Deepa Narayanan

High-Resolution Fluorodeoxyglucose Positron Emission Tomography with Compression (“Positron Emission Mammography”) is Highly Accurate in Depicting Primary Breast Cancer

Abstract:  We sought to prospectively assess the diagnostic performance of a high‐resolution positron emission tomography (PET) scanner using mild breast compression (positron emission mammography [PEM]). Data were collected on concomitant medical conditions to assess potential confounding factors. At four centers, 94 consecutive women with known breast cancer or suspicious breast lesions received 18F‐fluorodeoxyglucose (FDG) intravenously, followed by PEM scans. Readers were provided clinical histories and x‐ray mammograms (when available). After excluding inevaluable cases and two cases of lymphoma, PEM readings were correlated with histopathology for 92 lesions in 77 women: 77 index lesions (42 malignant), 3 ipsilateral lesions (3 malignant), and 12 contralateral lesions (3 malignant). Of 48 cancers, 16 (33%) were clinically evident; 11 (23%) were ductal carcinoma in situ (DCIS), and 37 (77%) were invasive (30 ductal, 4 lobular, and 3 mixed; median size 21 mm). PEM depicted 10 of 11 (91%) DCIS and 33 of 37 (89%) invasive cancers. PEM was positive in 1 of 2 T1a tumors, 4 of 6 T1b tumors, 7 of 7 T1c tumors, and 4 of 4 cases where tumor size was not available (e.g., no surgical follow‐up). PEM sensitivity for detecting cancer was 90%, specificity 86%, positive predictive value (PPV) 88%, negative predictive value (NPV) 88%, accuracy 88%, and area under the receiver‐operating characteristic curve (Az) 0.918. In three patients, cancer foci were identified only on PEM, significantly changing patient management. Excluding eight diabetic subjects and eight subjects whose lesions were characterized as clearly benign with conventional imaging, PEM sensitivity was 91%, specificity 93%, PPV 95%, NPV 88%, accuracy 92%, and Az 0.949 when interpreted with mammographic and clinical findings. FDG PEM has high diagnostic accuracy for breast lesions, including DCIS. 

Positron Emission Mammography: High-Resolution Biochemical Breast Imaging:

Positron emission mammography (PEM) provides images of biochemical activity in the breast with spatial resolution matching individual ducts (1.5 mm full-width at half-maximum). This spatial resolution, supported by count efficiency that results in high signal-to-noise ratio, allows confident visualization of intraductal as well as invasive breast cancers. Clinical trials with a full-breast PEM device have shown high clinical accuracy in characterizing lesions identified as suspicious on the basis of conventional imaging or physical examination (sensitivity 93%, specificity 83%, area under the ROC curve of 0.93), with high sensitivity preserved (91%) for intraductal cancers. Increased sensitivity did not come at a cost of reduced specificity. Considering that intraductal cancer represents more than 30% of reported cancers, and is the form of cancer with the highest probability of achieving surgical cure, it is likely that the use of PEM will complement anatomic imaging modalities in the areas of surgical planning, high-risk monitoring, and minimally invasive therapy. The quantitative nature of PET promises to assist researchers interested studying the response of putative cancer precursors (e.g., atypical ductal hyperplasia) to candidate prevention agents.

PET‐Guided Breast Biopsy

Abstract:  Molecular imaging, using positron emission tomography (PET), has become an integral step in the evaluation of many patients with malignancy. However, its use in patients with breast cancer has been limited by the lower levels of 18F‐fluorodeoxyglucose (FDG) uptake in some breast malignancies compared to other cancers, the small size of many breast cancers, and the need for biopsy under PET guidance. High‐resolution breast PET, or positron emission mammography (PEM), with biopsy guidance software, now addresses these issues. We report a prospective, multicenter study designed to test the efficacy and safety of PEM biopsy guidance software in women with FDG‐avid breast lesions worrisome for malignancy. The intervention chosen was vacuum‐assisted core biopsy. Nineteen subjects underwent a total of 24 PEM‐guided biopsies. All lesions were successfully targeted and sampled as determined by post‐biopsy image scan evaluation, specimen imaging, and pathologic concordance. Invasive cancer was identified in 13 of 24 lesions (54%), while four (17%) were high‐risk lesions and three of these were upgraded to malignancy at excision. No serious adverse events occurred and all patients found the procedure to cause only minimal to mild discomfort. High‐resolution PEM‐guided breast biopsy is both safe and effective for the sampling of PET‐depicted breast lesions.

