Breast Density Should Play a Greater Role in MRI Screening Guidelines.

Guidelines for breast cancer screening with MRI were first proposed in 2007, only a few years after its clinical introduction. Those initial guidelines, which were generated by a committee sponsored by the American Cancer Society (ACS), have served as the template for similar recommendations by several organizations, with a singular goal regarding patient candidacy for MRI screening, a qualifying threshold based on risk stratification. Higher risk in those patients recommended for MRI screening translates to higher cancer detection rates, which in turn impacts cost-effectiveness. But there is another variable that should be as important as risk stratification in selecting patients for MRI screening: the probability that screening mammography will fail to detect developing cancer. That failure rate is a function of breast density, included in the MRI screening guidelines as a traditional risk factor but neglected when one considers its role as the primary cause of false-negative mammograms. The two implications of dense mammograms are essentially independent: (1) refining risk stratification and (2) predicting the "miss rate" of mammography. In the 2007 guidelines, indications for annual screening MRI, in addition to mammography, were based on patients having a calculated probability of "greater than 20-25% lifetime risk" for developing breast cancer. Other categorical risks, such as BRCA positivity, are listed in the ACS guidelines, but in effect, the threshold for adding MRI to the screening regimen has been a 20% lifetime risk for the development of breast cancer. While risk stratification in the original MRI screening guidelines had a number of inconsistencies, the focus herein is the questionable placement of high-density patients into the category described as "no policy for or against MRI, more research needed," a category where lifetime risks were grouped as 15-19%. Thus, mammographic density was relegated to its role as a traditional risk factor, while its potentially more significant impact, predicting the "miss rate" of mammography, had no role in patient selection for screening MRI. The 2007 ACS guideline committee was limited by the lack of available data, and since there was no evidence for mortality reduction at the time, the decision was made to follow the patient selection criteria that had been used in the six international MRI screening trials, even though there was little consistency among those trials. Since then, the number of screening MRI trials has more than doubled, and new trials are being designed and implemented with a focus on both features of density: risk and cancer camouflage. Enough evidence has accumulated during the 16 years subsequent to the original ACS high-risk screening guidelines to consider a complete revision that accounts for both numerical risk levels and density levels, much like what was used in the ACRIN 6666 trial. In establishing a new set of guidelines, our first question should be: What is the "miss rate" of mammography in this patient? If the chance of a false-negative mammogram is as high as we see with Level D density, then the decision to include MRI becomes straightforward. The traditional risk assessment would then be used to help determine the optimal interval between MRI screens while maintaining cost-effective cancer detection rates.

Historical Origins for the Overestimation of Mammographic Sensitivity.

The sensitivity of screening mammography for the early detection of breast cancer has improved over the years due to advances in technology. However, guidelines for screening mammography are often based on the mortality reductions demonstrated in the historic trials, where sensitivity with the first-generation mammography was relatively low. With attempts to establish risk:benefit ratios for population screening, it is important to understand the wide range of sensitivities that have been reported for mammography.  Original calculations for mammographic sensitivity were often based on studies that included palpable tumors, thus generating inflated numbers not fully applicable to non-palpable tumors. If restricted to asymptomatic screening, sensitivity calculations were often based on the inverse of interval cancers, a relatively inaccurate method since breast cancers missed on mammography can remain undetected clinically for several years. It was not until multi-modality imaging was developed, primarily ultrasound and MRI, where sensitivity determinations could be made in real time by cross-checking outcomes with each modality. From this, it became apparent that there was a strong correlation between breast density levels and sensitivity levels, such that a single number to denote mammographic sensitivity was disingenuous. The increasing awareness that mortality reductions in the historic trials were achieved with a low sensitivity tool has prompted great interest in additional technologic improvements in mammography, as well as multi-modality imaging approaches for women with high density and/or high risk. In order to appreciate the potential benefit of these new approaches, it is helpful to understand the historical basis behind overestimating the sensitivity of screening mammography.

