| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Breast Imaging, CentraHealth/Virginia Baptist Hospital, 1330 Oak
Ln., Suite 202, Lynchburg, VA 24503.
2 Department of Radiology, CB 7515, Chapel Hill, NC 27599-7515.
3 Pathology Consultants of Central Virginia, Lynchburg, VA 24501.
Received April 13, 2004;
accepted after revision April 6, 2005.
Address correspondence to M. C. Alexander
(camille.alexander{at}centrahealth.com).
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. Patients with primary invasive breast cancer diagnosed in 1993-1997 were grouped according to the size of the lesion: 0.1-0.9 cm and 1.0-1.4 cm. Each tumor was placed in one of five mammographic prognostic categories: stellate without calcifications; circular without calcifications; and calcifications with or without tumor mass in a casting, crushed-stone, or other (powdery, punctuate, or round) pattern. To assess reproducibility, a second radiologist gave an independent interpretation in the first 109 cases. Descriptive data were stratified by tumor size, and tests of association were done with an extension of Fisher's exact test. Odds ratios and confidence intervals were computed. Weighted log-rank test and Kaplan-Meier survival curves were used to compare breast cancer survival in the stellate group compared with the other groups.
RESULTS. Two hundred one consecutive patients with a median follow-up period of 7.4 years were identified. There were nine breast cancer deaths. The stellate morphologic pattern was most common (91 [45.3%] of 201 cases), yet there was only one breast cancer death in this group (survival rate, 98.9%; 95% confidence interval [CI], 96.7-100%). In the group of stellate lesions smaller than 1.0 cm, 67.6% (25/37) of the tumors were well-differentiated without lymph node metastasis (30 [96.8%] of 31 cases), and there were no deaths. In the group of stellate lesions measuring 1.0-1.4 cm, 66.7% (36/54) of the tumors were well-differentiated with a 19.6% risk of lymph node metastasis, one death, and a survival rate of 98.1% (52/53; 95% CI, 94.4-100%). Circular tumors accounted for 29.9% (60/201) of tumors and 55.6% (5/9) of breast cancer deaths. Casting and crushed-stone microcalcifications were associated with 33.3% (3/9) of disease-specific deaths. A weighted kappa value of 0.89 (CI, 0.83-0.94) indicated very high agreement of pattern assignment.
CONCLUSION. Stellate tumors had a significantly better survival prognosis than tumors with other patterns even though there were no differences in treatment. By recognizing these small malignant lesions, trained radiologists may be able to identify tumors that pose negligible risk of breast cancer death.
Keywords: breast cancer • mammography • prognosis • stellate
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
In most large research studies of outcomes and attributes of small lesions
according to clinical TNM classification, investigators have used relatively
large groupings of size ranges, T1 (
2.0 cm) or even T1 and T2 (> 2 cm
and 5 cm). Smaller pathologic TNM subsets such as T1a ( 0.5 cm) and
T1b (> 0.5 cm and 1.0 cm) are subject to digit preference, and
relatively few publications describe isolation of these size groups for study
of their mammographic-prognostic characteristics
[6,
7]. Stage I grouping criteria
allow tumors up to 2.0 cm, do not subclassify by smaller size, and exclude
patients with positive nodes. Variations in surgical techniques and pathologic
processing of lymph nodes may lead to 30% underestimation of metastatic
deposits [8], and the
importance of micrometastasis when found is controversial
[9]. Furthermore, the growing
trend toward administration of neoadjuvant chemotherapy before surgical
removal of malignant tumors confounds accurate pathologic determination of
tumor size and thus TNM classification. These factors raise the question
whether the body of scientific evidence regarding the outcomes associated with
broad size ranges of small breast tumors accurately reflects the reality of
the presentation and outcome of very early breast cancer. The inability of the
medical community to separate very small breast tumors with an excellent
long-term prognosis from those resulting in death remains despite knowledge of
histologic grade and lymph node status.
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
The study design focused on prognostic categories composed of already diagnosed cases of breast cancer with certain mammographic features. Each category was composed of tumors with one of several mammographic appearances. The categories closely followed category descriptions reported previously by other authors [7, 10]. Both mammograms and pathology results for each patient were required for inclusion in the study. Mammographic findings of cancer had to be visible, even if only in retrospect. The mammograms of four patients were lost, or their tumors were mammographically occult. When neoadjuvant chemotherapy preceded histologic tumor measurement, the case was disqualified (n = 2). A total of 201 patients with a new diagnosis of primary invasive breast cancer met the criteria for the study.
|
|
|
Tumor size was determined by the initial pathol ogist's microscopic (preferred) or gross assessment as documented in the pathology report; however, inconsistencies or omissions in size or other pathology data were clarified by slide review by a pathologist. For assessing multifocal disease, the largest-diameter invasive tumor focus was used as the tumor size in accordance with American Joint Committee on Cancer guidelines [14]. Histologic type was based on the final pathology report. Grading was scored with the Nottingham method of Elston [15]. Nuclear grading of ductal carcinoma in situ (DCIS) in cases presenting with the crushed-stone and casting mammographic patterns was determined according to the 1997 Consensus Conference Committee Guidelines [16].
