Post mortem magnetic resonance imaging in the fetus, infant and child: A comparative study with conventional autopsy (MaRIAS Protocol)

Background Minimally invasive autopsy by post mortem magnetic resonance (MR) imaging has been suggested as an alternative for conventional autopsy in view of the declining consented autopsy rates. However, large prospective studies rigorously evaluating the accuracy of such an approach are lacking. We intend to compare the accuracy of a minimally invasive autopsy approach using post mortem MR imaging with that of conventional autopsy in fetuses, newborns and children for detection of the major pathological abnormalities and/or determination of the cause of death. Methods/Design We recruited 400 consecutive fetuses, newborns and children referred for conventional autopsy to one of the two participating hospitals over a three-year period. We acquired whole body post mortem MR imaging using a 1.5 T MR scanner (Avanto, Siemens Medical Solutions, Enlargen, Germany) prior to autopsy. The total scan time varied between 90 to 120 minutes. Each MR image was reported by a team of four specialist radiologists (paediatric neuroradiology, paediatric cardiology, paediatric chest & abdominal imaging and musculoskeletal imaging), blinded to the autopsy data. Conventional autopsy was performed according to the guidelines set down by the Royal College of Pathologists (UK) by experienced paediatric or perinatal pathologists, blinded to the MR data. The MR and autopsy data were recorded using predefined categorical variables by an independent person. Discussion Using conventional post mortem as the gold standard comparator, the MR images will be assessed for accuracy of the anatomical morphology, associated lesions, clinical usefulness of information and determination of the cause of death. The sensitivities, specificities and predictive values of post mortem MR alone and MR imaging along with other minimally invasive post mortem investigations will be presented for the final diagnosis, broad diagnostic categories and for specific diagnosis of each system. Clinical Trial Registration NCT01417962 NIHR Portfolio Number: 6794


