Paediatric and perinatal postmortem imaging: the need for a subspecialty approach

Evaluating the true value of postmortem imaging is difficult, but direct comparison between postmortem imaging and autopsy findings have recently been published [11, 13]. In adults, the concordance rates of adult postmortem imaging and full autopsy findings in the only blinded study published to date were disappointing, at around 50% [13]. This included 180 cases, and the major discrepancy rate between cause of death identified by radiology and autopsy was 32% for postmortem CT and 43% for postmortem MRI. Postmortem CT appears to be a more accurate imaging technique than postmortem MRI for providing a cause of death in adults, with a similar error rate to traditional autopsy/clinical death certification. Typical errors included ischaemic heart disease, pulmonary embolism and pneumonia. Whilst there is room for improvement, postmortem CT and postmortem CT angiography (CTA) are already becoming the workhorse imaging modality for adult imaging [14‐16].

Recent data confirm that postmortem MRI is likely to become the standard for postmortem imaging in children. A recent benchmark prospective validation study of postmortem imaging in foetuses and children at a specialist children’s hospital in London found a >90% concordance rate between noninvasive postmortem assessment (including postmortem MRI and ancillary investigations not requiring invasive procedures, such as placental examination) and conventional full autopsy findings in 400 cases (277 foetuses, 123 children) [11]. Interestingly, this was greatest for foetuses (<24 weeks 95%, >24 weeks 96%), less good for newborns (81%) and infants (85%), and least good for children ages 1–16 (54%), although these data probably reflect to some extent smaller sample sizes of the latter groups. Based on noninvasive postmortem findings, around 40% of traditional autopsies were judged to be unnecessary, and in these cases there was 99% concordance between conventional autopsy and minimally invasive autopsy. The lower concordance in children was primarily due to undetected infectious pathologies, including pneumonia and myocarditis, for which conventional postmortem MRI currently has a low detection rate [11].

The study also demonstrated that postmortem MRI was particularly useful for intracranial pathology in foetuses, in which the inherent fragility of the foetal brain leads to difficulties with traditional neuropathological examination even following fixation, and in whom adequate pathological examination of the brain may not be possible in around 20% of cases due to changes of autolysis, maceration and artefacts. Of the cases in which intracranial postmortem MRI examination was normal, detailed formal neuropathological examination provided clinically significant new information in less than 1% [11].

Importantly, however, these data also demonstrate that use of postmortem MRI alone, without involvement of a pathologist to perform and interpret ancillary postmortem investigations in conjunction with the clinical features currently has a poor diagnostic yield (around 50%). If this approach is offered in conjunction with noninvasive external examination, then the limitations must be appreciated and adequately explained. Parents should be informed that noninvasive autopsy, including postmortem MRI and ancillary, minimally invasive investigations, performed jointly by a pathologist and radiologist, can in the majority of cases have a similar accuracy to that of conventional autopsy for detection of cause of death and/or major pathology. This approach could therefore allow a triage process in which initial postmortem MRI and other investigations are performed with progression to full or modified autopsy as indicated by the results of the noninvasive postmortem findings. This approach is likely to be useful in improving the uptake of postmortem evaluation for parents in whom current approaches are unacceptable.

Whilst neurological and cardiovascular disease (such as coronary artery related disease) are likely to account for the vast majority of sudden death in adults, in children the spectrum of disease is markedly different [11, 17]. For foetuses and neonates, congenital abnormalities and complications of delivery account for the majority of deaths, and for foetuses terminated for antenatally detected structural anomalies, genetic and syndromic disorders are the predominant category [18] (Figs. 2,3, and 4). In infancy, the single most common group is sudden and unexpected deaths, often termed SIDS, cot-death or crib-death, in which the mechanism remains uncertain and the diagnosis is one of exclusion following autopsy. Of unexpected infant deaths in which a medical cause is identified, infections, such as unrecognised respiratory tract infection, represent the largest group [19]. Therefore, the optimal strategies for investigating such deaths, including both imaging and ancillary investigations, vary significantly with age. A thorough understanding of the spectrum of likely pathologies is required in order to direct a rational death investigation strategy.

Furthermore, in the paediatric setting, especially foetal and neonatal deaths, there has often been some form of antenatal imaging, usually US. Evaluation of this imaging is useful in several ways, since antenatal findings may be confirmed or refuted on postmortem imaging (such as ventriculomegaly [20]). An individual with no abnormality on either antenatal US or postmortem cross-sectional imaging is unlikely to have a significant anatomical abnormality missed (85% agreement) [21]. Excellent antenatal sonography skills combined with good postmortem imaging may preclude the need for formal invasive autopsy, but this has yet to be fully evaluated.

In some circumstances, histological evaluation of tissue samples is required to determine the cause of death or provide a specific diagnosis and may be the most useful ancillary investigation [22]. In such cases, following postmortem imaging to direct further procedures, tissue may be obtained by either an open or minimally invasive approach, such as the use of endoscopic guided tissue sampling [23], which is likely to represent a more acceptable approach compared to traditional autopsy for many parents [24, 25].

Just as in adult postmortem imaging, the field of paediatric and perinatal postmortem imaging is highly specialised and relatively new, with relatively few interested and qualified people around the world with the interest and expertise to develop this service further. There is a small but substantial overlap with a range of other fields including forensic imaging (predominantly adult-oriented), paediatric imaging in live children (which share many of the disease and imaging characteristics), and paediatric and perinatal pathology. Perinatal pathology is a separate subspecialty in the United Kingdom with a dedicated training pathway and accreditation.

