| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
ORIGINAL ARTICLES |
From the Departments of Surgery (LE, EK, SH, HK), Bioengineering (SP), Pathology (DS), Radiology (LE, YL, NH), and Medicine (DT, HR, JP), University of California, San Francisco, California.
Correspondence: Address correspondence and reprint requests to: Laura Esserman, MD, 2356 Sutter Street, 6th Floor, San Francisco, CA 94115; Fax: 415-353-9571; E-mail: laura.esserman{at}ucsfmedctr.org
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Methods: Thirty-three women who received neoadjuvant doxorubicin and cyclophosphamide chemotherapy for 4 cycles and serial breast MRI scans before and after therapy were evaluated for this study. Response to therapy was measured by change in the longest diameter on the MRI.
Results: Five distinct imaging patterns were identified: circumscribed mass, nodular tissue infiltration diffuse tissue infiltration, patchy enhancement, and septal spread. The likelihood of a partial or complete response as measured by change in longest diameter was 77%, 37.5%, 20%, and 25%, respectively.
Conclusions: MRI affords three-dimensional characterization of tumors and has revealed distinct patterns of tumor presentation that predict response. A multisite trial is being planned to combine imaging and genetic information in an effort to better understand and predict response and, ultimately, to tailor therapy and direct the use of novel agents.
Key Words: Magnetic resonance imaging • Neoadjuvant chemotherapy • Locally advanced breast cancer
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Since the initial study by Heywang et al. that demonstrated contrast enhancement of breast carcinomas on magnetic resonance imaging (MRI), numerous subsequent studies have been conducted to evaluate the efficacy of contrast-enhanced MRI for detecting, diagnosing, and staging breast disease.3 These further studies established that essentially all breast malignancies enhance with gadolinium. They also provided evidence that contrast-MRI is highly sensitive to cancers in the breast as small as a few millimeters in size.4–10 Sensitivities were reported in the range of 95%–100%. The limitation was low to moderate specificity, ranging from 37%–97%, with false positive enhancement occurring frequently in benign breast lesions.4,5,8,9,11–18
With the demonstrated performance of contrast-enhanced breast MRI, the current focus is toward determining the appropriate role of breast MRI in the clinical management of breast cancer. Potential roles include providing differential diagnosis as an adjunct to mammography, ultrasound, and clinical exam, local staging of disease extent, and high-risk screening. MRI can be developed as a staging tool and can accurately represent the extent of cancer when compared to pathology specimens.19 For local staging, MRI can be used to estimate the extent of disease prior to or following neoadjuvant chemotherapy. Numerous studies have examined the ability of MRI to accurately estimate residual disease after neoadjuvant chemotherapy.20–23 Abraham and colleagues evaluated MRI as a tool for determining response and extent of disease after neoadjuvant therapy. The responses of 39 patients with Stage II, III, and IV disease were assessed by MRI, clinical examination, and mammography. Clinical examination assessments corresponded with MRI assessments 52% and 55% (surgical and medical oncologist, respectively) of the cases. Mammography agreed with MR response in 52% of the cases. MR accurately predicted the pathologic determination of residual disease in 97% (30/31) of the cases.20 An additional study by Davis et al. reported the accuracy of MRI, ultrasound, and mammography for measuring the largest diameter in comparison with the largest cancer diameter at pathologic assessment. MR was found to have the highest correlation coefficient with pathologic assessment (r = .98) and the smallest standard error (.34). The correlation coefficients of ultrasound and mammography were found to be .45 and .46, respectively. MRI was found to be particularly more accurate for larger cancers.21 Moreover, Rieber et al. evaluated MR as a method to monitor therapy response by classifying 13 patients as responders or nonresponders to neoadjuvant chemotherapy based on MR findings. MR assessment correctly identified all cases of malignancy. In the responders, the investigators observed a flattening of the Gd-DPTA uptake curve after the first cycle of chemotherapy and an absence of Gd-DTPA uptake after the fourth cycle. Despite this change in Gd-DPTA behavior, which led to an underestimation of tumor extent in two patients and false negative assessments in four patients, the study concludes that MR provides information regarding response to therapy after the first cycle.22 Gilles et al. used dynamic contrasted enhanced MR subtraction to assess residual tumor after neoadjuvant chemotherapy in 18 women with locally advanced breast cancer. MR showed early contrast enhancement in all women with residual tumor except one patient with nodular residual tumor. The study reports that MR images of contrast-enhanced lesions correlated well with histopathologic analysis in all but two patients.23 These studies demonstrate substantial support for the use of MR for assessing residual disease and therapeutic response after neoadjuvant chemotherapy. The imaging strategy for staging uses high spatial resolution, fat-suppressed, three-dimensional imaging covering the entire symptomatic breast. MRI is particularly robust in stage III breast cancers for the capture of size, pattern, and heterogeneity of large tumors and thus is potentially a very useful tool to measure response to therapy.
