Skip to main content
Premium Trial:

Request an Annual Quote

Cancer Centers Investing in Research Autopsies to Explore Why Patients Fail Precision Medicine


NEW YORK (Precision Oncology News) – Over the course of a year, Carol Gibbs watched her husband of 31 years succumb to a cancer so rare that there are only around 100 cases described in the scientific literature. A tumor was initially detected in his neck in the summer of 2016, but over the course of that year, the cancer cells had deftly mutated, spread throughout his body, even to his brain, and evaded all attempts at treatment. 

She prayed for a miracle but knew things were changing and that she had some tough decisions to make. As she began making the arrangements and preparing for the worst, she realized she wanted to do something to honor her husband's memory. 

Jeffrey Gibbs was a dentist in his mid-50s in Columbus, Ohio. His wife described how he loved the outdoors, worked out every day, and was a "researcher and scientist at heart." She recalled how when he got his genetic test results from Foundation Medicine, reporting that his cancer had more than 100 mutations per million nucleotides — what researchers described as "ultra-hypermutated" — he pored over the 150-page report, fascinated by the details of all the genetic alterations. He embraced his diagnosis with a deep religiosity and a wicked sense of humor. When his colleagues gave him a lifetime achievement award, in his acceptance speech he said his purpose in life was to serve God and help others as much as he could. Privately, he quipped to his wife, "Of course, they give the award to the guy dying of cancer."

When Carol Gibbs learned about the rapid autopsy program at Ohio State University's Comprehensive Cancer Center, where her husband was being treated, she and her three sons agreed this was what he would have wanted. Although he was too sick to decide for himself, they were sure he would have wanted to go in service to science, so that researchers could learn more about his disease: interdigitating dendritic cell sarcoma (IDCS), a poorly understood cancer that lacks standard treatment. 

The family's decision allowed researchers, led by OSUCCC's Sameek Roychowdhury, access to precious blood and tissue samples that they analyzed via whole exome sequencing. Using the molecular clues, they mapped the evolution of Jeffrey Gibbs' cancer cells as they multiplied, spread, and avoided attack from treatments. In publishing their findings, Roychowdhury and his colleagues wrote that this may be the first time IDCS had been so extensively sequenced and analyzed. 

Top cancer centers around the country are starting research autopsy programs like the one at OSU in an effort to overcome the blind spots and limitations of precision oncology approaches. Currently, when patients have their tumors genomically profiled as Jeffrey Gibbs did, it is done from tissue samples from the primary tumors, and patients are often not re-biopsied when their cancers progress. This may be due to cost, or because patients may not be able to endure invasive procedures as they get sicker.

Eric Haura, director of the Lung Cancer Center of Excellence at Moffitt Cancer Center, illustrated the problem in lung cancer, where doctors usually start considering precision oncology approaches for patients only after they have developed late-stage or advanced disease. But genomic testing to identify potentially beneficial treatments is done on tissue samples from surgical resection of early-stage tumors with very different molecular characteristics. The molecular mechanisms driving cancer cells in a lung tumor may shift and use different pathways than those targeted by treatments as they spread to the brain or liver. "We're just not studying the right cancers," said Haura. 

Even when advanced cancer patients can endure biopsies, it's likely from a single site, which would miss capturing other molecular mechanisms driving individual cancer cells in the tumor. "When we perform a needle biopsy in a patient, you’re accessing one metastatic lesion, in one area, where the radiologist can safely put the needle, but there are multiple metastatic sites," said Haura.

Currently, "researchers are sequencing a mash of tumor cells," agreed Roychowdhury, an associate professor and co-director the rapid autopsy program at OSUCCC and The James Cancer Hospital. "In order to really understand [intra-tumor heterogeneity] we have to study multiple tumor samples over time, utilize research autopsy, incorporate single-cell sequencing, and develop new computational approaches to describe tumor cells one at a time and map their evolution." 

However, as the term "rapid autopsy" suggests, time is of the essence if researchers want to use the collected biological samples for molecular analysis, such as single-cell sequencing. And speed requires investment in expertise and infrastructure: a multi-disciplinary team of genetics experts, oncologists, and pathologists, on-call autopsy technicians to collect the samples, and a system of transport for the deceased to be quickly taken to a facility for the autopsy and returned at a prearranged location for funeral services.

Importantly, research autopsies require the consent of the patient or, in some cases, the family. Oncologists must evaluate if a patient is ready to consider the possibility that treatments might fail and find the right way to ask if they want to continue to contribute to research in the event of their death. 

"These are sensitive discussions and they have to be done very carefully," said Dejan Juric, who co-developed the rapid autopsy program at Massachusetts General Hospital and directs the Termeer Center for Targeted Therapies & Investigational Cancer Therapeutics Program. "Our physicians have to show a lot of empathy for patients' fears and hopes and carefully balance these questions."  