Feasibility of Breast MRI as the Primary Imaging Modality in a Large Asian Cohort.

Purpose Contrast-enhanced MRI has repeatedly demonstrated significantly enhanced sensitivity compared to mammography and ultrasound in breast cancer detection. The purpose of this study was to evaluate the feasibility and outcomes of using breast MRI as the initial imaging study for screening and diagnosis.  Materials and methods In this retrospective review of a cohort of 10,374 breast MRI scans in 7967 patients in Taitung County, Taiwan, a total of 5619 participants met inclusion criteria and were included in our analysis. We reviewed all biopsies that were performed subsequent to MRI studies in women (screening vs. diagnostic). The primary outcomes were false-positive (FP) biopsy rates and positive predictive value (PPV) of MRI - parameters that have historically been associated with performance that restricts more widespread use of MRI. False-positive rate based on benign biopsies (FPR-3) and the positive predictive value (PPV-3) were calculated. Results Without complementary imaging or follow-up to identify false negatives, the study of performance characteristics was limited to false positives and PPV. There were 351 benign biopsies generated by MRI out of the cohort of 5555 participants (5619 minus the malignant biopsies), generating a false-positive rate of 6.3%. Sixty-four patients out of 415 biopsies were malignant, generating a PPV-3 of 15.4%. Conclusion In this Asian cohort, utilizing breast MRI as the initial study for screening and/or diagnosis appears to be limited more by practical considerations such as cost and patient flow efficiency than by feasibility based on performance characteristics. With well-established superior sensitivity, coupled with improved interpretive skills and techniques that allow for low false-positive rates, MRI should be further studied for its role as the primary imaging modality in breast screening and diagnosis.

Breast cancer survival following MRI detection in a high-risk screening program.

Guidelines for high-risk screening with MRI were introduced by the American Cancer Society in 2007, based on superior sensitivity of MRI over mammography, albeit without proven mortality reduction. The mortality end point is still unconfirmed, but international data are maturing with improved survival apparent, albeit subject to lead time and length bias. In this observational study of survival, we review 41 consecutive patients whose cancers (85.3% invasive) were detected through 2039 asymptomatic MRI screenings. With a minimum follow-up of 5 years and median follow-up of 10.2 years (range: 5.0-15.1), disease-specific survival is 100%.

Development and Assessment of a New Global Mammographic Image Feature Analysis Scheme to Predict Likelihood of Malignant Cases.

This study aims to develop and evaluate a new computer-aided diagnosis (CADx) scheme based on analysis of global mammographic image features to predict likelihood of cases being malignant. An image dataset involving 1,959 cases was retrospectively assembled. Suspicious lesions were detected and biopsied in each case. Among them, 737 cases are malignant and 1,222 are benign. Each case includes four mammograms of craniocaudal and mediolateral oblique view of left and right breasts. CADx scheme is applied to pre-process mammograms, generate two image maps in frequency domain using discrete cosine transform and fast Fourier transform, compute bilateral image feature differences from left and right breasts, and apply a support vector machine (SVM) to predict likelihood of the case being malignant. Three sub-groups of image features were computed from the original mammograms and two transformation maps. Four SVMs using three sub-groups of image features and fusion of all features were trained and tested using a 10-fold cross-validation method. The computed areas under receiver operating characteristic curves (AUCs) range from 0.85 to 0.91 using image features computed from one of three sub-groups, respectively. By fusion of all image features computed in three sub-groups, the fourth SVM yields a significantly higher performance with AUC = 0.96±0.01 (p<0.01). This study demonstrates feasibility of developing a new global image feature analysis based CADx scheme of mammograms with high performance. By avoiding difficulty and possible errors in breast lesion segmentation, this new CADx approach is more efficient in development and potentially more robust in future application.

Applying a new quantitative image analysis scheme based on global mammographic features to assist diagnosis of breast cancer.