Studies have shown that pathologic response after neoadjuvant chemotherapy predicts survival. Studies from researchers from MD Anderson Cancer Center24–28 and others29,30 suggests that complete pathologic response (elimination of tumor) following neoadjuvant therapy, even in patients with stage IIIB and IIID disease, is strongly predictive of an excellent long-term survival. Additional studies have shown that tumor size and lymph node status retain predictive value after neoadjuvant therapy.27,28 Thus, it is reasonable to try to determine if comparison of serial MRI exams could predict which patients would be likely to have a good outcome, either a complete pathologic response or <1 cm of residual invasive tumor in the breast. The ability to predict response to chemotherapy, hormonal therapy, or other biological therapy and ultimate outcome after treatment would permit therapy to be tailored and prognosis to be estimated with greater degree of accuracy. If MRI proves to be accurate in demonstrating response to therapy, those with >1 cm of residual tumor might be better served with a change in regimens (if there is an equal likelihood of response) or novel therapeutic agents as they become available.
The purpose in studying MRI in patients undergoing doxorubicin and cyclophosphamide neoadjuvant therapy is to determine if MRI can identify change and if residual disease on MRI after chemotherapy is correlated with residual disease at pathology. Over the course of the clinical protocol to measure response to therapy by MRI, we found that there were distinct imaging patterns of phenotypes that appeared to predict response to treatment. In this manuscript, we describe our characterization of stage III tumors according to their initial phenotype and the associated response after induction chemotherapy.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
Patient characteristics are shown in Table 1. A total of 45 patients were entered into the doxorubicin and cyclophosphamide neoadjuvant protocol with serial MRI, but only 33 patients were considered assessable for the purposes of this study. One patient was not able to have her initial MRI scan prior to the first cycle of chemotherapy; one patient decided to undergo surgical excision 1 week after the first cycle of therapy; three patients were found to have metastatic disease during the course of their chemotherapy; one patient started tamoxifen prior to beginning her neoadjuvant chemotherapy; three patients had undergone excisional biopsies to establish the diagnosis of cancer; one patient died of a myocardial infarction during chemotherapy and never underwent surgical treatment; one patient never began therapy; and one patient was treated outside UCSF (University of California, San Francisco) and we were unable to obtain her tissue to compare with the MRI scans.
|
Pathology and MRI Size Correlation
For the purposes of this analysis, MRI size was measured as the longest diameter. We used the new WHO (World Health Organization) criteria for response as measured by the longest diameter, where 30% or more change has been demonstrated to be equivalent to a partial response or >50% volume reduction.31 When multiple nodules were observed over a large area, a composite measurement of all nodules was taken and change in the longest diameter of the composite measurement was recorded. When widely separate masses were observed, the longest diameter of the largest mass was recorded. Automated methods for total tumor size to represent changes in total volume as well as longest diameter are being developed and will be reported separately. Pathology specimens were serial sectioned from medial to lateral in a sagittal fashion. Tumor size was taken as the longest diameter of the infiltrating and in situ tumor observed in the gross specimen.
Categorization of Initial MRI Phenotype
After carefully studying patients and their responses, we noted differential responses by the tumor pattern at the time of the pretherapy MRI scan. Five dominant patterns were noted: dominant nodule with rim enhancement, infiltrating nodular, infiltrating diffuse, patchy enhancement, and septal spread. The patterns are described in Table 2.