Oncologists are often afraid to broach this topic, scared it might make patients lose hope, but physician scientists involved in these programs have been surprised by the altruism of patients and their families. "It's the providers' fear that prevents them from asking patients," said Juric. "More often than not, patients and their families are almost relieved that something else can be done, and despite the ultimate loss, it can help someone else."

Advancing precision oncology

The US Food and Drug Administration in May approved the first PI3K inhibitor alpelisib (Novartis' Piqray) for advanced breast cancer patients with PIK3CA mutations. But years before, using samples from a breast cancer patient who had given permission for her tumor tissues to be used for research when she passed, researchers were already tracking how cancer cells driven by these alterations survive and grow, and they had begun strategizing ways to stop them. 

A team led by MGH's Juric described in a 2015 Nature paper the case of a 60-year-old metastatic breast cancer patient who had received alpelisib as part of a Phase I study, experienced tumor shrinkage for more than nine months, but then, the cancer that had spread to her lungs started to grow again. Two months after stopping treatment, she died. 

However, because she had consented to partake in MGH’s tissue donation and rapid autopsy program, researchers had access to tumor tissue from before she received treatment, as well as from more than a dozen metastatic sites upon her death. They sequenced these tumors,  mapped the molecular characteristics, and found that all the metastatic tumors that eventually became resistant to alpelisib did so through loss of PTEN expression, though via different genomic events. One tumor site that was still responding to the drug expressed the protein. 

Researchers then created a model of the patient’s lung tumors that had lost PTEN expression by growing these tissues in mice and demonstrated that they, too, were resistant to alpelisib. However, these mice were responsive to buparlisib, a pan-PI3K inhibitor that targets all the PI3K isoforms, as well as to the combination of an alpha- and beta-PI3K inhibitor. 

"As we were conducting the Phase I trial of alpelisib, we already started looking at acquired resistance so we can understand how subsequent trials should be designed," said Juric, who started MGH's rapid autopsy program three years ago with pathologist James Stone, and who also co-lead the Phase III SOLAR-1 study that resulted in alpelisib's approval. 

With the knowledge garnered through the autopsy of this breast cancer patient "we were able to think about ways to address PTEN null tumors with other targeted therapies," he said, noting that AKT inhibitors, such as Genentech’s investigational drug ipatasertib, are emerging as an option.

This case also underscores why the word "rapid" is so important in the term "rapid autopsy." Tissue degrades quickly after death, limiting certain types of analysis, such as phosphoproteomics, and growing tumor tissue in mice becomes harder. For this reason, MGH aims to do these autopsies within a six-hour window after death, and has even performed an autopsy as fast as 30 minutes, allowing for phosphoproteomics analysis.   

In the case of the breast cancer patient, the autopsy was done three hours after she passed, allowing researchers to perform in-depth sequencing and build mouse models. The models were critical to demonstrating that because alpelisib inhibited only the PI3-alpha kinase, the patient's metastatic tumors were able to evade death and progress using a variety of other genomic alterations. They hypothesized that the tumors could have been thwarted by a drug that shut down both alpha and beta kinases. 

"You can profile the tumor using next-generation sequencing or other approaches, but what you end up with is a list of putative resistance factors," said Juric. But with a model of the tumor, if there are three resistance drivers, then one can test if targeting one or all three can reverse resistance, he explained. 

Tumor models within rapid autopsy programs are also useful for preclinical testing of combination therapeutic strategies that go after different resistance pathways. The first precision oncology drugs to come to the market have been targeted drugs that are combined with chemotherapy or given as single agents. But, because tumors can easily become resistant to a single targeted drug via different genetic mutations, researchers are now exploring combinations of targeted therapies or regimens containing targeted drugs and immunotherapies, betting that it'll be harder for tumors to develop mutations that cause resistance to multiple drugs.   

One drawback of rapid autopsy programs, however, is that they often focus on studying the most deadly tumors late in their course, when patients have undergone multiple rounds of treatment, making it difficult to figure out which molecular changes emerged in response to which therapeutic pressures. To address this, MGH researchers are collecting tissues more systematically, giving patients the chance to donate tissue and plasma at diagnosis. A subset of patients will be followed longitudinally, and if they consent will have autopsies. "By collecting samples at different time frames, when patients are alive and upon death, you have a powerful set of samples [for analysis, and] you can combine the data and find patterns that you can't find otherwise," said Juric. 

When precision oncology fails

There are plenty of success stories in precision oncology. For example, in certain types of breast cancer, lung cancer, melanoma, and leukemia, researchers have uncovered markers that can be targeted by treatments, which have dramatically improved the outlook for patients. But even when precision medicines work, they only work for a time, and cancer cells almost always find a way around them.

