Keywords

Historical value of the autopsy in research

The autopsy procedure and preservation of specimens

Research benefits of the autopsy

– Investigating the biology of malignancy

– Evaluate the effects of medical and surgical therapy

– Ensuring accurate epidemiological data

What makes a good research autopsy program work?

Historical Value of the Autopsy in Research

An autopsy , also known as a postmortem , is the medical examination of the deceased. It is a careful and detailed examination of the body and internal organs, in order to determine the cause of death and answer any clinical questions. The term “autopsy” means “to see for oneself” and refers to the fact that manifestations of disease are directly observed rather than relying solely on clinical findings and investigations [1, 2]. Modern medical understanding of disease originated when autopsies were first used to examine the structure and function of normal tissue . This could then be compared to the alterations seen in disease [3, 4]. Autopsies contributed to Virchow’s theories of cellular pathology in 1876, and to Osler’s great advancements of medical knowledge in the early 1900s [2]. Throughout the twentieth century autopsies played a key role in the explosion of medical knowledge. In recent times, autopsy research has contributed to the understanding of diseases as diverse as oesophageal adenocarcinoma, sudden cardiac death in young people, avian H1N1 influenza, and Creutzfeldt–Jakob disease [5–8].

Sadly, many believe there is no longer a place for autopsies in modern medical research ; that an autopsy cannot provide specimens that are adequate for modern research techniques, and that all we need to know about a patient and his disease can be derived from premortem clinical investigation and imaging . So is there any place for autopsies in modern research programs? The answer to this is a resounding “YES”! Autopsies can be used to obtain large quantities of tissue for research, assess response to therapy, map the distribution of metastases, highlight rare complications, provide feedback for quality assurance assessment of protocols and procedures, and provide reliable cause of death information.

The decline in hospital based autopsies and the more rigorous ethical standards and consent requirements for tissue retention can make research using autopsy data and tissue difficult [3, 9, 10]. However despite these hurdles, there is a recent resurgence of interest in the benefits of utilising autopsies for research, including obtaining tissue for molecular studies [10–12].

The Autopsy Procedure and Preservation of Specimens

The body is examined externally, then the internal organs are dissected, examined macroscopically, and tissue taken for histologic examination. An autopsy may be authorised by the state to establish a cause of death (coronial autopsy). Alternatively one may be performed at the request of clinicians or families to answer a clinical question, as part of a quality assurance measure, or to obtain tissue for research (hospital based autopsy). Coronial autopsies are a legal requirement to establish if the death was due to natural causes. Hence, there are additional legal and ethical standards that must be met if these autopsies are to be used for research. In contrast, hospital based autopsies are performed after a death certificate has been issued, and an autopsy is only performed with the consent of the patient (obtained prior to death) of the family.

An autopsy can be targeted, it can exclude designated organs (often the brain), or be a detailed examination of the whole body [1]. Appropriate consent for the autopsy, and in many countries the retention of tissue samples or organs, must be obtained. Autopsy research programs may also require research ethics approval, although not all countries mandate this [13]. Tissue and fluids retained for research purposes can be subject to a wide range of interrogation that includes histology, biochemistry, microbiological studies, immunohistochemistry, histomorphometric analysis, and molecular and biochemical studies [2, 14, 15].

“Rapid” autopsies are performed as soon after death as possible, specifically to obtain high quality tissue for research . The autopsy is ideally performed within 4 h of death to minimise postmortem tissue degradation [16]. Rapid autopsies are performed within an established research program. They require hospital infrastructure through which patients can be recruited and give consent prior to death. A team of pathology and laboratory staff experienced in biobanking need to be available on call and able to respond within a few hours of the tissue donor’s death [17, 18]. To expedite the removal and processing of tissue samples, clinical and radiologic information can be used to indicate which sites are to be biopsied or organs examined [17]. The extent of autopsy examination varies, with some programs continuing with a full autopsy protocol [18], while others simply obtain tissue using core needle biopsies of sites of interest. Of course, if a complete autopsy is not performed unexpected findings are much less likely to be identified.

Adequacy of Autopsy Tissue for Molecular and Biochemical Research

It is a common misconception that autopsy tissue is not good enough for molecular and biochemical studies [9, 19]. Although there is no doubt that degradation of nucleic acids can be an issue, this may be much less significant than many believe. For many research questions, the disadvantages are well outweighed by the availability of ample tumour and control tissue.

