Less‐invasive autopsy for early pregnancy loss

Autopsy investigations provide valuable information regarding fetal death that can assist in the parental bereavement process, and influence future pregnancies, but conventional autopsy is often declined by parents because of its invasive approach. This has led to the development of less‐invasive autopsy investigations based on imaging technology to provide a more accessible and acceptable choice for parents when investigating their loss. Whilst the development and use of more conventional clinical imaging techniques (radiographs, CT, MRI, US) are well described in the literature for fetuses over 20 weeks of gestational age, these investigations have limited diagnostic accuracy in imaging smaller fetuses. Techniques such as ultra‐high‐field MRI (>3T) and micro‐focus computed tomography have been shown to have higher diagnostic accuracy whilst still being acceptable to parents. By further developing and increasing the availability of these more innovative imaging techniques, parents will be provided with a greater choice of acceptable options to investigate their loss, which may in turn increase their uptake. We provide a narrative review focussing on the development of high‐resolution, non‐invasive imaging techniques to evaluate early gestational pregnancy loss.


| INTRODUCTION
Pregnancy loss is a common event with an estimated 23 million miscarriages (<24 weeks gestation) and nearly 2 million stillbirths (>24 weeks gestation) occurring worldwide every year. 1,2 These numbers are increasing and could be far higher as many women may not present to the health services and data is therefore often based on approximate estimates. [3][4][5] Any pregnancy loss can be traumatic for the family and can lead to prolonged grief, depression, anxiety, and post-traumatic stress, with effects possibly lasting for several years. [6][7][8][9][10] Alongside this, miscarriage is little understood by both the general public and healthcare professionals and has been recognised as an under investigated area of medicine. 1,11 Compounding these issues around awareness, parents often believe that they are somehow at fault, which can stop them seeking help and support from healthcare professionals, preventing the true number of miscarriages in the community to be identified. 4,12 In this traumatic context, the delivery of information and its reception by the couple is particularly challenging. Therefore, providing support and a supportive environment where families can make an informed and appropriate choice for themselves is key.
Investigating the possible causes of pregnancy loss has multiple advantages including guiding future pregnancies and advancing medical knowledge, [13][14][15] but helping bereaved parents understand the implications of their choices at this difficult time can be extremely challenging. 16 Staff training is key to imparting complex information at this difficult time, and experience and guidance in this area for the staff is important to clearly convey key complex information. 14,17 This training should come in differing formats including formal courses, case discussions, and involvement in multidisciplinary meetings to raise awareness of the impact autopsy, and alternative less invasive investigations can make. 17,18 2 | HISTORICAL BACKGROUND Conventional autopsy (CA) is the traditional reference standard investigation into pregnancy loss and consists of multiple investigations also called feto-placental examination including a review of the clinical history, an external examination with photography, internal macroscopic examination and histological analysis, tissue sampling for genetic testing and macroscopic/microscopic examination of the placenta. 15,19 Together these investigations provide new clinical information in 40%-70% of cases, depending on the clinical question to be answered. 15,[20][21][22] Most perinatal and fetal autopsies are carried out with parental consent for diagnostic purposes, rather than being directed by the coroner/medical examiner for forensic purposes. Involving parents in the decision-making process is essential to increase uptake rates and provide care for the families. 13,23 Some parents choose conventional autopsy as the "reference standard" or maximum level of investigation possible as they wish to be sure they have excluded all possible causes for their loss, 15,24 but this may not be appropriate for all cases: there remains a significant proportion of pregnancy losses where the cause is still 'undetermined' despite full investigation. 15,25,26 Others decline all autopsy investigations as they wish to avoid 'further harm' to their child and are not prepared to undergo a more mental turmoil at this challenging time but can later regret this decision. 14,16,24,27,28 Parents choose to investigate their loss for multiple reasons, including to understand their child's death more fully, 29 the recurrence risk of future pregnancies, 30 to advance scientific knowledge, 31 for a degree of 'closure' after their loss 32 and to rule out selfblame 15,24 However, autopsy uptake has been declining for many years 10,17,22,[33][34][35][36][37] with parental reasons for declining including wishing to protect the baby, practical aspects over transfer to a specialist hospital, misconceptions regarding the benefits of autopsy, poor communication between families and healthcare professionals, uncertainty about the value of the procedure, a mistrust of the hospital system, religious and personal beliefs, and the invasiveness of the procedure. 10,14,38,39 Added to this, there is now a relatively small number of specialist paediatric pathologists (relative to the number of deaths to be investigated 40 ), meaning that access to specialist paediatric pathology services in the UK is becoming less accessible.
In this narrative review article, we aim to focus on the development of high-resolution, non-invasive imaging techniques to identify the possible causes of early pregnancy loss in this challenging area. We defined up to 20 weeks as "early" gestation but

