Handheld optoacoustic scanner reveals molecular differences in breast tumors without biopsy

Handheld multispectral optoacoustic tomography (MSOT) can distinguish molecular subtypes of malignant breast cancer in living patients, according to an exploratory study published in European Radiology Breast. Researchers at the Technical University of Munich scanned 33 women with suspected breast cancer before biopsy and found that tumors lacking hormone receptors produced significantly stronger optoacoustic signals than hormone receptor-positive tumors, particularly in the tumor core. The findings suggest that light and sound together may one day offer a non-invasive window into tumor biology that currently requires tissue sampling.

Molecular subtyping of breast cancer currently depends on histopathological, immunohistochemical, and molecular analysis of biopsy specimens. Pathologists classify tumors according to the expression of the estrogen receptor, the progesterone receptor, the human epidermal growth factor receptor 2 (HER2), and the proliferation marker Ki-67. These four markers sort tumors into the generally defined subtypes luminal A, luminal B, HER2-positive, and triple-negative, each of which carries a different recommended therapeutic regimen, ranging from endocrine therapy to neoadjuvant or systemic chemotherapy. Because biopsy processing takes time, the search for imaging biomarkers that provide complementary information about tumor biology has become a priority, although the authors stress that such biomarkers cannot replace tissue diagnosis.

The technology behind the new study is optoacoustic, or photoacoustic, imaging. The tissue of interest is illuminated with short light pulses in the visible or near-infrared spectrum, and the light is absorbed by endogenous absorbers such as oxygenated hemoglobin (HbO2), deoxygenated hemoglobin (Hb), lipids, and water. That absorption generates mechanical waves with frequencies in the ultrasound range, which detectors pick up and reconstruction algorithms convert into images of tissue vasculature, perfusion, oxygenation, metabolic activity, and inflammation. Conventional optical imaging of the breast, which relies on near-infrared light alone, is hampered by photon scattering that degrades resolution and by the limited penetration depth of near-infrared light, which restricts its use in clinical routine. Optoacoustic imaging sidesteps some of these constraints by listening for sound rather than collecting scattered photons.

Previous work had already shown promise. Diot and colleagues identified high-resolution photoacoustic patterns of benign and malignant breast tumors using a handheld MSOT operating at 28 wavelengths, though that study involved only 10 participants with different types of cancers. Neuschler and colleagues then demonstrated in a pivotal multicenter study that optoacoustic imaging combined with ultrasound significantly improved the specificity of breast tumor assessment compared with benign tumors using a first-generation MSOT scanner. Dogan and colleagues subsequently used the same scanner to differentiate molecular subtypes, showing differences in optoacoustic internal vessels, internal blush, and internal hemoglobin between luminal cancers and triple-negative and HER2-positive cancers. The new study extends this line of work with a second-generation system and semi-quantitative signal measurements rather than qualitative scoring.

Between July 2021 and August 2022, the team prospectively recruited 33 patients with suspected breast cancer at a single center, with a mean age of 54 plus or minus 14 years. Each patient underwent digital mammography, gray-scale ultrasonography, an MSOT scan, and a core needle biopsy of the suspicious lesion as part of the routine diagnostic workflow. Biopsies confirmed malignant disease in all patients and provided material for molecular subtyping: 11 patients (33 percent) had luminal A tumors, 5 (16 percent) had luminal B tumors, 6 (18 percent) had HER2-positive tumors, and 11 (33 percent) had triple-negative tumors. No special type was the most common histological subtype, accounting for 23 of the 33 cases. There was no significant difference in mean age among the molecular subtypes.

The imaging protocol used a second-generation hybrid clinical MSOT/ultrasound system, the Acuity Echo, with a hand-held probe that both illuminates tissue and performs ultrasound scanning. Patients lay supine while the tumor was examined in two perpendicular directions, transverse and longitudinal, for about 30 seconds each. The system operates at a repetition rate of 25 Hz for real-time optoacoustic imaging, with co-registered ultrasound images captured at approximately 5 Hz. Each light pulse delivered nearly 10 millijoules of energy, well below safety limits for laser use in medical applications. Signals were acquired at 65 different wavelengths from 660 to 1300 nanometers in increments of 10 nanometers, reconstructed with a model-based approach, and then separated by linear unmixing into relative signal components corresponding to Hb, HbO2, lipids, and water. Total blood volume (TBV) was calculated as the combined relative hemoglobin-related signal. The authors emphasize that the resulting maps do not represent absolute chromophore concentrations or direct measurements of tissue oxygenation, only component-specific relative signal characteristics.

