Ohio State researchers developing AI model to detect cancer early

The Ohio State University’s James Cancer Hospital makes advancements on lobular breast cancer screening

While researchers work to improve lobular breast cancer screening, innovative treatments are improving outcomes for patients with the condition, which is diagnosed in more than 40,000 Americans each year.

Lobular breast cancer — or invasive lobular carcinoma (ILC) — is a form of breast cancer that starts in patients’ lobules, or milk-producing glands. ILC may not be a common term among the larger public, but it makes up approximately 10-15% of all breast cancers diagnosed in the U.S.

“There are different types of breast cancer — invasive ductal carcinoma, the most common type, which begins in the ducts that carry milk to the nipple, and invasive lobular carcinoma, which starts in the lobules that produce milk,” said Arya Roy, MD, a medical oncologist at The Ohio State University Comprehensive Cancer Center – James Cancer Hospital and Solove Research Institute (OSUCCC – James).

While ILC is usually cured when caught early, a lack of widely-known symptoms, along with screening challenges and higher rates of late recurrence, make it vital that researchers continue their efforts to improve diagnostic techniques and technologies.

“We can give hope to ILC patients by developing more tools to identify the cancer as much as possible so that we can make sure it’s completely cured,” Roy said.

Invasive lobular carcinoma diagnosis challenges

Unlike some other types of breast cancer that form in masses, ILC cells tend to grow in single-file lines.

“Usually when people think about breast cancer, they think about lumps, but in most lobular breast cancer cases, there won’t be one,” Roy said. “ILC does not have a protein (called E-cadherin) in the cells which makes them stick together, so the cancer actually forms in lines or chains.”

While lumps usually aren’t present with ILC, there are some potential symptoms, including:

• Swelling of the breast
• Thickening of breast skin
• Changes in the nipple, including inversion

Because of the way ILC develops, mammograms may not detect it in its early stages, making it even more important that patients report any changes in their breasts.

When ILC is diagnosed, doctors use MRI and other tests to determine size and stage so that unique treatment plans can be created for each patient.

“ILC is almost always estrogen receptor–positive, so we use an FES PET scan to identify the spread of the cancer,” Roy said. “This newer imaging technology helps detect cancers that have high levels of estrogen receptor expression.”

While doctors and researchers are working to improve ILC screening, identifying possible risk factors could raise the chances of early diagnosis, including:

• Family history of ILC or stomach cancer
• LCIS lesion in the breast
• CDH1 genetic mutation
• Dense breasts

Lobular breast cancer treatment and research

Though ILC tumors are typically larger than other types of breast cancer at the time of diagnosis, they also tend to grow more slowly, which can raise the chances of successful treatment.

Doctors sometimes administer chemotherapy before surgery to try to reduce the size of ILC tumors before removal, but the presence of estrogen can reduce its effectiveness.

“ILC usually does not respond to chemotherapy very well because the cancer is highly hormone positive,” Roy said.

However, patients with estrogen-positive ILC may benefit from endocrine therapy administered before surgery. Multiple clinical trials are currently underway at Ohio State involving this treatment, which directly targets estrogen to shrink tumors or stop their growth.

While therapy and surgery often lead to successful initial treatment of lobular breast tumors, ILC patients can be at higher risk of recurrence than invasive ductal survivors.

“We don’t currently have a specific tool to identify lobular breast cancer recurrence risk,” Roy said. “We are using techniques for invasive ductal breast cancer, but that’s not always accurate.”

Roy and her colleagues are working to fill that gap through clinical research that uses artificial intelligence to improve ILC recurrence risk assessment.

“We’re developing an AI model that analyzes higher risk features obtained from digital pathology images of removed tumors, and combining those with the patients’ clinical features,” Roy said. “Then, an algorithm will help us predict the patients with higher risk of ILC recurrence.”

Roy’s team is currently using patient information to train the AI model with the goal of developing a scoring system that identifies patients at high risk for recurrence. The training process is lengthy, but the benefits could be well worth the effort.

“We can keep the patients in long-term surveillance to diagnose recurrences early, when the cancers are at their most treatable stages,” she said.

SOURCE: The Columbus Dispatch