Several research groups, mostly using tissue from rapid autopsy programs , have successfully performed genomic copy number analysis of primary and metastatic tumours. The methods used include comparative genomic hybridisation [20, 21], fluorescent in situ hybridisation [22], and single nucleotide polymorphisms (SNP) analysis [20]. Using high-resolution genome-wide SNP arrays, Liu et al. [20] investigated 58 metastatic prostate cancer samples obtained at rapid autopsy from 14 subjects. They found subject specific data clustering of the 58 samples, suggesting a common origin of the metastatic cells [20]. Zarghooni et al. [40] also used SNP based DNA microarrays to analyse 11 diffuse intrinsic pontine gliomas (DIPG), which are lethal paediatric brainstem tumours. Nine of their cases were obtained at postmortem, with a postmortem interval range of 9–40 h. They found the genomic alterations were different to paediatric supratentorial tumours and identified two novel potential biological targets. To control for any postmortem alterations the samples were matched with their own normal brain tissue, and compared with the two available surgical biopsies [23]. This particular study was not from a rapid autopsy program, and several of their subjects died during terminal care at home.

Adequate preservation and evaluation of RNA is important for functional genomic studies. RNA is known to deteriorate with increasing time from death to autopsy. To consistently obtain high quality RNA requires an organised program that can efficiently obtain and process tissue. The median RIN (RNA Integrity Number) of a rapid autopsy program with a median postmortem interval of 3 h has been recorded as 8.9 for brain and 7.0 for body tissues [24]. Tissue samples are considered to be of high quality if the RIN ≥ 6.5, while samples with RIN ≥ 8.0 are considered suitable for all downstream molecular techniques [25]. Studies with a longer postmortem interval do show more significant deterioration in RNA quality [26]. Messenger RNA (mRNA) levels have also been determined from brain tissue obtained at autopsy, using reverse transcription followed by real-time polymerase chain reaction (PCR). While this showed a general decline in measured mRNA levels in the autopsy tissue, when the measured mRNA level was adjusted according to a reference gene mRNA level, most genes evaluated were not affected by the postmortem status. One gene did have significantly decreased adjusted mRNA levels. The results suggest that overall the pattern of gene expression in postmortem tissues is similar to surgical biopsy tissue, but carefully chosen controls are required. Factors that may alter RNA expression in postmortem tissue include both individual variation in gene expression and reduced production at the time of death, and possibly relate to the mode of death rather than the postmortem delay [9, 27].

Tissue obtained after a long postmortem interval may have partially degraded RNA, but this can still be utilised for PCR amplification of smaller fragments, so that tissue need not be wasted [28]. Increasingly, new technologies are being developed that can tolerate lower quality RNA samples, for example, NanoString® technology, meaning that gene expression can still potentially be evaluated [29]. DNA can also be obtained from postmortem tissue, including formalin-fixed and paraffin embedded tissue; however, larger DNA fragments are more prone to degradation than in surgically obtained formalin-fixed and paraffin embedded tissue [30].

Studies using proteins are more difficult and complex, as there is significant variation in degradation that is not predictable. Each study must therefore commence with an evaluation of the preservation of that particular protein [9, 31].

Expansion of Autopsy Derived Material: Resource Generation and Applications

Viable tumour tissue can be harvested to establish both in vitro and in vivo models, allowing in-depth studies of both primary and metastatic tumours [18, 32]. Several studies have confirmed the viability of growing fibroblasts from autopsy tissue, which can be reprogrammed into induced pluripotent stem cells [33, 34]. While mouse xenograft s of cultured human tumour cell lines have been in the researcher’s toolkit for many decades, patient-derived xenograft (PdX) resources are becoming increasingly sought after to provide greater clinically predictive insights. The PdX benefits pre-clinical research by preserving both tumour heterogeneity and tissue architecture, and by facilitating the modeling of specific stages of disease progression (for example, local metastasis, distant metastasis, and/or broad disease dissemination) in the absence of the clonal selective pressures of culture in monolayer [35]. Furthermore, clinical trials are routinely undertaken in cohorts with advanced disease, PdX models generated from metastatic deposits collected in rapid autopsies are certainly a more relevant pre-clinical model as opposed to the use of surgical resections of primary tumours [36–38].

Research Benefits of the Autopsy

Investigate the Biology of Malignancy

Use of autopsy tissue for research is particularly valuable for rare malignancies, those that are not managed by surgical excision, tumours that are frequently disseminated at the time of diagnosis, and metastases. For many of these special groups obtaining enough tissue for research studies can be problematic. If biopsies are sufficient for clinical diagnosis and management , taking more tissue may be unethical. However, biopsies may not be large enough for both clinical diagnosis and research studies. If subsequent treatment does not include surgical excision, further tissue may never be obtained. Malignancies such as pancreatic carcinoma are often disseminated at the time of diagnosis, consequently those patients will often not undergo surgical resection. Metastatic or recurrent malignancy may not undergo repeat biopsy, especially if the tumour is deep seated or difficult to biopsy, hence comparison with the primary may never occur.