BACKGROUND AND OBJECTIVE: This study aims to develop and evaluate a unique global mammographic image feature analysis scheme to predict likelihood of a case depicting the detected suspicious breast mass being malignant for breast cancer. METHODS: From the entire breast area depicting on the mammograms, 59 features were initially computed to characterize the breast tissue properties at both spatial and frequency domain. Given that each case consists of two cranio-caudal and two medio-lateral oblique view images of left and right breasts, two feature pools were built, which contain the computed features from either two positive images of one breast or all the four images of two breasts. Next, for each feature pool, a particle swarm optimization (PSO) method was applied to determine the optimal feature cluster followed by training a support vector machine (SVM) classifier to generate a final score for predicting likelihood of the case being malignant. To test the scheme, we assembled a dataset involving 275 patients who had biopsy due to the suspicious findings on mammograms. Among them, 134 are malignant and 141 are benign. A ten-fold cross validation method was used to train and test the scheme. RESULTS: The classification performance levels measured by the areas under ROC curves are 0.79 ± 0.07 and 0.75 ± 0.08 when applying the SVM classifiers trained using image features computed from two-view and four-view images, respectively. CONCLUSIONS: This study demonstrates feasibility of developing a new global mammographic image feature analysis-based scheme to predict the likelihood of case being malignant without lesion segmentation.

Redefining the sensitivity of screening mammography: A review.

From its inception, screening mammography has enjoyed a perceived level of sensitivity that is inconsistent with available evidence. The original data that imparted erroneous beliefs about sensitivity were based on a variety of misleading definitions and approaches, such as the inclusion of palpable tumors, using the inverse of interval cancer rates (often tied to an arbitrary 12 month interval), and quoting prevalence screen sensitivity wherein tumors are larger than those found on incidence screens. This review addresses the background for the overestimation of mammographic sensitivity, and how a major adjustment in our thinking is overdue now that multi-modality imaging allows us to determine real time mammographic sensitivity. Although a single value for mammographic sensitivity is disingenuous, given the wide range based on background density, it is important to realize that a sensitivity gap between belief and reality still exists in the early detection of breast cancer using mammography alone, in spite of technologic advances. Failure to recognize this gap diminishes the acceptance of adjunct methods of breast imaging that greatly complement detection rates.

Applying a new computer-aided detection scheme generated imaging marker to predict short-term breast cancer risk.

This study aims to investigate the feasibility of identifying a new quantitative imaging marker based on false-positives generated by a computer-aided detection (CAD) scheme to help predict short-term breast cancer risk. An image dataset including four view mammograms acquired from 1044 women was retrospectively assembled. All mammograms were originally interpreted as negative by radiologists. In the next subsequent mammography screening, 402 women were diagnosed with breast cancer and 642 remained negative. An existing CAD scheme was applied 'as is' to process each image. From CAD-generated results, four detection features including the total number of (1) initial detection seeds and (2) the final detected false-positive regions, (3) average and (4) sum of detection scores, were computed from each image. Then, by combining the features computed from two bilateral images of left and right breasts from either craniocaudal or mediolateral oblique view, two logistic regression models were trained and tested using a leave-one-case-out cross-validation method to predict the likelihood of each testing case being positive in the next subsequent screening. The new prediction model yielded the maximum prediction accuracy with an area under a ROC curve of AUC = 0.65 ± 0.017 and the maximum adjusted odds ratio of 4.49 with a 95% confidence interval of (2.95, 6.83). The results also showed an increasing trend in the adjusted odds ratio and risk prediction scores (p < 0.01). Thus, this study demonstrated that CAD-generated false-positives might include valuable information, which needs to be further explored for identifying and/or developing more effective imaging markers for predicting short-term breast cancer risk.

Prediction of breast cancer risk using a machine learning approach embedded with a locality preserving projection algorithm.

