Optical tomography/US monitors preop chemo response

SAN ANTONIO-Optical tomography with ultrasound localization has the potential to monitor tumor vascular changes during neoadjuvant chemotherapy, according to a pilot study in which the modality was able to distinguish between responders and nonresponders, and even between complete and partial pathologic responses. Susan Tannenbaum, MD, of the University of Connecticut, Farmington, described the approach at the 2007 San Antonio Breast Cancer Symposium (abstract 45).

Optical tomography uses diffused light in the near-infrared (NIR) spectrum for imaging tissue. As light passes through the body, a quantifiable portion of it is absorbed by blood vessels. Tumors have higher microvessel density or higher blood volumes than nonmalignant tissues, and total hemoglobin (tHb) concentration, as measured by NIR, is directly related to tumor vascular content. Prior studies by the University of Connecticut and Hartford Hospital researchers and physicians have shown that tHb concentration is much higher in cancerous than in nonmalignant breast masses.

“Due to the intense light scatter that occurs when NIR travels through tissue, three-dimensional image localization is poor and quantification of light is imprecise,” Dr. Tannenbaum said. To optimize light localization and quantification, co-author Quing Zhu, PhD, professor of bioengineering, and colleagues coupled NIR with ultrasound (US) to produce an NIR-US imaging modality that simultaneously acquires US images and optical data via a handheld probe

“This volumetric data is then reconstructed tomographically in 5 mm cuts from skin down to chest wall,” Dr. Tannenbaum said. Each slice shows the volume and distribution of tHb concentration (the tHb map).

Patients with stable disease tend to show virtually no difference in tHb maps before and after treatment, she said. Complete pathological responders show marked reductions in tHb maps after treatment. And partial responders demonstrate some degree of volume loss and reduction in tHb concentration, “although it is less pronounced than in the complete responders,” she explained.

The investigators then quantitated these changes in tHb during treatment for each patient in the study to determine whether these absolute changes might reflect pathologic response groups.

To quantify tHb content, they calculated the blood volume index (BVI) (total volume of tHb within the tHb map × average tHb). Then, to measure vascular changes during treatment, they used the percent BVI (PBVI) at any cycle, compared to the pretreatment BVI.

Changes in PVBI

If at the end of therapy, PVBI showed little change, this value would be 100%; if it was markedly reduced, it would approach 0%. Changes in PBVI were then evaluated for their potential to monitor tumor vascular changes during chemotherapy and predict pathological response.

The study reported at SABCS involved 11 breast cancer patients undergoing neoadjuvant chemotherapy. The women were monitored with the NIR-US imager after every two treatment cycles until surgery. After surgery, based on the pathological assessment of the specimen, the patients were classified into three response groups: complete or near-complete responders, partial responders, and nonresponders.

The results showed that PBVI correlated with pathological response. At the end of treatment, nonresponders showed little change in PBVI from baseline (86.8%), whereas PBVI was significantly reduced from baseline in partial responders (46.3%) and in complete or near-complete responders (29.1%).

Complete responders had lower PBVI than partial responders (P < .02) and partial responders’ PBVIs were lower than those of nonresponders (P < .04).

“Even at the end of the second treatment cycle, the PBVI of the complete or near-complete responders was noticeably lower than that of the partial responders (P = .091) and nonresponders (P = .075),” added co-author Dr. Zhu.

The investigators concluded that NIR-US using PBVI to monitor vascular changes accurately predicted complete responders and nonresponders in this pilot study, and may offer the potential for early recognition of treatment efficacy and tailoring of the intervention to the response.

“Our future studies will include further technology development and increased cohort size,” Dr. Tannenbaum said.

VantagePoint Monitoring tumor response: The more information, the better

Debu Tripathy, MD - The point of monitoring patients receiving neoadjuvant therapy is to separate responders from nonresponders early-on, before a full course of neoadjuvant chemotherapy or endocrine therapy is given, said Dr. Tripathy, clinical professor of internal medicine, The University of Texas Southwestern Medical Center, Dallas.

While information obtained at baseline can be useful, he said, “we may learn even more about how a tumor will behave by watching what happens during treatment.”

The assessment methods discussed at San Antonio-optical tomography with ultrasound and the risk algorithm-both ask a fundamental question: Can we look into the biology of tumors by observing what happens in a dynamic time frame?

Dr. Tripathy said the two methods may reveal slightly different aspects of a tumor. “My feeling is that the more information, the better,” he said.

Near-infrared technology can determine the total amount of blood in the tumor and suggest the number of blood vessels feeding it, he said. The computer interprets changes in the parameters, performs pattern recognition, and recognizes those things that are associated with worse outcomes, he said, “although what those parameters are measuring we cannot yet be sure.”

Once we have constructed an algorithm, he said, “we will have a simple instrument that will measure light and scatter with software and produce a predictive score, which will need to be validated.”

The clinical risk algorithm, which includes pre- and post-treatment Ki67, offers yet another way to evaluate the tumor. Ki67 level is a significant predictor, “though in itself, it is not huge,” he said. “With the mixing of independent indices, as in Dr. Ellis’ risk model, one obtains better discrimination and better predictive value.”

Dr. Tripathy believes that a combination approach will work best: “We will probably get better prediction using multiple parameters from different perspectives, getting information from imaging and proteins and genes. But it will be too expensive to have everything. We will need to determine which approach gives us 90% of the information we want.”