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
30%; minimal (MR) for a change in LD <30%, and stable or progressive disease (SD/PD) for no change or increase in LD. Pathology was evaluated by the absence or size of residual invasive disease. In our series, nine patients had a complete response by clinical exam and 15 patients had a partial response. Only two patients had absence of residual disease at the time of pathology; only three had 
1 cm of residual disease. These are the only categories that correspond to better survival, 90% and 70% five-year survival, respectively.24 Evaluation by MRI showed that only four patients had 1 cm of residual tumor.
Size Correlation Between MRI and Pathology After AC Neoadjuvant Chemotherapy
We sought to discover whether or not MRI after doxorubin and cyclophosphamide neoadjuvant chemotherapy captures residual disease as measured by pathology. In our series, residual tumor size and MRI longest distance after chemotherapy correlates with a Spearman’s correlation coefficient of 0.84 (P < .0001) (Fig. 1). Upon performing the final analysis with a regression of zero intercept, we found that the model has an R2 of 0.92 (P < .0001). As seen in Fig. 1, correlation of tumor size and MRI size is excellent, especially with smaller specimens. It can be difficult to obtain an exact size of larger tumors. Given the nature of pathology sampling in mastectomy specimens, true size of these large tumors is often hard to estimate, especially in the doxorubicin and cyclophosphamide neoadjuvant setting where masses are less apparent and diffuse infiltrates are more common. The outliers above the line represent large tumors and mastectomy specimens with and without prior lumpectomy. It is important to note that we did not make a distinction between ductal carcinoma-in-situ (DCIS) and invasive disease. Whether or not MRI is capable of this distinction will be the subject of further investigation. Of the two patients who had an absence of all residual disease at the time of pathology, one was clearly predicted by MRI. The second patient is less clear; the MRI showed enhancement over a 2 cm area. The patient had a complete clinical response, but residual calcifications and a mass on ultrasound. The residual disease was excised using ultrasound wire guidance. There was no residual disease upon pathologic examination, and no residual calcifications upon repeat mammography. The MRI was not repeated after the lumpectomy and the patient declined further therapy. She returned for follow-up 5 months later and was found to have a local recurrence. A repeat MRI confirmed a recurrence in the exact same location that the original MRI had shown to contain residual disease; however, the recurrence was much larger, measuring 4.5 cm. In this unusual case, we cannot say whether MRI accurately predicted the presence and extent of residual disease but circumstances suggest that the MR image postneoadjuvant therapy had accurately detected residual disease. However, for the purpose of analysis, we considered this case to be a complete pathologic response. By percent volume response, both cases had >90% response. In the future, volume response may be a better method for measuring change, but we are still developing the methods for reliably predicting volume response.
|
1 cm of residual disease and those who had >1 cm of residual disease. Percent change in tumor size, as measured by longest diameter, correlates with residual tumor size (Table 3). Change in volume, as measured by change in longest diameter, is our best MRI measurement; our automated tumor volume measurements are in the process of being refined. The number of positive nodes also appears significant by this particular analysis.
|
|
Imaging Phenotypes
From our pilot study, we have identified five predominant MRI patterns in women who present with LABC. Over time, as we add more data, we will probably refine the categories, but have chosen to have more initial categories to maximize our chances of seeing distinctions among groups. The first pattern is the appearance of a single predominant focal mass with rim enhancement, displacing surrounding tissue. The second pattern is one where the tumor has irregular borders, but is nodular in character. Rather than one dominant mass, there are several masses over a region of tissue. Pattern three includes tumors where the tumor appears to replace the tissue and presents as a sheet of enhancement. The fourth pattern is one of an appreciable mass with sparse enhancement. The fifth category is one of a tumor pattern that exhibits septal spreading and skin thickening. In Table 5, we present the predominant histologic types, mean Scarff Bloom Richardson scores (tumor grade), and percent of patients with a complete or partial response by MRI (as measured by longest diameter) for each MRI. Pattern 1 is a high grade tumor that presents as a circumscribed mass and is the most likely to respond well to neoadjuvant chemotherapy. Another view of the response data can be seen in Fig. 2, which shows the range of response for each MRI pattern. Pattern is highly significant for predicting response to therapy, with P < .015 by the Kruskal-Wallis test.