In order to automatically identify a set of effective mammographic image features and build an optimal breast cancer risk stratification model, this study aims to investigate advantages of applying a machine learning approach embedded with a locally preserving projection (LPP) based feature combination and regeneration algorithm to predict short-term breast cancer risk. A dataset involving negative mammograms acquired from 500 women was assembled. This dataset was divided into two age-matched classes of 250 high risk cases in which cancer was detected in the next subsequent mammography screening and 250 low risk cases, which remained negative. First, a computer-aided image processing scheme was applied to segment fibro-glandular tissue depicted on mammograms and initially compute 44 features related to the bilateral asymmetry of mammographic tissue density distribution between left and right breasts. Next, a multi-feature fusion based machine learning classifier was built to predict the risk of cancer detection in the next mammography screening. A leave-one-case-out (LOCO) cross-validation method was applied to train and test the machine learning classifier embedded with a LLP algorithm, which generated a new operational vector with 4 features using a maximal variance approach in each LOCO process. Results showed a 9.7% increase in risk prediction accuracy when using this LPP-embedded machine learning approach. An increased trend of adjusted odds ratios was also detected in which odds ratios increased from 1.0 to 11.2. This study demonstrated that applying the LPP algorithm effectively reduced feature dimensionality, and yielded higher and potentially more robust performance in predicting short-term breast cancer risk.

Overestimating Overdiagnosis in Breast Cancer Screening.

Overdiagnosis in breast cancer has been a focus of increasing concern with wide ranges of calculations made indirectly through the study of prospective randomized trials and analyses of large registries. While most admit that some degree of overdiagnosis is inherent with ductal carcinoma in situ (DCIS), the rate of overdiagnosis with invasive disease is highly controversial. Although it is generally accepted that overdiagnosis is calculated through indirect means and deductive reasoning, this is not entirely the case. Patients who refuse treatment, yet curiously return for follow-up, allow a direct glimpse at the natural history of screen-detected cancers. And historic autopsy studies offer information as to undiagnosed disease prevalence from the pre-screening era. While these autopsy studies support a modest degree of overdiagnosis in DCIS, they do not support widespread overdiagnosis for invasive cancer. The 1.3% mean incidence of invasive disease from seven autopsy studies correlates closely with disease prevalence, a direct observation that cancers do not remain quiescent in the breast until death. If invasive breast cancer does not regress in untreated patients and does not remain quiescent, then the high estimates being calculated for overdiagnosis are more likely to be length bias from long natural histories rather than true overdiagnosis.

Integration of Serum Protein Biomarker and Tumor Associated Autoantibody Expression Data Increases the Ability of a Blood-Based Proteomic Assay to Identify Breast Cancer.

Despite significant advances in breast imaging, the ability to accurately detect Breast Cancer (BC) remains a challenge. With the discovery of key biomarkers and protein signatures for BC, proteomic technologies are currently poised to serve as an ideal diagnostic adjunct to imaging. Research studies have shown that breast tumors are associated with systemic changes in levels of both serum protein biomarkers (SPB) and tumor associated autoantibodies (TAAb). However, the independent contribution of SPB and TAAb expression data for identifying BC relative to a combinatorial SPB and TAAb approach has not been fully investigated. This study evaluates these contributions using a retrospective cohort of pre-biopsy serum samples with known clinical outcomes collected from a single site, thus minimizing potential site-to-site variation and enabling direct assessment of SPB and TAAb contributions to identify BC. All serum samples (n = 210) were collected prior to biopsy. These specimens were obtained from 18 participants with no evidence of breast disease (ND), 92 participants diagnosed with Benign Breast Disease (BBD) and 100 participants diagnosed with BC, including DCIS. All BBD and BC diagnoses were based on pathology results from biopsy. Statistical models were developed to differentiate BC from non-BC (i.e., BBD and ND) using expression data from SPB alone, TAAb alone, and a combination of SPB and TAAb. When SPB data was independently used for modeling, clinical sensitivity and specificity for detection of BC were 74.7% and 77.0%, respectively. When TAAb data was independently used, clinical sensitivity and specificity for detection of BC were 72.2% and 70.8%, respectively. When modeling integrated data from both SPB and TAAb, the clinical sensitivity and specificity for detection of BC improved to 81.0% and 78.8%, respectively. These data demonstrate the benefit of the integration of SPB and TAAb data and strongly support the further development of combinatorial proteomic approaches for detecting BC.

