System (62734). Inclusion criteria were – pregnant women aged ≥16 years whose fetus had a brain abnormality detected by USS at a gestational age of 18 weeks or more, with no contraindications to iuMR and gave written, fully informed consent to enter the study. Recruitment was from 16 fetal medicine units in the UK. Ultrasonography and iuMR imaging All of the USS studies were performed by appropriately trained NHS consultants in fetal medicine and they recorded each brain abnormality using nomenclature from the “ViewPoint” antenatal ultrasound reporting software (GE Healthcare, Chalfont St Giles UK). Each woman then had an iuMR examination performed on a 1·5T superconducting clinical MR system at one of six centres. The base requirement for the iuMR study was T2-weighted images of the fetal brain in the three orthogonal planes and a T1-weighted ultrafast sequence in at least one plane (usually axial). The reporting radiologist was aware of the diagnoses made on USS from the study paperwork and also had access to the full clinical USS report. The radiologist was required to comment on each brain abnormality recognised on USS (using ‘diagnosis excluded’ if the finding was not present on iuMR) and added extra anatomical diagnoses where appropriate onto a similar form as used for USS. This article is protected by copyright. All rights reserved. A cc ep te d A rti cl e In cases where the pregnancy continued and the child survived, outcome reference data (ORD) was obtained from the clinical notes based on diagnoses made on postnatal neuroimaging studies up to the age of six months, with the exception of those fetuses described in b) below. In cases of Termination of pregnancy (TOP), stillbirth or neonatal deaths the ORD was based on autopsy and/or post-mortem MR imaging. An independent paediatric neuroradiologist determined whether a full review by a Multidisciplinary Independent Expert Panel (MIEP) was required. Review was required in every case unless; a) There was complete and unequivocal agreement between the abnormal findings on USS, iuMR and the ORD; or b) VM was the only finding described on both USS and iuMR examinations but the size of the ventricles had returned to normal as shown on USS later in pregnancy or on neonatal imaging. The latter was counted as agreement in recognition that VM commonly resolves spontaneously during later pregnancy. The MIEP consisted of three consultants working in the NHS (neuroradiologist, fetal medicine consultant, and paediatric neurologist) from a centre that did not recruit into MERIDIAN. They were asked to judge if the USS-based diagnoses and iuMR-based diagnoses were in complete agreement with the ORD. The primary analysis in MERIDIAN centred on participants who underwent iuMR within 14 days of USS and for whom ORD was available and consisted of a calculation of diagnostic accuracy for USS and iuMR studies defined as; [True positives + True negatives]/Total number of cases This article is protected by copyright. All rights reserved. A cc ep te d A rti cl e Potential differences in the diagnostic accuracies of USS and iuMR were assessed by McNemar’s paired binomial test. The subgroup reported in this paper consists of women who were recruited into MERIDIAN carrying a fetus with VM as the only finding on USS. Fetal VM was categorised into three groups based on the size of the largest trigone measured on USS as used clinically in the referring centres namely; • mild (10, 11 and 12mm) • moderate (13, 14 and 15mm) • severe (16mm and over). Presence of other brain abnormalities The main analysis consists of calculations of diagnostic accuracy of USS and iuMR by comparing of the ORD with the imaging-based diagnoses in the entire sub-group and then in relation to the severity of VM diagnosed on USS. Assessment of ventricular size The fetal medicine consultant performing the USS was asked to record the ventricular measurements as left or right trigone and to specify which one was the near-field measurement. There are frequently problems in visualising the near-field ventricle on USS and hence measurements may be difficult for that hemisphere. The trigone measurements made on iuMR imaging were recorded and compared with the USSderived measurements. We first compared the category of VM made on USS and iuMR by taking the size of the largest trigone and classified it as mild, moderate or severe using the definitions given above. We then compared the absolute measurements of the left and right trigones made on USS and iuMR and describe the difference in order to This article is protected by copyright. All rights reserved. A cc ep te d A rti cl e judge the degree of agreement. We also compared the degree of agreement in relation to which measurement was near-field and which was far-field on USS. We also assessed the risk of other brain abnormalities in relation to unilateral or asymmetric VM as judged by iuMR imaging. In accordance with our previous publications, unilateral VM is defined in a fetus with one t