| Post-mortem radiographs
Fetal post mortem radiographs (PMXR) are cheap, quick, widely available, easy to perform, and largely used to estimate gestational age and assess the developing skeleton. 47 As most skeletal dysplasias and abnormalities are now typically identified through ante-natal ultrasound, 58 PMXR of all fetuses has been shown to add diagnostically useful information in less than 1% of routine cases. 59 Whilst PMXR is diagnostic in suspected skeletal dysplasia, improvements in antenatal diagnosis mean that these abnormalities are being detected earlier at earlier gestations, when the bones are less ossified, and therefore PMXR are less discriminatory. 60-62

| Post-mortem computed tomography
Post-mortem computed tomography (PMCT) is also widely available, cheap, and quick to perform 47,63 but the combination of small size and poor soft-tissue contrast limits its diagnostic ability in small fetuses. 56,64,65 Although PMCT angiography can increase the soft tissue contrast, large validation studies are still required. 56,66,67 PMCT can also provide information on dysplasia imaging through 3D reconstructions, but with limited additional diagnostic utility due to limited ossification at early gestation negating this advantage in early pregnancy. 56

| Post-mortem magnetic resonance imaging (1.5 and 3T)
Post-mortem magnetic resonance imaging (PMMRI) is widely available and provides excellent soft-tissue information. However, due to the increasing clinical pressures combined with the time required to perform detailed PMMRI (approx. 60 min), 52,68 there is often difficulty in accessing these scanners for post-mortem imaging.
The main challenge for PMMRI at field strengths below 3T is the low spatial resolution and subsequent difficulty in visualising structures in <20 weeks GA due to poor tissue contrast and low signal-to-noise. 41,45,[52][53][54]69 PMMRI at 3T field strength provides fewer non-diagnostic scans than 1.5T for fetuses <20 weeks GA through better image contrast, with higher sensitivity (34.6% vs. 17.3%)), specificity (65.1% vs. 45.3%), and concordance (55.1% vs. 36.1%) with CA. 52 Field strengths above 3T have reported further improvement for fetuses <20 weeks gestation. 49,57,70 Soft tissue contrast and signal-to-noise ratio (SNR) can be improved through immersion of smaller fetuses in a Gadolinium/formaldehyde solution prior to scanning at 3T, but this technique has not been optimised for clinical use. 50 Overall, conventional PMMRI has limited diagnostic accuracy below 20 weeks gestation. 41 It may also be better understood by parents when consenting for post-mortem imaging having already undergone ante-natal ultrasound scanning and is also unaffected by tissue fixation. 61,72,[74][75][76][77] PMUS also has the advantage of being able to direct needle biopsies for tissue sampling. 47