Post-processing relied on manual segmentation by an experienced optoacoustic imaging expert, verified by a board-certified radiologist with more than six years of experience. From each image series, one frame showing the least motion on the ultrasound greyscale image was selected, and the optoacoustic image was segmented into core, rim, and total tumor regions according to the corresponding ultrasound image. Mean wavelength-specific signal intensities at 750, 800, and 850 nanometers were calculated for each region, alongside the relative spectrally unmixed intensities for Hb, HbO2, and TBV. Statistical comparisons used one-way ANOVA with Bonferroni corrections across the four subtypes, and Welch’s t-tests or Mann-Whitney U-tests for the hormone receptor subgroup analysis, with two-sided testing at a significance level of 0.05.

The results showed a clear pattern when tumors were grouped by hormone receptor status. In the total tumor volume, wavelength-specific signal intensities at 750, 800, and 850 nanometers were significantly higher in the hormone receptor-negative group, comprising HER2-positive and triple-negative tumors, than in the hormone receptor-positive group, comprising luminal A and luminal B tumors, with p-values of 0.016, 0.046, and 0.046 respectively. Relative Hb-related signal intensities were also significantly higher in the hormone receptor-negative group (p = 0.005), while TBV- and HbO2-related intensities in the total volume showed no significant differences (p = 0.159 and p = 0.054). In the tumor core, the differences sharpened: wavelength-specific intensities at all three wavelengths were significantly higher in the hormone receptor-negative group (p < 0.001, p = 0.002, and p = 0.042), as were relative Hb- and TBV-related intensities (p ≤ 0.001 and p = 0.004). The tumor rim showed fewer differences, with only the 750-nanometer intensity and relative Hb-related signal reaching significance (p = 0.033 and p = 0.008).

Comparisons among all four molecular subtypes yielded fewer significant results, which the authors attribute to the small sample size of each group and the overlap in signal intensities between subtypes. In the tumor core, mean signal intensities at 750 nanometers were significantly lower in luminal A tumors than in triple-negative tumors (1.24 × 10−6 ± 0.61 × 10−6 versus 2.10 × 10−6 ± 0.75 × 10−6 arbitrary units, p = 0.026), and mean relative TBV-related intensities were significantly lower in luminal B tumors than in triple-negative tumors (0.11 ± 0.52 versus 0.22 ± 0.89 arbitrary units, p = 0.041). No other significant differences emerged when the other subtypes in the core or all subtypes in the rim were compared. The authors note that the most pronounced group-level differences occurred in the tumor core, suggesting this region may contain relevant biological information for subtype-related comparisons, although overlap between groups was noted, indicating that the current data do not yet support reliable patient-level molecular subtype classification.

The authors interpret the higher signals in hormone receptor-negative tumors as possibly reflecting the more aggressive biology of these tumors, which has been associated with vascular proliferation, altered oxygen metabolism, and hypoxia-related pathways. Intratumoral hypoxia has been linked to a higher risk of metastasis, more aggressive tumors, and higher mortality, so relative MSOT signal characteristics may eventually provide complementary prognostic information, though the authors caution that their prognostic relevance will need to be validated in future studies. The study’s limitations are considerable: it was exploratory and single-center with a limited number of patients; segmentation was manual and may introduce observer-dependent variability; the analysis relied on mean signal intensities and did not capture spatial distribution, texture, or heterogeneity; advanced multivariate approaches such as principal component analysis, radiomic analysis, or machine-learning-based classification were not performed because the limited cohort size would carry a substantial risk of overfitting; diagnostic performance metrics such as receiver operating characteristic analysis, area under the curve, sensitivity, and specificity were not calculated because the study was not powered or designed to develop a diagnostic classifier; and reproducibility analysis was not possible because all patients underwent treatment within two weeks after biopsy. The authors call for validation in larger multicenter cohorts and evaluation of whether MSOT-derived biomarkers provide incremental value beyond established imaging and histopathological parameters.

Subject of Research: Non-invasive molecular subtyping of breast cancer using handheld multispectral optoacoustic tomography

Article Title: In vivo molecular subtyping of malignant human breast cancer using multispectral optoacoustic tomography

Article References: Feuerriegel, G. C., Katsouli, N., Reidl, M., Fasoula, N.-A., Sridharan, P., Wiedemann, A., Makowski, M. R., Karlas, A., Ntziachristos, V., & Fallenberg, E. M. (2026). In vivo molecular subtyping of malignant human breast cancer using multispectral optoacoustic tomography. European Radiology Breast, 1(1), Article 6. https://doi.org/10.1007/s44502-026-00004-x

Image Credits: AI Generated

DOI: 10.1007/s44502-026-00004-x

Keywords: multispectral optoacoustic tomography, breast cancer, molecular subtyping, hormone receptors, triple-negative, photoacoustic imaging, tumor core, hemoglobin, total blood volume, biopsy, non-invasive imaging, exploratory study