Applying a new quantitative global breast MRI feature analysis scheme to assess tumor response to chemotherapy.

PURPOSE: To develop a new quantitative global kinetic breast magnetic resonance imaging (MRI) features analysis scheme and assess its feasibility to assess tumor response to neoadjuvant chemotherapy. MATERIALS AND METHODS: A dataset involving breast MR images acquired from 151 cancer patients before neoadjuvant chemotherapy was used. Among them, 63 patients had complete response (CR) and 88 had partial response (PR) to chemotherapy based on the RECIST criterion. A computer-aided detection (CAD) scheme was applied to segment breast region depicted on the breast MR images and computed a total of 10 kinetic image features to represent parenchyma enhancement either from the entire two breasts or the bilateral asymmetry between the two breasts. To classify between CR and PR cases, we tested an attribution selected classifier that integrates with an artificial neural network and a Wrapper Subset Evaluator. The classifier was trained and tested using a leave-one-case-out (LOCO)-based cross-validation method. The area under a receiver operating characteristic curve (AUC) was computed to assess classifier performance. RESULTS: From the pool of initial 10 features, four features were selected by more than 90% times in the LOCO cross-validation iterations. Among them, three represent the bilateral asymmetry of kinetic features between two breasts. Using the classifier yielded AUC = 0.83 ± 0.04, which is significantly higher than using each individual feature to classify between CR and PR cases (P < 0.05). CONCLUSION: This study demonstrated that quantitative analysis of global kinetic features computed from breast MRI-acquired prechemotherapy has potential to generate a useful clinical marker that is associated with tumor response to neoadjuvant chemotherapy. J. Magn. Reson. Imaging 2016;44:1099-1106.

Computer-aided breast MR image feature analysis for prediction of tumor response to chemotherapy.

PURPOSE: To identify a new clinical marker based on quantitative kinetic image features analysis and assess its feasibility to predict tumor response to neoadjuvant chemotherapy. METHODS: The authors assembled a dataset involving breast MR images acquired from 68 cancer patients before undergoing neoadjuvant chemotherapy. Among them, 25 patients had complete response (CR) and 43 had partial and nonresponse (NR) to chemotherapy based on the response evaluation criteria in solid tumors. The authors developed a computer-aided detection scheme to segment breast areas and tumors depicted on the breast MR images and computed a total of 39 kinetic image features from both tumor and background parenchymal enhancement regions. The authors then applied and tested two approaches to classify between CR and NR cases. The first one analyzed each individual feature and applied a simple feature fusion method that combines classification results from multiple features. The second approach tested an attribute selected classifier that integrates an artificial neural network (ANN) with a wrapper subset evaluator, which was optimized using a leave-one-case-out validation method. RESULTS: In the pool of 39 features, 10 yielded relatively higher classification performance with the areas under receiver operating characteristic curves (AUCs) ranging from 0.61 to 0.78 to classify between CR and NR cases. Using a feature fusion method, the maximum AUC=0.85±0.05. Using the ANN-based classifier, AUC value significantly increased to 0.96±0.03 (p<0.01). CONCLUSIONS: This study demonstrated that quantitative analysis of kinetic image features computed from breast MR images acquired prechemotherapy has potential to generate a useful clinical marker in predicting tumor response to chemotherapy.

Diagnostic and prognostic microRNAs in the serum of breast cancer patients measured by droplet digital PCR.