In contrast, tissue samples of both primary and metastatic tumours can be obtained at autopsy. Multiple metastases can be sampled, including those in surgically inaccessible sites. The true extent of disease can be determined, including any metastases not detected antemortem. Tissue for controls can easily be obtained, unlike surgical biopsies that target diseased tissue. Research involving rare malignancies also benefits from collection of tumour tissue at autopsy, because larger amounts of tissue can often be obtained and retained for more extensive investigation [10].

DIPG for example, are diagnosed based on clinical and radiologic findings , and biopsy is often not performed at all. The current treatment is ineffectual and they are uniformly lethal. Minimal surgically obtained tissue is available for research, so autopsies can provide material that is critical to understanding the underlying tumour biology [39]. Parents of children with DIPG have been actively encouraging other parents to consider an autopsy, via DIPG cancer support networks. Recent studies using postmortem tissue have finally begun unraveling the molecular alterations present, hopefully allowing more targeted treatments to be identified [40].

With the use of endocrine , chemotherapy , and targeted therapies, the survival for a number of cancer types such as breast cancer has improved dramatically over the last two decades. While survival is good for localised disease, the outcome remains poor once metastases develop [10]. Currently, most metastatic deposits are not biopsied and the treatment of metastatic disease is based on the phenotype (including molecular phenotype) of the primary tumour. There is now compelling evidence that this may be inappropriate. Changes in biomarkers between primary and metastatic sites such as oestrogen receptor (ER), progesterone receptor (PR), and the oncogene HER2 have been demonstrated in breast cancer [16, 41]. In fact, the American Society of Clinical Oncology Clinical Practice Guidelines recommend the use of the ER, PR, and HER2 status of the metastasis to direct therapy, if supported by the clinical scenario and patient’s goals for care [42]. Studies investigating metastatic pancreatic carcinoma have found reduced expression of DPC4 in tumours that are widely disseminated as compared to localised, surgically amenable tumours, suggesting an important role for this tumour suppressor gene [43]. Indeed, whole exome sequencing of metastatic pancreatic ductal adenocarcinoma sampled at autopsy has gone some way towards illuminating the oncogenic drivers of this lethal disease progression [38]. Similarly, studies in prostate carcinoma have begun to document the clonal evolution and molecular changes from primary to metastatic sites. Rapid autopsy derived metastatic deposits from lethal castration-resistant prostate cancer were utilised to describe the discordance in ERG gene rearrangements and ERG protein expression between tumour sites in heavily treated patients [44]; a subset of these samples have been exome sequenced, identifying recurrent mutations in androgen receptor transcriptional cofactors [45].

These approaches will hopefully shed light on the mutations required to metastasise and the “genomic archeology” of multiple metastatic sites [20, 21, 31]. Understanding the biology of metastatic disease will become increasingly important in order to develop targeted therapies [46] understand why treatment fails and find biomarkers of aggressive disease [10].

Evaluate the Effects of Medical and Surgical Therapies

Autopsy research can provide valuable insights onto the effectiveness of both surgical and medical treatment of cancer [17, 47–49]. This is particularly important for new and rapidly evolving areas, such as transplant medicine [47] and stereotactic surgery [50]. Autopsy studies can reveal the effects of treatment on malignancy , providing information on both responders and non-responders and exposing “privileged sites” not reached by systemic therapies [51]. Autopsy allows the most aggressive disease to be sampled, and for samples to be obtained when treatment has failed. The genetic makeup of distant metastases following treatment failure in patients with breast cancer has been shown to be different to that of local lymph node metastases sampled during primary surgical treatment [31]. Toxic effects on adjacent normal tissue and the spectrum of side effects can also be documented [52]. Many survivors, particularly of paediatric and early adulthood malignancies, now live long enough to develop complications from their oncological treatment. Complications such as cirrhosis and bronchiolitis obliterans can be severe and lead to further morbidity and mortality [51]. A thorough understanding of the range of possible complications and their relative incidence is therefore required in determining treatment protocols .