Interpretation of Positron Emission Mammography: Feature Analysis and Rates of Malignancy

OBJECTIVE The purpose of our study was to define and illustrate standard terminology for describing findings on positron emission mammography (PEM) and provide associated rates of malignancy. SUBJECTS AND METHODS Three hundred eighty-eight women with newly-diagnosed breast cancer anticipating breast-conserving surgery completed a multicenter trial comparing PEM to MRI in assessment of disease extent. Morphologic terminology to describe PEM findings was patterned on BI-RADS for MRI, and investigators were trained in the PEM lexicon. PEM imaging features of known malignancies and additional PEM lesions were recorded and correlated with outcome. The reference standard was biopsy or at least a 6-month follow-up. RESULTS Of 166 additional lesions on PEM, 54 (33%) proved malignant, with median invasive tumor size 8 mm (range, 2-60 mm). Among 43 round or oval masses, 16 (37%) were malignant, compared with 16 of 21 (76%) of lobulated or irregular masses (p = 0.003). Among 14 findings of focal or regional nonmass uptake, two (14%) were malignant compared with four of 12 (33%) findings of linear-ductal or segmental uptake (p = 0.350). Malignancy rates for BI-RADS-type final assessments were category 2, one of 31 (3.2%); 3, three of 32 (9.4%); 4a, four of 18 (22%); 4b, nine of 33 (27%); 4c, 15 of 24 (63%); and 5, 22 of 28 (79%). On the basis of modeling, irregular or lobulated morphology was the strongest predictor of malignancy, followed by lesion laterality (i.e., ipsilateral to known cancer) then increasing semiquantitative (18)F-FDG uptake. CONCLUSION Use of standardized terminology to report PEM findings will facilitate effective communication of results and consistent management. A probably benign category 3 assessment carried a substantial rate of malignancy for lesions seen on PEM, and biopsy may be more appropriate than follow-up.

X-ray stereotactic lesion localization in conjunction with dedicated scintimammography

We are developing a dual modality system that combines digital X-ray mammography with gamma emission scintigraphy, on an upright mammography gantry in which the breast is held under mild compression by a support structure that is independent of the detectors. The X-ray source and detectors can be rotated around a fixed rotation axis permitting multiple views of the breast with fixed compression. Two such views can be combined as a stereotactic pair to obtain the three-dimensional location of breast lesions. Information about the location of the lesion within the breast permits corrections for attenuation and detector spatial resolution, resulting in more accurate estimation of the true lesion-to-background concentration ratio, based on the image lesion-to-background counts ratio. In this paper, we describe the model used to make these corrections, and present the results of phantom experiments designed to test the accuracy of our calculations.

Analysis of position-dependent Compton scatter in scintimammography with mild compression

During the past several years, we have been developing a breast scanning system that combines digital X-ray mammography with breast scintigraphy using a dedicated small field of view gamma camera. The relatively low uptake of /sup 99/mTc-sestamibi in the breast compared to other organs such as the heart results in a large fraction of the detected events being Compton scattered gamma rays. In this study, our goal was to determine whether generalized conclusions regarding scatter-to-primary ratios at various locations within the breast image are possible, and if so, to use them to make explicit scatter corrections to the breast scintigrams. Energy spectra were obtained from patient scans for contiguous regions of interest (ROIs) centered left to right within the image of the breast, and extending from the chest wall edge of the image to the anterior edge. An anthropomorphic torso phantom with fillable internal organs and a compressed-shape breast containing water only was used to obtain realistic scatter-only spectra for each ROI. The measured patient energy spectrum was fitted with a linear combination of the scatter-only spectrum from the anthropomorphic phantom and the scatter-free spectrum from a point source. This procedure was repeated for each of the ROIs. We found that although there is a very strong dependence on location within the breast of the scatter-to-primary ratio, the spectra are well modeled by a linear combination of position-dependent scatter-only spectra and a position-independent scatter-free spectrum, resulting in a set of position-dependent correction factors. These correction factors can be used along with measured emission spectra from a given breast to correct for the Compton scatter in the scintigrams. However, the large variation among patients in the magnitude of the position-dependent scatter makes the success of universal correction approaches unlikely.