The skills of a paediatric radiologist, according to the revised European training curriculum for radiology [26], should include:

Core radiological skills required include basic anatomy, radiation protection training and physics as applied to several different types of imaging modality. The appropriate use of each imaging modality requires expertise and varies according to the clinical question being addressed and the skill of the radiologist.

Several aspects of foetal development and antenatal imaging are now performed by obstetricians or foetal medicine specialists around the world, as antenatal and paediatric radiology services may be physically separate within a hospital campus. As a result, there is a risk of relatively limited knowledge of foetal imaging among many radiologists, and limited knowledge of MR techniques amongst those who perform antenatal imaging (mostly US). Above and beyond those requirements, forensic imaging requires an in-depth understanding of those cases where the cause of death is uncertain, including suspected inflicted or non-accidental injury, and the concept of SUDI. Forensic imaging also requires a working knowledge of national legal procedures and in-depth knowledge of forensic paediatric medicine, including how to prepare reports and interpret radiological findings for court.

The perinatal pathology skill set has long been recognised to be separate from that in general pathology [27], and now forms its own separate specialty in the United Kingdom. Whilst a common curriculum of basic histopathology training encompassing surgical pathology, autopsy and cytopathology and generic professional discipline are core to any pathologist, subspecialty training in paediatric and perinatal pathology should include:

Training specifically includes the ability to perform autopsies across a range of clinical scenarios including early foetal loss, spontaneous abortion, termination of pregnancy for foetal abnormality, stillbirth, intrapartum death, SUDI and other settings such as specific cardiac, hepatic or metabolic disease. The Royal College of Pathologists (England) states that knowledge of newer radiological techniques such as postmortem MRI, the ability to request radiology appropriate to the case and to appreciate the importance of obtaining expert radiological opinion should be part of the curriculum [28].

Just as paediatric imaging in the living requires a gentle approach in a child-friendly environment to attain maximal success rates, a paediatric postmortem service requires an equally sensitive and ethically sound approach, with an understanding of the specific medicolegal framework in which the work is performed, which may vary with institution, jurisdiction or country.

The International Society of Forensic Radiology and Imaging (ISFRI) is developing a comprehensive training and accreditation programme for postmortem imaging, predominantly organised around the use of radiographs and postmortem CT and CTA [29]. The recent ISFRI 2013 meeting was dedicated to developing a training strategy for adult postmortem imaging, with focus in five areas including data acquisition, analysis and storage, recommendations for reading/reporting, development of a basic curriculum, certification/revalidation issues, and a collaborative platform for data sharing [30]. All of these issues are equally valid for paediatric postmortem imaging. The need for education has previously been addressed highlighting the specific needs and knowledge base mandatory for reporting postmortem imaging [31].

Furthermore, national reference guidelines are now being written regarding practical postmortem service delivery. Postmortem MRI should be performed within already established centres of specialist paediatric/neonatal pathology, with postmortem MRI carried out by trained MR radiographers, potentially outside normal working hours [32]. In view of the specialist nature of the paediatric examination, those with dedicated specialist imaging skills should perform the assessment. This would be facilitated by centralising services in experienced regional centres. Some of the skills for postmortem MRI would need to be learned through an established national training scheme for paediatric radiologists, and funded research programmes are needed to build the evidence base to determine the types of death in which cross-sectional imaging is an adjunct to, and those in which it can be used as a replacement for, traditional autopsy. Standards of practice will need to be developed, probably establishing the pathologist in a central coordinating role in the investigation of the cause of death, working closely with practitioners who perform and interpret postmortem imaging studies [33].

A training programme dedicated to perinatal and paediatric postmortem imaging interpretation and reporting is clearly needed, given the wide range of congenital and genetic defects commonly seen in the obstetric antenatal screening population, as well as the effects of postmortem changes on imaging appearances. Medical personnel reporting such postmortem imaging must be appropriately trained, irrespective of their background, and will require combinations of skills from different disciplines. These may include common anatomical and radiological issues, including embryology, an extensive knowledge of congenital abnormalities including skeletal dysplasias, imaging techniques and optimisation, and the limitations of each imaging technique, as well as pathological issues, including decomposition, mechanisms of death and the normal range of pathologies that are likely to be encountered.

One limitation to the implementation of postmortem imaging in children is the availability of CT and especially MR scanners. In most hospitals, scanner availability is limited and full to capacity during daytime working hours of clinical imaging for live patients, as would be expected. This implies that, in general, postmortem imaging needs to be performed outside of normal working hours, either in the evenings, at night or on weekends, to fit around conventional clinical lists, although additional costs may be incurred in doing so. To facilitate changes in working practices, collaboration with radiographers or radiological technicians is paramount, to involve them early on in the need for good-quality postmortem imaging.

What needs to be achieved in paediatric and perinatal postmortem imaging in the next decade to establish a working service? Collaborative work within the imaging community is required to optimise imaging protocols and postmortem MRI sequences, with imaging predominantly taking place in specialised centres to a consistently high standard. Clinical guidelines should be developed for the optimal use of different imaging modalities (radiography, US, CT and MRI), such that the most appropriate imaging investigations are carried out according to the clinical scenario. Clarification is required regarding the need for ethical approval or parental consent for imaging investigations, above and beyond that required for clinical assessment. A teaching programme should be developed to help equip those who wish to pursue this subspecialty further with the best current knowledge available from a range of experienced personnel. Ultimately, this will help to better understand difficult cases, such as SUDI — increasing the information collected in this particular cohort of individuals is likely to yield new diagnostic information. Close collaboration among pathologists and radiologists with mutual respect and recognition for different expertise is essential to simultaneously provide a high-quality patient-focused clinical service and avoid turf battles.