Within drug therapeutic trials, autopsy examination can be used to accurately differentiate between deaths due to treatment (the so-called toxic death), disease progression, and deaths from unrelated causes [53, 54]. In 1997 a survey of clinical research papers published in the British Medical Journal, Lancet, Annals of Internal Medicine, and New England Journal of Medicine indicated that less than a quarter used autopsy to evaluate the cause of death [54]. A review in 2012 of studies conducted within the European Organisation for Research and Treatment of Cancer (EORTC) showed autopsies had been performed in just 26 treatment related deaths, from a total of 255. Of the 26 cases that underwent autopsy, 46 % had a final diagnosis that was discrepant with the clinical diagnosis. The reviewers also felt a further 64 cases which did not undergo autopsy had a clinical course which did not fit with the reported cause of death [53]. The vast majority of deaths were considered not treatment related, and no information is available on these.

Discrepancy between clinical diagnoses and autopsy findings are well documented [14, 15, 47, 48, 53, 55–63], and involve all levels of clinical practice from community hospitals to intensive care units [56]. The rate of major errors , where a principle underlying disease or cause of death is missed is approximately 30 %, and ranges from 5.5 % to over 45 % [53, 62]. Only a few studies show demonstrable improvement in the major error rate over the past decades [46, 64]. Other studies suggest the discrepancy rate has not changed significantly, but the type of unexpected pathology found has [58, 61]. This shift in the conditions that are most likely to be missed is attributed to both changing diagnostic criteria and changing treatments resulting in novel complications. When clinicians are more certain of their diagnoses the discrepancy rate is somewhat lower, but still significant, being 25 % in a large study of 1152 cases [15]. The lowest discrepancy rates (from 5.5 to 7 %) are reported from centres where the autopsy rate remains consistently over 50 % [62, 65].

The diagnosis of neoplastic disease may have a lower discrepancy rate when compared to other disease categories [58, 66], although some cultural factors and mental health disorders may result in under-diagnosis [67, 68]. Misdiagnosis of treatment complications is more problematic; opportunistic infections and cardiac complications are the most frequently missed diagnoses in cancer autopsy series [53, 63]. For example, invasive mycotic infection in patients following stem cell transplant was missed clinically in half of the cases in one study, despite being investigated with cultures, antigen testing, and high-resolution CT scans [69]. In a case described by Allan et al [47] a man with upper gastrointestinal bleeding thought to be secondary to graft versus host disease died despite appropriate therapy. At autopsy he was discovered to have instead succumbed to severe fungal infection. The patient had been investigated with liver and rectosigmoid biopsies, and the clinical diagnoses were compatible with the biopsy results. As this case demonstrates, clinical history and investigations may appear consistent with a particular diagnosis but that doesn’t mean the diagnosis is correct.

Autopsy data should be an essential part of clinical research protocols, particularly in the early stages of patient safety assessment . If autopsy following death during clinical trials is neglected, an under-reporting bias may be present that preferentially favours the death being due to disease, reducing credibility.

In addition, without autopsies the errors that may occur from misplaced clinical bias or suboptimal test performance cannot be documented and learnt from, and unexpected events may not be detected.

Ensuring Accurate Epidemiological Data

Accurate epidemiologic information is required when determining the significance within a given population of specific cancer types, and whether screening or treatment protocols are effective. For cancers that have the potential to remain occult, epidemiological data is not accurate unless it includes a survey of presumed normal subjects. Autopsies of unselected patients provide very accurate epidemiological data as tissue from organs presumed to be normal can be obtained and extensively examined. Prostate carcinoma is one such disease; without autopsy examination an accurate prevalence is unknown. With accurate prevalence data in hand, researchers can better determine the actual effects of prostate screening, and focus their attention on separating the more aggressive carcinomas from those that are indolent [70].

Autopsies also provide comprehensive information about the distribution of metastatic disease, which may be much more widespread than clinical records suggest [40]. A recent review has noted the difficulty in ascertaining the true incidence of brain metastases given the marked reduction in autopsies [46].

While newer imaging techniques may improve detection of metastases, like all diagnostic tests, false positive, and false negative results occasionally occur [71]. Even new, sensitive modalities may not detect disease that is present. For instance, positron emission tomography is one of the most sensitive imaging modalities clinically available and has a lower limit of 10 mm when imaging lung nodules [72]. Over-diagnosis may also occur, with positive scans resulting from active inflammatory nodules or the so-called metabolic flare reaction after chemotherapy [73, 74]. Histology on autopsy samples can be much more sensitive, and may detect tiny residual foci of malignancy, missed by imaging studies [47, 48].

As a definitive and detailed examination of the deceased, autopsies play a vital role in determining the incidence of cancer and proximate cause of death . This will naturally affect population statistics of disease incidence. Given the discrepancies documented in all studies, the reliability of death certificates has been questioned [15, 74]. Accurate population health records are also essential for assessing screening program effectiveness and developing evidence based public health policy [68, 74].

What Makes a Good Research Autopsy Program Work?