Many, like Jeffrey Gibbs, don't respond at all. "He had some of the highest tumor mutational burden we'd ever seen," Roychowdhury said. "He never responded to anything we tried."

When genomic profiling through Foundation Medicine revealed that Jeffrey Gibbs' tumor was ultra-hypermutated, for example, his care team was cautiously hopeful that treatment with the immunotherapy nivolumab (Opdivo) might work. Some cancer patients with a high tumor mutational burden respond well to immunotherapies, and TMB is now one of the most widely studied biomarkers in precision oncology research. 

Jeffrey Gibbs wasn't one of the lucky ones. After getting nivolumab, he continued to progress, and by January 2017 his scans revealed metastasis. Another bout of immunotherapy and chemotherapy only made him weaker. By the summer, he lost the use of his right arm and experienced a grand mal seizure. The cancer had spread to his brain. 

"I saw my husband slowly losing ground and began thinking that even if he survives this, his quality of life would never be the same," Carol Gibbs recalled. "I started thinking about how everything was going to be different and so hard for him." 

During all this, Gibbs hadn't talked about funeral arrangements. There wasn't even a burial plot. "[Jeff] wasn't fearful, as he knew where he was going, but in the next breath he would say, 'But I don't want to leave you guys,'" she recalled. "I didn't want to bring up the fact that we had really hard decisions to make, because he was still fighting this disease with everything he had and still wanted to be with us." 

By the fall, his health had declined so much that Carol Gibbs called her three sons, who were working and studying in far flung places and told them it was time to come home. During those last weeks with the family together, Jeffrey Gibbs got a chance to tell his sons his hopes for them for the future. Carol Gibbs found quiet moments with her husband to help him write the letters he wanted to leave their sons. 

In the process of considering the arrangements she wanted to make for her husband, a friend had mentioned OSU's rapid autopsy program and that Roychowdhury, who was already part of the care team, was involved. Now, with all the family members at the same place, Roychowdhury came to the Gibbs' home on a crisp fall day to explain what rapid autopsies entailed. Carol Gibbs remembers it as the kind of day the family would normally have spent outdoors, hiking or fishing, "but there was nowhere else I'd rather have been." 

Roychowdhury explained that the family would have to notify OSU soon after Jeffrey Gibbs died so his body could be transported back to the university for the autopsy. Amid all the science, there was a lot of compassion in how Roychowdhury discussed the program, Carol Gibbs said. "He didn't minimize what a huge decision this was for us." 

The family took the night to think about it. The next morning, they all agreed that this was the right decision. "Unanimously, we felt strongly that Jeff was such a scientist that he would have loved to be one of those researchers studying this," Carol Gibbs said. 

Jeffrey Gibbs died early the next morning. Within an hour of contacting OSU, his body was transported to the autopsy center, and by mid-afternoon the research team had taken all the necessary samples and his body was taken to the funeral home where services would be held. "They really honored his body," Carol Gibbs said. 

Roychowdhury and his team eventually published this case in Oncotarget last year and described what they learned. They performed whole-exome sequencing on nine metastatic tumor sites and the pre-treatment primary tumor, and conducted a variety of biomarker analysis, including calculation of TMB and microsatellite instability. They categorized mutations according to if they occurred in all tumors, in certain tumors, or were unique to one tumor, and identified three distinct mutational signatures. Using bioinformatics approaches, researchers further differentiated six clonal populations of cancer cells in each tumor, differentiated by 10 unique groups of genomic alterations. 

Ultimately, OSU researchers were unable to pinpoint exactly why Jeffrey Gibbs failed to respond to immunotherapy or other drugs. However, the immense genomic heterogeneity catalogued in his tumors highlighted the insufficiency of current biomarker approaches in precision oncology, which rely on genomic profiling of tissue from a single biopsy specimen, and the need for more research on how tumor heterogeneity impacts patient care.

"Doing an autopsy doesn't solve this question of tumor heterogeneity and evolution right way, but it gives us a path forward and hope," Roychowdhury said. 

Funding challenge

Before the 1970s, an autopsy was done on between 40 percent and 60 percent of patients who died in a hospital, but since then the rate of autopsies has declined sharply. The US Centers for Disease Control and Prevention estimated that in 1972, one out of five deaths were autopsied, and around 80 percent of them were done for clinical purposes to learn more about people who died from disease.  

In the latter part of the 20th century, as medical professionals began using new imaging tools, autopsies came to be thought of as old-fashioned procedures that were costing the hospital money without bringing in profits. By 2007, the CDC estimated that hospitals were performing clinical autopsies for around 4 percent of patients. 

