At the University of South Alabama, University of Alabama at Birmingham and HudsonAlpha Institute for Biotechnology researchers are looking for clues in the human body and even utilizing artificial intelligence in studies to advance early disease detection and more effective treatments for maladies such as cancer and Parkinson’s disease.
Their research could one day lead to new medications, better ways to detect disease and even enable physicians to predict with greater accuracy which treatments will benefit each patient the most.
Early Detection of Pancreatic Cancer
Pancreatic cancer is one of the most aggressive and deadliest forms of cancer because early detection is difficult, and only 1% of patients in stage 4 of the disease live five years.
But researchers at the University of South Alabama’s Frederick P. Whiddon College of Medicine and the USA Health Mitchell Cancer Institute made a discovery recently that could one day lead to new treatments for pancreatic cancer and methods of diagnosing it early.
“The number of estimated cases that will be diagnosed in this country this year is around 66,440 patients, and, unfortunately, 51,000 will die within one year,” says Santanu Dasgupta, Ph.D., a USA pathology faculty member and principal investigator for the study.
He says pancreatic cancer often spreads to other organs before doctors can make a diagnosis.
“But if you could identify the tumor in some way, like in a blood test, while the tumor is within the pancreas and it doesn’t spread, then you can cut it out and you can kill the cancer,” Dasgupta says.
So, he and the other researchers looked to the body’s cell mitochondria. Mitochondria are organelles inside of cells, and they contain some 1,500 proteins that they use to regulate cellular metabolism, stress and even cell death.
Unfortunately, cancer cells wisely use the proteins in mitochondria, too, which makes the cancer cells more aggressive, Dasgupta says.
Most of the 1,500 proteins travel from cytoplasm into mitochondria through a protein import system, and this system works through proteins called translocases of the outer mitochondrial membrane or TOMM.
Dasgupta says he discovered that the central outer mitochondrial membrane translocase, or TOMM22, is expressed in high levels in pancreatic cancer patients.
“So, I identified that TOMM22 is a protein that is present outside the mitochondrial membrane, which acts as a receptor, like a door, like an open road, through which all the food that can help cancer cells to grow comes in,” he says.
With that knowledge, the researchers engineered pancreatic cancer cells to express high levels of TOMM22 and found that the cancer cells became aggressive. But when they engineered cancer cells to block TOMM22, the cells became significantly less aggressive, he says.
“That TOMM22 could be used as a marker in blood tests to detect cancer early,” says Dasgupta, “and, we can also make a drug targeting this door because if I have a drug that can block this door, then pancreatic cancer cells cannot move to other parts of the body and cannot kill the body.”
“That’s the whole idea.”
The researchers published their findings in Molecular Cancer Research.
Predicting Breast Cancer Responses to Treatment
At the University of Alabama at Birmingham, researchers are using machine learning, a branch of artificial intelligence, to predict at the time of diagnosis how well a patient with triple-negative breast cancer will respond to chemotherapy.
“TNBC is aggressive; it spreads fast,” says Ritu Aneja, Ph.D., associate dean for research and innovation at the UAB School of Health Professions.
“Treatment options for TNBC are limited, with chemotherapy being an integral part of the treatment regimen,” says Aneja, “but studies show that only 30-40% of tumors shrink in response to chemotherapy.”
The remaining 60-70% don’t respond, and patients who undergo chemotherapy often suffer from unpleasant and distressing side effects such as nausea, vomiting and hair loss.
“Newer immunotherapies have improved outcomes by about 10-15%, yet chemotherapy remains the mainstay of TNBC treatment,” she says.
To understand why some triple-negative breast cancer tumors respond well to chemotherapy and others respond partially or not at all, scientists at UAB, Georgia State University and the University of Galway in Ireland, used machine learning to examine the microenvironments of tumor samples before chemotherapy.
A tumor’s microenvironment is the ecosystem surrounding a tumor that consists of both immune cells and tumor cells, as well as blood vessels, stroma and other cells that nurture tumor cells. Microenvironments are known for influencing how well a tumor responds to chemotherapy.
Machine learning involves creating complex algorithms that can be trained on preexisting data and can also spot new or unique perception data patterns that may be missed by the human eye, she says.
Aneja and her team used machine learning to develop an algorithm or predictive model that could recognize specific features and patterns in the breast tumor microenvironments and tell the difference between responders and non-responders.
The results of the study showed 85% accuracy, with the algorithm correctly predicting that 42 out of 51 cases would respond positively to chemotherapy and 29 out of 34 cases would be unresponsive to the treatment.
“The future is bright,” says Aneja. “If one is diagnosed with TNBC, the prediction algorithm will be able to tell whether the person will benefit from chemotherapy and if the cancer will be aggressive or metastasize later, just by examining their biopsy sample.”
Scientists look for easier Parkinson’s diagnostic method
At the HudsonAlpha Institute for Biotechnology in Huntsville, researchers are on the hunt for a way to detect Parkinson’s disease early and monitor its progress.
Parkinson’s disease is a neurological disorder that causes stiffness, intense tremors in the hands and legs, a shuffling gait and other debilitating symptoms. There is no cure, and the symptoms can grow worse over time.
Moreover, no test exists to determine if someone has the disease. Consequently, to make a diagnosis, doctors must conduct physical and neurological exams, evaluate patients’ symptoms, and see how patients respond to certain medications before making a diagnosis.
So, scientists at HudsonAlpha are launching a study that will involve taking blood samples to look for biological molecules called biomarkers that could signal when Parkinson’s disease is present.
“Could we be better at detecting it earlier? The earlier you detect something, the better chances you have for developing effective treatments,” says Richard Myers, Ph.D., HudsonAlpha’s chief scientific officer and president emeritus.
In addition, biomarkers may provide new information about the biology of Parkinson’s that could be helpful for researchers in the future, Myers says.
Benjamin Henderson, Ph.D., a senior scientist at HudsonAlpha, says their researchers will examine hundreds of blood samples from people with Parkinson’s disease and other control subjects — people without the disease — in hopes of identifying biomarker signatures.
“The main goal of this is to enroll hundreds of controls and hundreds of patients because that’s going to give us the ability to perform sequencing assays and statistically find things that are relevant,” Henderson says.
The researchers will collect blood samples at the Smith Family Clinic for Genomic Medicine on the HudsonAlpha campus. All participants must be age 19 or older, and they must be either Parkinson’s patients; unaffected, biological relatives of Parkinson’s patients; or people who live with Parkinson’s disease patients.
“We’re going to get basic medical and family history information and blood samples. Then we’ll have them to do a sniff test, which is a measurement of their ability to know and distinguish different scents,” says Meagan Cochran, director of clinical education for HudsonAlpha and director of the Smith Family Clinic for Genomic Medicine.
“That’s because it turns out people with Parkinson’s disease often lose that ability. That’s a good example of a known biomarker for Parkinson’s disease.”
To participate, email [email protected] or call 256-327-9640.
Gail Allyn Short is a Birmingham-based freelance contributor to Business Alabama.
This article appears in the July 2024 issue of Business Alabama.