Background
For over 500 years a post mortem examination has been used to establish cause of death. This procedure provides valuable information on pathological processes -one of the key foundations of medical education. Perinatal and neonatal post mortem examination has a particularly valuable role; this was formally recognised some 15 years ago when the Royal College of Obstetricians and Gynaecologists (RCOG) and Royal College of Pathologists recommended that a perinatal post mortem examination rate of less than 75% was unacceptable and that the ideal was 100% [1,2]. Autopsy rates have steadily declined over the years since this document was published [2][3][4][5]. This decline has been accelerated by adverse publicity surrounding alleged organ retention without formal parental * Correspondence: s.thayyil@ucl.ac.uk 1 Institute for Women's Health, University College London, London, UK Full list of author information is available at the end of the article consent in the Bristol Royal Infirmary Inquiry [6] and the Royal Liverpool Children's Inquiry [7]. Neonatal post mortem examination consent rate was less than 20% in England and Wales in the most recent report of the Confidential Enquiry into Stillbirths and Deaths in Infancy (CESDI) [8].
The loss of a fetus, baby or child is devastating to parents. As well as coping with their loss, parents often want to know why their child died, and if there is an increased risk for existing children or for future pregnancies. A post mortem examination may provide this information. In 14-46% of perinatal and infant post mortem examinations, additional clinically significant information is found beyond that known prior to the examination, which would affect counselling or recurrence risks [3,4,9,10]. The findings may confirm or refute clinical diagnoses made during life. Many studies report significant disagreement between the pre-morbid diagnosis and post mortem examination in at least 10% of cases. This impacts both upon recurrence risks and the approach to prenatal diagnosis in future pregnancies [3,11]. Post mortem examination thus has a valuable place in confirming or refuting pre-morbid diagnoses, making further diagnoses and identifying genetic and obstetric factors of relevance to the management of future pregnancies, allowing appropriate counselling of families who can then make informed, reproductive choices. The post mortem examination will also provide useful information for clinicians, helping them to understand the causes and effects of diseases as well as the effectiveness and complications of treatment. In addition, the post mortem examination can play a crucial role in research and so advance the progress of fetal and paediatric medicine.
Should a post mortem examination be performed, recent alterations to the post mortem examination procedure and consent process may reduce the amount of information available, especially for central nervous system abnormalities [12][13][14]. Until recently, the usual practice was to remove and fix the brain before dissection, a process that could take up to 3 weeks. Parents now frequently request that all organs are replaced before burial. As adequate fixation is difficult within this time, the brain has to be examined following a suboptimal period of fixation, which can make interpretation of the developing brain difficult. Delay between intrauterine death and delivery, leading to maceration of the fetus, makes brain examination more difficult for the pathologist. The RCOG guidelines state that any pregnancy terminated after 22 weeks gestation should be accompanied by fetocide to ensure that the fetus is not born alive. This procedure is usually accompanied by the administration of mifepristone (a cervical ripening drug), which has its optimum efficacy in shortening the time between induction and delivery after 48 hours [15]. This effectively means that most delivered fetuses undergoing termination of pregnancy after 22 weeks gestation will have been dead for at least 48 hours, rendering post mortem examination of the brain difficult.
In addition to the difficulties of acquiring consent to perform conventional autopsy, and sufficient time to perform optimal histological preparation, various religious communities find conventional autopsy unacceptable [1]. Provision of a less invasive, accurate and widely available method of post mortem assessment has been advocated [1] and would enable access to post mortem information for the first time for many in these communities.
In summary, a less invasive method of accurately assessing detailed anatomical and pathological changes in all body systems after death would be of great value. Information for diagnosis and clinical audit can be obtained as well as creating a permanent electronic record of findings, whilst allaying parental concern with regard to organ retention or conventional invasive post mortems.
Though conventional radiology to assess the chest and bones has been used for some time in post mortem examination (for example, a skeletal survey is performed on all paediatric cases referred to the coroner and in all perinatal cases), MR imaging would be well suited as a non-invasive imaging modality for post mortem assessment. Standard imaging protocols could be performed in any hospital equipped with an MR scanner and the images sent to a centre of expertise for reporting. MR imaging would potentially overcome some of the weaknesses of conventional autopsy, providing a complete multisystem analysis that is non-invasive.
MR imaging of the excised brain [16], spine [17] and heart [18] has been successfully performed. However, although an initial feasibility study of whole-body post mortem MR imaging was reported in 1996 [19], its use in clinical practice has remained controversial. Several small studies of whole-body post mortem MR in fetuses have been reported [20][21][22]. In all fetal studies, imaging of the central nervous system (CNS) proved the most accurate, whilst body imaging, in particular imaging of the heart proved more problematic. A recent study, focussed on the diagnosis of CNS abnormalities in fetuses and stillbirths, reported a sensitivity of 100% and specificity 92% for MR compared with conventional post mortem examination [23]. Other studies have confirmed the accuracy of CNS fetal post mortem MR [24,25]. Imaging of the other body systems has been less well documented. Our own recent experience is that accurate post mortem body MR image acquisition is possible with modern MR imaging sequences (unpublished data), but post mortem MR imaging of the heart is less accurate [26][27][28].
Furthermore, MR imaging has grown in clinical importance in the living fetus and newborn infant [29], especially for brain anomalies. There is now extensive literature describing the normal MR imaging appearance of the in utero fetal brain from around 17 weeks gestation [30] and the ex-utero preterm infant brain from around 25 weeks gestation [31].
The decline in parental consent for autopsy, and technical limitations of conventional autopsy to define some nervous system abnormalities, together with a reduction in number of skilled perinatal pathologists and morphologists, has lead to a need to seek alternative less invasive methods for post mortem examination of the fetus, neonate and child. In 2001, the UK Chief Medical Officer recommended that modern imaging methods should be evaluated [32]. Since then several reports on forensic aspects of post mortem imaging have been published, however, these studies are limited to post mortem imaging in adults, primarily using computerised tomography (CT) and many studies were of poor quality [33].
Our previously published systematic review on post mortem MR imaging in fetuses, newborns, children and adults demonstrates that there is insufficient evidence to recommend the use of post mortem MR imaging as an alternative for conventional autopsy [34]. Most comparative studies to date have been small and/ or have compared single systems such as the brain and did not have adequate blinding of radiologists and pathologists due to their retrospective nature. In particular, none have (a) systematically examined all the body systems in a large series of fetal, neonatal and childhood deaths (b) assessed the MR appearance of death-induced artefacts or the effect that death and maceration may have on the MR image (c) have assessed the possible disadvantages or advantages of a minimally invasive post mortem examination in combination with MR imaging.
Thus, over a decade after the first description of post mortem MR imaging, we still lack the evidence for routine implementation. Here we describe a large, prospective, blinded, comparative study to evaluate MR as an alternative to conventional invasive autopsy in fetuses, newborns and children.

Hypothesis
MR imaging can provide an accurate, detailed, threedimensional post mortem record of structural abnormalities and the disease processes of the whole body in the fetus, neonate and child, with similar diagnostic information to a conventional autopsy.

Primary objective
To compare the accuracy of whole body post mortem MR imaging for detecting the cause of death and/or major pathological lesions with that of conventional autopsy in fetuses, newborns and children.

Secondary objectives
To compare ante mortem imaging assessment (ultrasound and MR) of fetuses with post mortem MR and CT images.
To compare ante mortem diagnosis, including imaging data, in neonates, infants and children with post mortem MR images.