"Autopsies used to be the only way to understand what happened to the patient," said Juric. "With the emergence of MRIs, PET and CT scans, we think we have a much better understanding of the extent of disease." 

At MGH, before the initiation of its rapid autopsy program in 2015, Juric estimated that between 7 percent and 9 percent of cancer patients had autopsies performed on them. Now, out of a need to understand tumor heterogeneity in the era of precision oncology, some cancer centers are turning to autopsies once again. 

But, because time is of the essence as pathologists and autopsy technicians race to collect and preserve tumor tissue for detailed molecular analyses, these programs require investment in infrastructure and expertise. It took Moffitt several years to figure out the logistics and set up its research autopsy program. For the time being, research autopsies are done only on lung cancer patients, and the program has so far done 15 post-mortem procedures. "It’s expensive to run this program," Haura admitted. 

When patients who consent to an autopsy die in the hospital it can be done fairly quickly, but many of Moffitt's lung cancer patients choose hospice for their last days. For those cases, Moffitt employs a medical examiner who goes to the funeral home to collect the samples, and transports them to the cancer center for storage inside a collection kit that preserves the tissues. Using this process, the program has been able to collect samples from patients within a hundred-mile radius of Tampa within 24 to 48 hours after their death.  

Although that time frame is longer than what other research autopsy programs are aiming for, Moffitt has preliminary evidence demonstrating the feasibility of its program. Researchers have shown that the program is collecting quality samples using immunohistochemistry and DNA and RNA sequencing. With IHC analysis, for example, researchers have been able to detect T cell and immune markers in lung cancer. In a patient who failed a third-generation ALK inhibitor, researchers identified a possible resistance mutation and suggested a treatment that may overcome it. 

Haura and his colleagues are in the process of publishing these early learnings, but he noted that the program must expand to really understand complex disease mechanisms, examine relapse following targeted agents, study patients who receive more complex immune therapies, such as CAR T cell therapies, and tackle other big unanswered questions in oncology today like why outcomes can differ so dramatically following immunotherapy. "It's taken a while to collect samples from 15 patients," he said. "We just don’t have enough numbers."

Haura would also like to do single-cell analysis and incorporate the use of organoids and mouse models, but then samples would have to be collected much faster and Moffitt's program logistics would have to change, requiring funding that has been difficult to procure.

"It can be very difficult to find sources of funding for infrastructure building," Haura said. "The challenge for us is a long-term sustainability plan." The program received a National Cancer Institute grant to study the ethical framework for the project, but since then it has largely relied on philanthropic donations. 

Haura is now exploring industry partnerships, in which drugmakers interested in identifying new targets would support Moffitt's autopsy program to learn from patients who failed therapies. But within these partnerships, he would push for the findings to be public. "These tissues are donated by patients and their families," he said. "This is a way for people to give back, [and] that spirit needs to be maintained in any partnerships we put together. We need to share the data." 

The hardest part

More than the funding challenges, for doctors, raising the topic with patients and loved ones is perhaps the hardest part of these research autopsy programs. At Moffitt, oncologists are trained on how to discuss the program and when to approach the subject. When the patient is progressing or at the end of life, it may not be the right time, Haura said, who despite 20 years of experience as a medical oncologist still finds it daunting. "It requires a deeper relationship between the oncologist and the patient," he said. 

Not all patients are right for research autopsies, and doctors must pick up on cues as to how patients understand their disease and the expectations they have for treatment. One patient Haura approached about the research autopsy program declined after considering it. "She felt like if she signed the consent, she would be acknowledging her imminent mortality," Haura recalled. "We still have a great relationship and she wasn't angry." 

On the other hand, Haura has approached patients who had already considered and looked into donating their bodies to medical schools or for research. After physicians get over their initial fear of discussing this topic, they are often surprised by patients' and families' openness to it. 

Since Roychowdhury started the rapid autopsy program at OSU in 2016, researchers have autopsied 46 patients with a variety of cancers. When the program was just starting, Hui-Zi Chen, an assistant professor at OSUCCC who co-directs the autopsy program with Roychowdhury, had her doubts. "Initially, I was thinking, this is pretty morbid. It's not what I signed up for," said Chen. "But, the more I started doing it, the more I realized how rewarding it was."

The Gibbs family's decision to participate in the autopsy program personalized their connection to OSU's research and deepened the relationship with the care team. Jeffrey Gibbs' middle son, who is pursuing a PhD and a medical degree, is interested in oncology, and Roychowdhury has offered guidance. Carol Gibbs wants to talk to families about the importance of rapid autopsy programs in the hopes they will consider the option. She would tell them, "It gives purpose to the pain.