Although many types of research can utilise autopsy data, prospective autopsy research programs are the most valuable as fresh and frozen tissue can be retained for molecular studies and future research [11]. Some oncology protocols such as from the Children’s Oncology Group (COG) provide a facility for storage and dissemination of tissue from specific malignancies [12]. Postmortem brain banks have been an integral part of neuropathology research for decades, and increasingly, similar banks for malignancies are being established within academic medical centres [75, 76]. Biobanking of normal tissue is also valuable, and recently the United States National Cancer Institute published recommendations regarding the postmortem recovery of such specimens for research [77].

To be successful, a research autopsy program requires good collaboration between clinicians, pathologists, and researchers in order that sufficient autopsy consents are obtained, and that autopsies are performed to a high standard. Much of the focus in the literature is on how to achieve a higher autopsy rate and the most sensitive method of obtaining consent. Other significant factors that are less frequently examined are the problems of funding and geographic issues .

The consent rate for autopsy research varies greatly between studies from 47 to 98 % [23, 78]. This variance may reflect the difference between obtaining consent to perform research on autopsies that are mandated (Coronial) versus requesting an autopsy specifically for research purposes, with the latter having a much lower consent rate. Some studies suggest that higher autopsy rates can be maintained within specialised programs [47]. Others emphasise the role that a good pathologist–clinician relationship plays [49]. Autopsy request can be part of the end of life discussion when treatment has failed, ideally through the treating oncologist who has already established a relationship with the patient and family [18, 23]. Feedback to the families can also be arranged through the oncological team [23], and this may provide closure and answer any lingering questions [10].

Obtaining consent for the use of tissue for research requires explicit consent in many countries [79]. Although clinicians who have a close relationship with the deceased’s family are often considered to be in the best position to request tissue samples, very high consent rates for obtaining tissue for research (96–98 %) have been obtained by nurse practitioners contacting bereaved families by telephone [7, 78, 80].

A common theme that emerges is that one of the major barriers to obtaining autopsies is the reluctance by medical staff to ask families for consent [12]. It is suggested that the response of families to requests for an autopsy is much more positive than medical professionals assume [78], and that when doctors ask, the autopsy rate increases [9]. A survey of parents of children who had died from cancer found that 93 % indicated they would have agreed to donate tissue for research if asked. Of those same parents, only half had been given the opportunity to do so [12]. Families of research participants are often positive about being given the opportunity to contribute to an area of knowledge that caused suffering for a loved one [7, 23, 78, 80]; in exceptional cases tissue donation has been initiated by parents [28]. Discussing possibly autopsy prior to death can allow for decisions to be made away from the grief of death, although sensitivity is clearly required [81]. Involving patient network and advocacy groups may allow researchers to understand and respond to potential concerns, as well as disseminate information [13].

The reasons for refusal are varied, and include emotional distress, religious and cultural issues, the feeling a loved one has suffered enough, and time pressures [14, 23]. Time pressures are one of the most commonly cited reasons, due to the additional delay imposed by an autopsy [82]; this may be alleviated by a rapid autopsy program . While families may refuse consent, few have indicated dissatisfaction with being asked [78]. Patients may have terminal care at home or in a hospice, so arrangements to transport the body to the mortuary after death will be required [18]. Organisation of the transportation issues and associated costs in advance was found to be helpful [23], and removes an otherwise significant barrier to participation. Some research groups have successfully coordinated external non-academic centres to perform the autopsy and obtain tissue to overcome geographic barriers [28, 32].

Education of both medical staff and families on the value of obtaining tissue at autopsy for research is necessary. Medical staff must be made aware of the presence of research protocols that use autopsy tissue, and the value of tissue donations. Families need better information regarding the potential benefit of donating tissue for research and the process of tumour banking. Education regarding the practical aspects of the autopsy procedure is also important [12]. In addition, tailoring request protocols according to the specific needs of racial and cultural minorities may improve representation of those groups within clinical studies [83].

Conclusion

The autopsy is an essential component of clinical audit as well as cancer research, but remains under-appreciated by many medical researchers. Autopsies can provide large quantities of high quality tissue suitable for modern research methods, as well as providing accurate information on extent of disease, treatment response, and cause of death. Major barriers to obtaining tissue from autopsies for research include a lack of awareness of both current research protocols and the potential value of autopsies, and reluctance to approach family members. However, patients and families are often positive about donating tissue for research, provided consent requests are carefully considered. Time pressures and transportation costs are some of the potential barriers that can be ameliorated [84]. Funding may also be an obstacle, as the infrastructure required is expensive and the benefits may take some years to be realised.