Combined structural and functional imaging of the breast.

Scintimammography, or single gamma nuclear imaging of the breast, has shown promise as a way of characterizing certain biological properties of suspicious breast masses. Conventional scintimammography, performed using large clinical gamma cameras and prone patient positioning suffers from several drawbacks including poor sensitivity for small (< 1 cm) lesions and no reliable method for correlating scintigraphic findings with those of other imaging modalities. We are developing a system designed to overcome some of these problems. The system combines x-ray mammography with scintimammography on a common gantry. The x-ray and gamma ray images are obtained in quick succession, with the breast in a common configuration under mild compression. A digital x-ray detector is used, permitting rapid assessment of lesion location prior to gamma imaging, and enabling fusion of the x-ray transmission and gamma emission information in a single digital image. In a pilot clinical diagnostic study, the system has demonstrated high pathology-proven accuracy in differentiating benign and malignant masses.

Phantom study of radiotracer concentration quantification in breast scintigraphy

We are developing a breast imaging system that combines digital x-ray mammography and gamma emission scintigraphy in an integrated unit. Here we present results of a phantom study designed to investigate the use of lesion position information from the x-ray images to improve quantification of lesion and background radioactivity concentrations. The preliminary results indicate that the effects of gamma ray attenuation, partial volume averaging, and scatter on lesion image contrast can be modeled.

Analysis of Spatial Correlation between 99mTc-Sestamibi Uptake and Radiographic Breast Density

Breast scintigraphy is a technique by which the biological properties of breast lesions can be assessed using an injected radiopharmaceutical. It may be particularly useful for women with radiographically dense breasts, in whose mammograms, lesions are often obscured by breast tissue. We are evaluating a dual modality breast scanner developed at the University of Virginia for its ability to distinguish between benign and malignant lesions. The scanner obtains a digital mammogram and a gamma ray emission image in quick succession with the breast held under mild compression, resulting in a fused image in which structures in the digital mammogram can be directly correlated with those in the scintigram. Our experience has shown that radiopharmaceutical uptake by normal breast tissue can sometimes obscure uptake by small lesions. It would therefore be advantageous to correct for this background uptake if possible. One potential way of accomplishing this is to use the information from the digital mammogram to help predict the background radiopharmaceutical distribution. With this in mind, we retrospectively investigated the degree of spatial correlation between the distribution of background activity and the distribution of radiodense breast tissue in normal breasts. Using a histogram-based analysis, we have quantified the degree of correlation in 16 images obtained from a total of 8 patients. We also used the mammographic images to quantify the radiographic density of each breast. Our results suggest that spatial correlation between areas of high radiopharmaceutical uptake and parenchymal density exists in the most dense regions of the breast for either extremely dense or heterogeneously dense breasts. High correlation was also observed for some homogeneously fatty breasts. In the latter case however, variation in breast thickness appeared to be the cause of the increased correlation. Correlation properties are approximately equal in both right and left breasts for a particular patient, except in cases exhibiting focal radiotracer uptake in a lesion. Although our preliminary results suggest that correlation between radiopharmaceutical uptake and parenchymal density exists, the number of cases thus far is too small for definitive conclusions. In addition, the planar nature of the dual modality scans imposes an inherent limitation on our ability to take into account attenuation of the emitted gamma radiation, which thus constitutes an uncontrolled variable in the correlation analysis. In principle, this problem can be eliminated by 3-dimensional imaging.

Improved lesion visibility in a dedicated dual head scintimammography system - phantom results

Phantom test results are presented simulating a dedicated dual head compact gamma imaging system for scintimammography. These studies were completed with various detector prototypes constructed from pixellated scintillators coupled to various arrays of position sensitive photomultiplier tubes. A 15cm /spl times/ 20cm dual-head prototype system is currently under development that will provide sufficient imaging area for breast imaging studies. The results indicate that the use of two detector heads placed on opposing sides of breast under mild compression substantially improves the lesion contrast regardless of lesion depth in the tissue (especially for small, centrally located lesions). In addition, there is significantly improved lesion signal by using a geometric mean technique, Square Root (Image 1* Image2), to combine the images from the opposing views. In count poor cases, a smoothing kernel was applied prior to image multiplication to reduce high spatial frequency noise.