Methods/Design
The study has been ongoing at two hospitals: Great Ormond Street Hospital for Children NHS Trust (GOSH) and University College Hospital NHS Foundation Trust (UCH), since March 2007. These hospitals are associated with a single academic institution -University College London (UCL). All recruited cases underwent post mortem MR imaging at 1.5 T (Avanto, Siemens Medical Solutions, Enlargen, Germany) as well as a conventional autopsy. CT imaging was also performed in cases with suspected traumatic injury or skeletal dysplasia.
In line with the CESDI recommendations, post mortem examination was offered in all cases of perinatal death and consent sought by the appropriately trained staff (consultant, experienced nurse, or experienced midwife). For consented autopsies, the standard National Health Service (NHS) consent form (produced by Department of Health) that includes consenting for the use of post mortem imaging for research was used (consent form and patient information sheet given in appendix 1 and 2) [35,36].
In Her Majesty's (HM) Coroner's cases no parental consent for autopsy was required and so, once the body was received by the GOSH mortuary, a member of the research team contacted the HM coroner's office for permission for a bereavement nurse to approach the parents by telephone to gain consent for MR. If the parents gave verbal consent, a pre-paid envelope with consent form and information leaflet was sent to the parents (see reference 35 for full details of this process). Once MR consent was obtained, a post mortem MR was performed prior to standard autopsy. Great Ormond Street Hospital and Institute of Child Health Research Ethics Committee (04/Q0508/ 41) approved the study.

Post mortem MR imaging
The first 20 subjects were used to optimise the imaging sequences. These subjects will not be included in the main study. The optimised MR study protocol is given in Table 1. A team of four specialist radiologists (paediatric neuroradiology, paediatric cardiology, paediatric chest & abdominal imaging and musculoskeletal imaging) Each reported the MR image, blinded to the autopsy report. Each radiologist reported the post mortem MR independently on to a large Microsoft access database (Microsoft Inc, Redmond, USA), with predefined drop down menus of categorical variables and codes (based on standard autopsy reporting). The MR data will be then re-classified jointly by a radiologist and a pathologist with regards to the likely final diagnosis, broad diagnostic categories and specific abnormalities of each organ system.

Conventional autopsy
Experienced paediatric or perinatal pathologists performed all autopsies according to the Royal College of Pathologists' guidelines, with input from specialist paediatric cardiac pathologists or neuropathologists as required. The pathology data were entered into the same database using an independent access portal, blinded to the MR data. The categorisation will be the same as for the MR imaging (i.e. specific final diagnosis, broad diagnosis category and separate system normal/abnormal rating) ( Table 2). This provides the gold standard against which the MR imaging is assessed.

Data analysis
Using conventional post mortem as the gold standard comparator, the MR images will be assessed for accuracy of the anatomical morphology, associated lesions, clinical usefulness of information and determination of the cause of death. The primary outcome will be the percentage of cases for which MR imaging correctly identifies the diagnostic category. Secondary analyses will make more detailed comparison of the individual measurements recorded within both autopsy types to facilitate understanding of how different diagnoses may occur. The sensitivities, specificities and predictive values of: (i) MR imaging plus clinical history, and (ii) MR imaging plus clinical history and other non-invasive post mortem investigations (e.g. external examination, genotyping, placental examination and skeletal survey) and to identify: (a) the specific diagnosis (b) one of predefined broad categories of diagnosis and (c) specific diagnostic category for each system, will be presented.
Changes according to weight of the fetus/infant/child will be investigated using logistic regression models. Diagnostic statistics will also be presented for the individual components (brain, chest, abdomen, heart and musculoskeletal) to determine areas of good and bad concordance. All estimates will be presented with 95% confidence intervals. Ante mortem assessments will be similarly compared with those made post mortem.

Sample size calculations
To determine the primary outcome of percentages correctly diagnosed to within +/-5% with 95% confidence will require 400 cases if the percentage correct is as low as 50%. We anticipate that the percentage correct will be substantially higher than this, in which case the estimate will be more precise. If the percentage correct is as high as 90%, then this will be estimated to within +/-3% with a sample of 400.
Similarly for the individual components, we anticipate a wide range in the age distribution of cases and this should allow some quantification of changes by weight. Further investigations within specific diagnoses will necessarily be based on smaller numbers and are hence more exploratory     than definitive in nature. Presentation with confidence intervals will indicate any under-powering of the comparisons.

