Prospective study of quantitative broadband diffuse optical spectroscopy for differential diagnosis of suspicious breast lesions.

31

Background: A definitive breast cancer diagnosis requires a biopsy. However, suspicious lesions classified BI-RADS 3 and lower are not indicated for biopsy, leading many patients to undergo extended imaging surveillance lasting 6 to 18 months. This period of uncertainty can cause significant stress and anxiety, and represents a missed opportunity for treatment if malignant. Light-based diffuse optical spectroscopy (DOS) has been investigated as a noninvasive and safe technique for breast cancer assessment due to its sensitivity to tissue composition [1-4]. In this prospective study, we evaluate the discriminative power of DOS for distinguishing suspicious benign and malignant breast lesions in patients undergoing biopsy. Methods: Sixty-eight women, eighteen years or older, with at least one suspicious breast lesion and scheduled for biopsy were enrolled. Imaging was done using a custom-made continuous-wave and frequency-domain DOS system. Measurements were acquired in a 4x4 or 5x5 grid pattern centered over the lesion with 1cm spacing and a 1cm buffer of normal tissue, depending upon lesion size. A mirrored measurement was collected on the contralateral breast. Tissue concentrations were obtained for oxyhemoglobin, deoxyhemoglobin, methemoglobin, water, lipids, and collagen [4]. The lesion region of interest was defined as the interior points of the grid. From those, the average absolute chromophore value, average lesion-to-normal (L/N) ratio, Z-score-normalized values, and malignancy probability were computed using a previously published model [3]. Results: Of the 68 enrolled subjects, 21 were excluded due to operator error (N=7), instrument failure (N=3), or insufficient signal-to-noise ratio (N=11). Among the remaining subjects, 14 had biopsy-confirmed malignant lesions, and 33 had benign lesions. Analysis of absolute concentrations revealed that collagen was significantly elevated (p=0.037) in malignant lesions compared with normal tissue, while methemoglobin approached significance (p=0.058); the other five chromophores did not present any significant differentiating power. For the z-score data, only 36 subjects were analyzed, and water was the only significant differentiator (p=0.039). Conclusions: Of the 68 enrolled subjects, 21 were excluded due to operator error. Among the remaining subjects, 14 had biopsy-confirmed malignant lesions, and 33 had benign lesions. Analysis of absolute concentrations revealed that collagen was significantly elevated (p=0.037) in malignant lesions compared with normal tissue, while methemoglobin approached significance (p=0.058); the other five chromophores did not present any significant differentiating power. For the z-score data, only 36 subjects were analyzed, and water was the only significant differentiator (p=0.039). Clinical trial information: N/A.