|
|
|
|
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
MRI, because of its ability to delineate anatomic detail of breast tumors, seems to be a valuable tool in the assessment of response to therapy. The results in our study series are consistent with other data on stage III disease. The MD Anderson series (Kuerer et al., personal communication), demonstrated that residual tumor was a strong predictor of survival. In this series, which included 356 patients, patients who had a complete pathologic response to chemotherapy in the breast had an 89% survival rate, whereas patients with a less than complete response to therapy had a poor long-term survival rate (64%). This analysis can be further broken down into those that had <1 cm of invasive tumor remaining (70% survival) and those that had >1 cm of remaining invasive cancer (50% survival) (Kuerer et al., personal communication). Furthermore, in the MD Anderson series, 25% of the patients had a minimal response to therapy or progressive disease and a very poor outcome, regardless of therapy after surgery, 30% and 10% survival, respectively.24 Ten percent and 65% of the patients had a complete and partial response, respectively. Those who had a complete clinical response had an 80% survival rate, and those with a partial response had a 70% survival rate. In our pilot series, with only 33 patients, 27% had a complete response by clinical exam and 45% had a partial response by clinical exam, which was similar to the MD Anderson series. Even with a short follow-up, in our series, residual disease at the time of surgery was predictive of metastatic recurrence. Pathologic response is clearly more specific than clinical response, but, regardless, response is an important factor in survival.
While residual size or pathologic response has been established as predictive of outcome, it is important to recognize that residual size does not capture change over time and does not allow incorporation in rate or degree of change. MRI, on the other hand, has the potential to provide a quantitative measurement of response to therapy by change over time in longest diameter and volume. An important question is, then, whether or not MRI after chemotherapy captures residual disease as measured by pathology. As discussed above, MRI after doxorubicin and cyclophosphamide neoadjuvant chemotherapy strongly correlates with residual tumor. Of the two patients with complete response pathologically, MRI only predicted absence of disease in one of the cases when longest diameter was used. Larger studies are being planned to determine the value of MRI in the neoadjuvant setting and will include careful correlation with mammography. Very careful attention to pathologic correlation in a timely manner is critical to this study because small residual tumors could be missed in a mastectomy specimen. Others have also looked at the ability of MRI to predict residual tumor size after chemotherapy.20–23
The value of imaging in assessing response to therapy is critical to study because imaging is noninvasive and can be measured multiple times, making it a powerful tool to guide future therapy. Residual disease at the time of surgery is highly predictive of recurrence, but it is a parameter which can be assessed only once, and never again. Whether MRI will become a standard for women with stage III tumors is not established at this time. It must be carefully evaluated in a larger prospective study using other imaging modalities as well. However, MRI is potentially exciting because of its sensitivity and ability to capture other biologic information.
MRI appears to have predictive value in determining the type and degree of response to therapy. It may aid in clinical management by helping to appropriately set patients’ expectations about potential success or about the potential for breast conservation. In this small series, our results were highly predictive of clinical response. There are several measurements from MRI that may make a significant impact on our predictive ability. One of these measurements is change in longest distance. Change in longest diameter alone is important, but other parameters may also further enrich our ability to predict response. One of these measurements is viewing total volume of tumor, three-dimensional pixel by pixel. We are developing techniques to automate volumetric change as well as change in peak signal enhancement. Another of these measurements is studying the change in the pattern of signal enhancement. In prior work, we have noted the strong correlation between the signal enhancement ratio (SER) and microvessel density.37 Thus, signal enhancement ratio change may be particularly meaningful. Each of these parameters will probably contribute different information about the tumor. Change in longest diameter will probably help the most in determining who will be able to successfully undergo breast conservation. Some combination of all three will probably be the best predictor of response to therapy. For example, the strength of signal enhancement and the total volume or density decreases in pattern 3 cases, although the tumor extent (longest diameter) does not change. It is interesting to note that, in two of these cases, many positive nodes were identified, but neither patient has relapsed after two and a half and four years of follow-up, respectively. Only a larger study and long-term follow-up can tell us whether these factors will be important for predicting recurrence or the timing of recurrence.