Discussion
Over the past decade, the consent rate for autopsy in the newborn has been less than 20% in the UK [8]; pediatric autopsies have become virtually non-existent, apart from the cases investigated by the HM Coroners or police (forensic cases), where parental consent is not required. Furthermore, this decline has occurred despite an increase in the number of cases in which autopsy consent (approximately 80% of cases following a perinatal death) is sought by the clinicians. One of the key reasons for parental refusal is the apparent invasive nature of conventional autopsy. Although post mortem MR imaging was reported as an alternative for conventional autopsy more than a decade ago, it has not been introduced into routine clinical practice as supporting evidence is based on small and poorly designed studies [35]. In the UK, its use has been limited to some private initiatives. The Chief Medical Officer (UK) recommended rigorous evaluation of post mortem MR imaging as an alternative for autopsy, before it is widely introduced into the UK clinical practice. Following this, the UK Department of Health funded the present study (MaRIAS-Magnetic Resonance Imaging Autopsy Study) for systematic and rigorous evaluation of post mortem MR imaging as an alternative for conventional autopsy with a view to making recommendations to the Department of Health with regard to the advisability of introduction into routine clinical practice, either on its own or along with other minimally invasive post mortem investigations [37].
Quality assurance of the data Transparent Research Audit System (TRUST) guidelines will be followed for analysis, authorships and publication of the MaRIAS data. AMT re-examined all MR images and cross checked against the radiology reports on all organ systems for quality assurance. Shea Addison (SA) will cross check all the pathology data for any data entry errors. AMT, NJS and SA will be responsible for cleaning the data. The final data used for analysis and accurate contributions of each author will be stored at UCL until 2032 for research data auditing. ST and AMT will be the custodians of the MaRIAS data. All manuscripts from the study will be reviewed by the Department of Health (UK), ST and AMT before submission, though the Department of Health (UK) does not influence the scientific content of any research output. Information leaflet for parents Thank you for taking the time to read this leaflet. We know that this is a difficult time for you and appreciate the time you are taking to read this leaflet.
Background to the study MRI (Magnetic resonance imaging) and CT scans, as you may be aware are special techniques to get images of the body. An MRI scan can examine internal organs in detail and may be able to identify some of the problems that can be detected by a post mortem examination. In some cases, we believe that MRI may even be better than a post mortem examination. Many parents are understandably upset about the thought of their baby undergoing a post mortem. We are doing this study to find out if an MRI scan of the whole body can give similar information to that of post mortem, so that in future we might be able to offer an MRI scan instead of post mortem.
What will happen if we agree to take part? If you agree to take part we will arrange for your baby to have an MRI scan (and in some cases a CT scan as well) as soon as possible at Great Ormond Street Hospital for Children. This involves taking a series of pictures using a special machine. We may take biopsy using small needles under MRI guidance, for examination under microscope. The whole process will take about 2 hours. As soon as the scan is done we will arrange for your baby to be taken for the traditional post mortem. We will ensure that at all times your baby will be treated with due respect and reverence.
The MRI scan will not delay the post mortem or the timing of burial or cremation. Taking part would not involve you in any extra hospital visits. Any additional information, if any from MRI/CT scan will be included in autopsy report to the coroner. We will also need to have access to the post mortem results and the results of any other tests that were done before or after birth. This is so that we can compare the results of tests which are done traditionally with the results from the MRI and work out which combination of tests give the most accurate results overall.
Will my taking part in this study be kept confidential?
All information that is collected about you or you baby or during the course of the research will be kept strictly confidential. Any information we collect will only be used by the research team for the purpose of the study.
Who will have access to the case/research records? All the data and images collected as part of this study will be stored on a secure computer. Only the researchers involved in this study will have access to the data collected in the course of this study. A representative of the hospital's Research Ethics Committee will also have access to data. The 1988 Data Protection Act safeguards the use of some types of personal information. This places an obligation on those who record or use personal information, but also gives rights to people about whom information is held. If you have any questions about data protection, please contact the Data Protection officer via the switchboard on 0845 155 5000. The results from our project will be published as papers in medical journals. No data will be published that allows for individuals to be identified in any way. If requested, we will be able to send you copies of any papers published when we have completed the study in 3-4 years time.
Do you have to take part? It is up to you to decide whether or not to take part. If you do decide to take part you will be given this information sheet to keep and will be asked to sign a consent form. You will be given a copy of the signed consent form for your records. If you do not feel able to take part it will not in any way affect the care your family receives.
Who do I speak to if I have further questions or worries?
In the first instance please contact xxx who is coordinating this project. His contact details are given below. xxx can also be contacted if you need any further information or xxx is not available. If you wish to speak to someone not directly involved in the study then please contact xxxxx If you have any complaints about the way in which the project is being or has been conducted, in the first instance please discuss them with any of the doctors listed below. If the problems are not resolved, or you wish to comment in any other way please contact xxxx Who is organising and funding the research? This study is being organised by the Cardiothoracic, Radiology and Pathology Departments at Great Ormond Street Hospital for Children and by the Fetal and Neonatal Medicine Units and Pathology Department at University College London Hospital, Funding is provided by the Department of Health. This study has been reviewed and approved by the Great Ormond Street Hospital Research Ethics Committee.
Thank you once again for all your time and trouble. Contacts for further information: XXXX