BACKGROUND: Breast cancer circulating biomarkers include carcinoembryonic antigen and carbohydrate antigen 15-3, which are used for patient follow-up. Since sensitivity and specificity are low, novel and more useful biomarkers are needed. The presence of stable circulating microRNAs (miRNAs) in serum or plasma suggested a promising role for these tiny RNAs as cancer biomarkers. To acquire an absolute concentration of circulating miRNAs and reduce the impact of preanalytical and analytical variables, we used the droplet digital PCR (ddPCR) technique. RESULTS: We investigated a panel of five miRNAs in the sera of two independent cohorts of breast cancer patients and disease-free controls. The study showed that miR-148b-3p and miR-652-3p levels were significantly lower in the serum of breast cancer patients than that in controls in both cohorts. For these two miRNAs, the stratification of breast cancer patients versus controls was confirmed by receiver operating characteristic curve analyses. In addition, we showed that higher levels of serum miR-10b-5p were associated with clinicobiological markers of poor prognosis. CONCLUSIONS: The study revealed the usefulness of the ddPCR approach for the quantification of circulating miRNAs. The use of the ddPCR quantitative approach revealed very good agreement between two independent cohorts in terms of comparable absolute miRNA concentrations and consistent trends of dysregulation in breast cancer patients versus controls. Overall, this study supports the use of the quantitative ddPCR approach for monitoring the absolute levels of diagnostic and prognostic tumor-specific circulating miRNAs.

Absolute quantification of cell-free microRNAs in cancer patients.

The hypothesis to use microRNAs (miRNAs) circulating in the blood as cancer biomarkers was formulated some years ago based on promising initial results. After some exciting discoveries, however, it became evident that the accurate quantification of cell-free miRNAs was more challenging than expected. Difficulties were linked to the strong impact that many, if not all, pre- and post- analytical variables have on the final results. In this study, we used currently available high-throughput technologies to identify miRNAs present in plasma and serum of patients with breast, colorectal, lung, thyroid and melanoma tumors, and healthy controls. Then, we assessed the absolute level of nine different miRNAs (miR-320a, miR-21-5p, miR-378a-3p, miR-181a-5p, miR-3156-5p, miR-2110, miR-125a-5p, miR-425-5p, miR-766-3p) in 207 samples from healthy controls and cancer patients using droplet digital PCR (ddPCR) technology. We identified miRNAs specifically modulated in one or more cancer types, according to tissue source. The significant reduction of miR-181a-5p levels in breast cancer patients serum was further validated using two independent cohorts, one from Italy (n = 70) and one from US (n = 90), with AUC 0.66 and 0.73 respectively. This study finally powers the use of cell-free miRNAs as cancer biomarkers and propose miR-181a-5p as a diagnostic breast cancer biomarker.

An alternative approach to selecting patients for high-risk screening with breast MRI.

Current guidelines for adding breast MRI to annual screening mammography are based entirely upon stratification of risk, with a heavy focus on lifetime calculations. This approach is fraught with difficulty due to the reliance on mathematical models that vary widely in their calculations, the inherent age discrimination of using lifetime risks rather than short-term incidence, and the failure to incorporate mammographic density, the latter being an independent risk as well as the greatest predictor of mammographic failure. By utilizing a system of patient selection similar to what was used in the American College of Radiology Imaging Network (ACRIN) 6666 trial for multi-modality imaging, 33 women without a prior diagnosis of breast cancer were found to harbor mammographically occult carcinoma through MRI screening. These 33 patients represent a 2% yield, closely approximating the yields seen in prospective MRI screening trials of women at very high risk of breast cancer. Using the "~20-25%" minimum established by the American Cancer Society and later the National Comprehensive Cancer Network, the Gail model would have prompted the use of MRI in only 9 of 33 (27.3%) patients, the Claus model 1 of 33 (3%), and the Tyrer-Cuzick model 12 of 33 (36.4%). Using all three models and opting for the highest calculated risk, then including BRCA-positivity, still would have identified only 16 of 33 (48.5%) patients with occult breast cancer discovered by MRI. Only one patient was BRCA-positive, and none had lobular carcinoma in situ, while 6 of 33 patients (18.2%) had atypical ductal hyperplasia (ADH). Measures are proposed to refine patient selection for MRI screening through the use of short-term or categorical risks, mammographic density, while maintaining cost-effectiveness through longer MRI screening intervals.