Predictive markers, such as imaging phenotype, may have potential value by enabling us to tailor therapy in the future and to optimize response for those who benefit from chemotherapy. What MRI shows us is that, possibly, response is determined by a combination of biologic makers rather than tumor size and node status; this leads to a new paradigm for managing LABC. Larger studies have shown that initial tumor size is not predictive of pathology response. After adjustment for initial tumor size, estrogen receptor (ER) status, and nuclear grade were associated with a higher likelihood of complete primary and axillary node pathologic response (absence of residual disease), independent of the initial tumor size.24 In our series, the patients with MRI pattern 1, who have the best response to chemotherapy, have the highest grade tumors and are largely ER negative.
Hopefully, biologic and molecular markers will also correlate with MRI types. We are currently trying to map the MRI phenotypes to mammographic patterns to determine if we can construct a similar association with an imaging modality that is inexpensive and widely accessible. In our study, we evaluated patients with a range of tumor sizes, and patterns appear more distinct as tumor sizes enlarge. The MRI patterns are very apparent by the time tumors are >3 cm in size but less apparent at an earlier stage. Molecular changes may be more reliable than MRI patterns in smaller tumors. As we accrue more patients, we should be able to refine our characterization by initial imaging phenotype and add molecular characterization.
A multi-institutional trial is being designed to validate the association between imaging phenotype and response to therapy and to establish the best methods for reliably measuring change in tumor size and volume. In addition, we will be exploring the biological and molecular characteristics of the tumors by response to therapy as well as the MRI pattern at the time of diagnosis. We will use the current Intergroup stage III neoadjuvant breast cancer protocol (CALGB 49808) as the stage for assessing the value of using MRI and surrogate markers to predict response and to tailor therapy. Response will be assessed before and after all neoadjuvant chemotherapy, as well as after 2 cycles of chemotherapy, and in between courses of doxorubicin and cyclophosphamide and paclitaxel regimens. We will determine whether or not these time points can serve as a point at which clinicians should consider changing therapy because of failure to respond. In our pilot study, the best technique to measure volume change is the percent change in the longest diameter. However, one of the study’s goals will be to improve automated methods for volume response, and these will be measured against change in longest diameter. The other important goal of the study is to lay the groundwork to enable novel therapeutics to be tested in the stage III setting, prior to surgical excision, for patients who fail to respond to chemotherapy and obviously have a very grave prognosis. It is important to note that our current study measures only the effect of doxorubicin and cyclophosphamide. We do not have sufficient data on the effects of paclitaxel. The upcoming multi-institutional study will pair MRI with a trial that looks at both doxorubicin and cyclophosphamide and paclitaxel. In this larger trial, we will be able to test the findings presented in this study and add biologic and genetic markers to further differentiate tumor types to predict response.