| Micro-focus computed tomography
Micro-CT provides non-invasive micron-level resolution imaging ( Figure 1), with increased spatial resolution as fetal size decreases ( Figure 2), making it ideal for early GA pregnancy loss and providing isotropic datasets which can be reconstructed for visualisation in different planes. 41,57,[83][84][85][88][89][90][91][92][93][94][95] To provide adequate soft tissue contrast, however, an exogenous contrast agent is required, typically potassium tri-iodide (I 2 KI), which stains the soft tissue through cellular diffusion. 46,83,85 Sufficient time must be allowed for complete diffusion, which can take several days or weeks and is dependent on the size of the fetus ( Figure 3). 83 Iodination can cause visible darkening of the fetus, which should be discussed with parents during the consent process and may be initially undesirable but can be partially reversed by sodium thiosulphate solution (Figure 4). 46 Maceration, the breakdown of tissue structure due to immersion within a fluid filled cavity (amniotic fluid) combined with autolysis following fetal demise, is a major challenge in identifying the cause of early pregnancy loss both through imaging and CA. 25,70,79,103 Through the late diagnosis of a miscarriage and a delay between fetal demise, identifying pregnancy loss and delivery of the fetus, it is exposed to an often unknown intra-uterine retention time. Maceration affects many early pregnancy losses and can affect the accuracy of PMUS even within 48 h, with between 25% and 40% of scans becoming non-diagnostic due to severe maceration. 51,72,78,79,[104][105][106] Equally, in-utero maceration deteriorates image quality for micro-CT, yet high quality is still possible in over 95% of cases despite the presence of a range of maceration states. 102 Micro-CT can also be used as an adjunct to CA providing high resolution imaging of excised organs, for example, of the kidney, 107 brain 92 and heart. 57,93,94,108 Diagnostic accuracy of ex-vivo imaging is also high, for example, for congenital heart disease equal to CA and superior for smaller specimens 108 and has been used to assess cardiac anatomy between 8-13 weeks GA to validate prenatal diagnosis following TOP for cardiac anomalies. 57 One practical limitation of micro-CT is that due to the highresolution of the technique, the data files are large (approximately 50 GB per clinical patient) requiring large provision of storage for clinical use relative to other suitable techniques (ultra-high-field PMMRI <1GB per patient). 83,85 Overall, due to the relative higher-resolution, low cost, and quick scan times, micro-CT has clear advantages 71 and is an excellent alternative to CA. 48,57,84 It can be used to create detailed scale 3D models or augmented virtual reality to aid parental counselling and education of congenital anomalies/fetal development, 41,57,[83][84][85][88][89][90][91][92][93][94][95]109 with a range of fetal gestations scanning from 6 weeks 87 to 24 weeks 84 with relatively short scanning times. 83

| Ultra-high-field post-mortem magnetic resonance imaging
Despite the numerous studies and multiple technical advances in recent years, the main disadvantage for PMMRI is the low image quality below <20 weeks GA or 300 g. 49,70,110 To overcome this, HF-MRI (7T and above) offers a significant increase in the SNR which can either improve the resolution or decrease the scan time, whilst also offering superior soft tissue contrast, whilst reducing the complexity of introducing an exogenous contrast agent. 55,70,85,[110][111][112] HF-MRI images are able to accurately identify anatomical structures, particularly of the heart, below 20 weeks GA 49 albeit with a lower resolution than seen with micro-CT (137 μm in HF-MRI vs. 22 μm in micro-CT), 71 whilst a high concordance between HF-MRI and CA is seen for a range of anatomical structures with sensitivity (94.6%), specificity (97.6%) and positive and negative predictive values respectively (93% and 98.2%). 112 One study comparing CA with 9.4 and 1.5T PMMRI of the brain demonstrated that CA provided no additional information to that of HF-MRI and where CA was non-diagnostic, HF-MRI could provide diagnostic information in 69% of cases. 70 HF-MRI provided greater spatial resolution, higher tissue contrast and better diagnostic information than 1.5T and was equivalent to CA when there was minimal maceration. 70 Micro-CT can also provide increased SNR and CNR than HF- fetuses. 55 The smaller the fetus, the more likely micro-CT would be a better imaging modality, as even higher spatial resolution should be possible by positioning the fetus closer to the X-ray source, 71,83 whilst also maintaining shorter scanning times. 55 Overall, as with micro-CT, most HF-MRI scanners are research based with limited clinical availability 41,55,61,70 ; Relative to HF-MRI, micro-CT provides greater cost efficiencies and higher resolution. 71

| AUTOPSY IN ADDITION TO LIA
Whilst LIA techniques are proven to provide answers to parents, it is important to identify whether any additional yield would be possible from completing an invasive autopsy. 113   -945