High-res imaging with elastography may accurately detect breast cancer in surgical margins

16 Apr 2020
High-res imaging with elastography may accurately detect breast cancer in surgical margins

A high-resolution, three-dimensional imaging technique, when combined with quantitative measurement of tissue elasticity, could accurately detect cancer within the resected margins of surgical specimens taken from patients undergoing breast-conserving surgery.

The findings from this study have been published in the journal Cancer Research

"Despite living in the 'digital age,' surgeons must routinely rely on their eyesight and sense of touch to determine if they have removed the entire tumour during breast-conserving surgery," said Kennedy. "Due to lack of adequate tools, 20 to 30 percent of patients must return for additional surgery, resulting in substantial physical and financial burdens and increased risk of complications."

Beyond surgeons' native senses of sight and touch, X-rays are often used for the intraoperative detection of tumour in the margin of surgical specimens, said Saunders. "While useful in some cases, this method can't detect small microscopic traces of tumour that surgeons often miss," she said. "As a result, it is widely accepted that higher-resolution intraoperative detection techniques are needed."

Optical coherence tomography (OCT) is a type of imaging technique that generates three-dimensional images of tissue.

"OCT can be described as the optical equivalent of ultrasound, using reflections of light waves rather than sound waves to form images of tissue microstructure," Kennedy explained.

The images generated by OCT can be used to visualise and detect cancer in mastectomy tissues, but the sensitivity and specificity of this technique in breast-conserving surgery specimens in several recent studies was relatively low.

"Cancer and its associated stroma are stiffer than benign tissues," Kennedy explained.

OCT technology can also be used to measure tissue deformation under an applied force, allowing for clinicians to ascertain the elasticity, or stiffness, of the surgical specimen.

Quantitative micro-elastography (QME) is a variant of OCT that can generate three-dimensional maps of local elasticity.

"We wanted to determine the diagnostic accuracy of QME for detecting tumour in the margins of breast-conserving surgical specimens, and compare this technique with images generated by OCT alone," Kennedy said.

In this study by Kennedy, Saunders, and colleagues, 90 patients were recruited who were undergoing surgical treatment for breast cancer.

Of these patients, 83 received breast-conserving surgery, and seven received a mastectomy.

Following surgery, simultaneous OCT and QME analyses were conducted on the resection margins of the freshly excised specimens.

After imaging, the surgical specimens were submitted for standard histopathological processing.

Histology sections were co-registered with the OCT and QME images to determine the types of tissue present in each scan.

To facilitate the comparison of OCT and QME images with histopathology, three-dimensional regions of interest (ROI) were selected from the scans.

At least one ROI was selected for every surgical margin used in the study.

Each ROI was determined to be positive for cancer if the pathologist identified tumour within 1 millimetre (mm) of the margin of the corresponding histology section.

To generate a set of pilot data, the researchers used all mastectomy surgical samples and surgical samples from 12 patients who received breast-conserving surgery.

These data were used to train seven readers (two surgeons, two engineers, one medical sonographer, one pathology assistant, and one medical resident) to determine if images generated by OCT and QME indicated the presence of cancer in the surgical margins.

Surgical samples from the remaining 71 patients who received breast-conserving surgery were used to determine the ability of OCT and QME to detect cancer within 1mm of the surgical margins, as compared with gold-standard post-operative histology.

Readers were blinded to the histopathology results.

Sensitivity, specificity, and accuracy of each method was calculated for each reader, and aggregate results for all of the readers were determined.

"While inter-reader agreement was nearly perfect for QME, agreement was only moderate for OCT," Kennedy noted.

Based on the aggregate results, OCT images resulted in a 69.0 percent sensitivity, 79.0 percent specificity, and 77.5 percent accuracy for detecting cancer within 1mm of the surgical margin.

QME images resulted in a 92.9 percent sensitivity, 96.4 percent specificity, and 95.8 percent accuracy for detecting cancer within 1mm of the surgical margin.

"Imaging the microscale stiffness of tissue using QME has the potential to reduce re-excision rates in breast-conserving surgery," Kennedy said. "Further, by quantifying tissue stiffness, we remove the subjectivity that is inherent to the surgeon's sense of touch.

"The ideal scenario would be to perform the imaging in the surgical cavity immediately after the specimen has been removed," Kennedy continued. "This would give surgeons a direct indication of whether any tumour had been missed. As such, our next goal is to develop a handheld QME probe to enable intraoperative imaging."

Roughly 12 percent of the selected ROIs were excluded from the study for a variety of reasons, including extensive thermal damage, imaging artifacts, or the presence of a rare form of mucinous ductal carcinoma in situ (DCIS), representing a limitation of the study.

"We are working on ways to accommodate for each of these excluded ROIs to ensure that the technique is as robust as possible," Kennedy said.

Source: American Association for Cancer Research