|
Acknowledgments |
|---|
Received for publication April 19, 2000. Accepted for publication December 5, 2000.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
 |
|
 |
|
  T. A. Buchholz, C. D. Lehman, J. R. Harris, B. A. Pockaj, N. Khouri, N. F. Hylton, M. J. Miller, T. Whelan, L. J. Pierce, L. J. Esserman, et al. Statement of the Science Concerning Locoregional Treatments After Preoperative Chemotherapy for Breast Cancer: A National Cancer Institute Conference J. Clin. Oncol., February 10, 2008; 26(5): 791 - 797. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
A. C. Wolff, D. Berry, L. A. Carey, M. Colleoni, M. Dowsett, M. Ellis, J. E. Garber, D. Mankoff, S. Paik, L. Pusztai, et al. Research Issues Affecting Preoperative Systemic Therapy for Operable Breast Cancer J. Clin. Oncol., February 10, 2008; 26(5): 806 - 813. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 M. J. Campbell, L. J. Esserman, Y. Zhou, M. Shoemaker, M. Lobo, E. Borman, F. Baehner, A. S. Kumar, K. Adduci, C. Marx, et al. Breast Cancer Growth Prevention by Statins. Cancer Res., September 1, 2006; 66(17): 8707 - 8714. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 I. Sachelarie, M. L. Grossbard, M. Chadha, S. Feldman, M. Ghesani, and R. H. Blum Primary Systemic Therapy of Breast Cancer Oncologist, June 1, 2006; 11(6): 574 - 589. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
L. Esserman Integration of Imaging in the Management of Breast Cancer J. Clin. Oncol., March 10, 2005; 23(8): 1601 - 1602. [Full Text] [PDF]
|
|
|
|
|
|
E. P. Mamounas Continuing Evolution in Breast Cancer Surgical Management J. Clin. Oncol., March 10, 2005; 23(8): 1603 - 1606. [Full Text] [PDF]
|
|
|
|
|
|
N. Hylton Magnetic Resonance Imaging of the Breast: Opportunities to Improve Breast Cancer Management J. Clin. Oncol., March 10, 2005; 23(8): 1678 - 1684. [Full Text] [PDF]
|
|
|
|
 |
|
 E. Yeh, P. Slanetz, D. B. Kopans, E. Rafferty, D. Georgian-Smith, L. Moy, E. Halpern, R. Moore, I. Kuter, and A. Taghian Prospective Comparison of Mammography, Sonography, and MRI in Patients Undergoing Neoadjuvant Chemotherapy for Palpable Breast Cancer Am. J. Roentgenol., March 1, 2005; 184(3): 868 - 877. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Meisamy, P. J. Bolan, E. H. Baker, R. L. Bliss, E. Gulbahce, L. I. Everson, M. T. Nelson, T. H. Emory, T. M. Tuttle, D. Yee, et al. Neoadjuvant Chemotherapy of Locally Advanced Breast Cancer: Predicting Response with in Vivo 1H MR Spectroscopy--A Pilot Study at 4 T Radiology, November 1, 2004; 233(2): 424 - 431. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 E. P. Mamounas Tailoring Loco-Regional Therapy with Neoadjuvant Chemotherapy: Another Step in the Right Direction Ann. Surg. Oncol., October 1, 2004; 11(10): 888 - 891. [Full Text] [PDF]
|
|
|
|
|
|
F. Thibault, C. Nos, M. Meunier, C. El Khoury, L. Ollivier, B. Sigal-Zafrani, and K. Clough MRI for Surgical Planning in Patients with Breast Cancer Who Undergo Preoperative Chemotherapy Am. J. Roentgenol., October 1, 2004; 183(4): 1159 - 1168. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
Y. Roth, T. Tichler, G. Kostenich, J. Ruiz-Cabello, S. E. Maier, J. S. Cohen, A. Orenstein, and Y. Mardor High-b-Value Diffusion-weighted MR Imaging for Pretreatment Prediction and Early Monitoring of Tumor Response to Therapy in Mice Radiology, September 1, 2004; 232(3): 685 - 692. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
S. H. Giordano Update on Locally Advanced Breast Cancer Oncologist, December 1, 2003; 8(6): 521 - 530. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 D. A. Mankoff, L. K. Dunnwald, J. R. Gralow, G. K. Ellis, E. K. Schubert, J. Tseng, T. J. Lawton, H. M. Linden, and R. B. Livingston Changes in Blood Flow and Metabolism in Locally Advanced Breast Cancer Treated with Neoadjuvant Chemotherapy J. Nucl. Med., November 1, 2003; 44(11): 1806 - 1814. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
R. A. Jong, M. J. Yaffe, M. Skarpathiotakis, R. S. Shumak, N. M. Danjoux, A. Gunesekara, and D. B. Plewes Contrast-enhanced Digital Mammography: Initial Clinical Experience Radiology, September 1, 2003; 228(3): 842 - 850. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
T. M. Breslin Preoperative Chemotherapy for Breast Cancer: Do We Need to Redefine Our End Points? Ann. Surg. Oncol., August 1, 2003; 10(7): 716 - 717. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||