Lymph node status data were derived according to size from the records of the 182 (90.5%) patients who underwent axillary node sampling. No sentinel node biopsies were performed in our community setting during the study period. Node status was recorded as the number of positive lymph nodes among the total number of nodes collected. Patients who did not undergo axillary sampling were removed from statistical analysis involving lymph node status.
Primary standard-of-care treatment administered to each patient was recorded. Surgical treatment options consisted of mastectomy and lumpectomy. Patients were not selected according to type of surgery. In most cases radiation therapy followed lumpectomy. Exceptions were made on the basis of very advanced patient age, clinical judgment, and patient refusal. Mastectomy for ipsilateral relapse of cancer, chest wall irradiation for recurrence after mastectomy, and palliative radiation for pain were not considered initial primary treatments. Chemotherapy administered after detection of relapse of disease was not defined as a primary intervention. Endocrine treatment was recorded if the patient took the drug for at least 3 years of therapy. Treatment data were collected to ensure no wide variations of treatment occurred that might have altered clinical outcome.
Clinical follow-up data were obtained through hospital inpatient records, outpatient records of the comprehensive breast imaging clinic, records of the radiation oncology clinic, records of attending medical and surgical specialists, and data submitted to the hospital cancer registry. Survival time was defined as the interval between the date of initial histologic diagnosis and last documented pertinent clinical or mammographic follow-up examination. Cancer status was determined by whether there was evidence of cancer at the last follow-up visit. If a patient had died, cause of death was obtained through the local cancer registry or through hospital chart review. After 3 years of no pertinent clinical or mammographic information and no indication of death, a patient was considered lost to follow-up. These patients were omitted from analyses of survival.
Patients with a history of cancer other than breast cancer were not excluded from the study. When metastatic disease was diagnosed, the clinicians' assessment of the cause of metastasis was used unless histologic evidence was documented, in which case the histologic evidence was used. Breast cancer-specific death and death of complications of breast cancer were listed as breast cancer-specific fatalities.
The assigning radiologist had no knowledge of clinical outcome or pathology results at the time of pattern classification. In a test of the reproducibility of pattern assignment, a second radiologist independently assigned mammographic patterns to the first 109 patients (1993-1995). Because only the first radiologist evaluated all mammograms for pattern assignment, the patterns assigned by the first radiologist were used for all statistical analyses, and a test for agreement in pattern identification was conducted between the two reviewers.
Descriptive proportions were compared by use of chi-square statistics. To test for an association between histologic grade and mammographic pattern, we stratified by tumor size and examined within-group and overall associations using an extension of the Fisher's exact test. For other two-way analyses, we computed odds ratios and confidence intervals using standard large-sample methods. Weighted log-rank tests and Kaplan-Meier survival curves were used to compare breast cancer survival across mammographic patterns. All analyses were done with SAS software (SAS Institute).
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Of the 201 invasive tumors, 41.3% measured less than 1.0 cm, 22.4% measured 1.0 cm, and 36.3% measured 1.1-1.4 cm. Table 2 displays the histologic types of breast cancer in our series by size group according to prognostic category. Infiltrating ductal carcinoma not otherwise specified was by far the most common histologic type in both size groups and in all prognostic categories. In the stellate group, 76% of the cases of cancer were ductal not otherwise specified; tubular and invasive lobular carcinoma each made up only 12% of the cases. Fifty percent of the tumors were well differentiated, and the frequency distributions of the two size groups were not significantly different (Table 3). Six patients were lost to follow-up. Overall, 41 (21%) of the patients died, but only 9 (4.6%) of the deaths were breast cancer specific. The disease-specific survival rate was 95.4% (186/195).
|
|
The predominant category of the 201 mammograms was stellate (45.3%) followed by round or oval (29.9%), casting calcifications (15.4%), crushed-stone calcifications (5.5%), and other calcifications (4.0%) (Table 1). One hundred fifty-one (75%) of the 201 tumors in the series were not associated with microcalcifications and were categorized as either stellate or circular. The most common mammographic pattern of tumors smaller than 1 cm was stellate (45%), followed by round or oval (23%). Proportionately, stellate and rounded features were less frequent in tumors smaller than 1 cm (56/83 [67.5%]) than in larger tumors (95/118 [80.5%]). Features associated with microcalcifications were more likely to be found in smaller tumors (27/83 [32.5%]) than in tumors 1.0 cm or larger (23/118 [19.5%]). However, because of the small numbers in our series, confidence intervals showed overlap and were therefore not statistically significant.
Association between mammographic category and histologic grade is displayed in Table 3. There was a statistically significant association between category and grade overall, the stellate group having a significantly higher proportion of well-differentiated tumors (p < 0.001) than the other four groups. Among tumors smaller than 1 cm, 67.6% of the stellate tumors were well differentiated, again a statistically significant finding compared with the lower percentage of well-differentiated tumors in the other patterns in this size category. Among the tumors measuring 1.0-1.4 cm, 66.7% of the stellate lesions were well-differentiated, and the proportion of well-differentiated tumors in the other patterns ranged from 37% (round) to 60% (crushed-stone calcifications). The association between category pattern and grade was not significant (p = 0.08) in the larger size group, but the trends were similar.
Among the 182 women who underwent axillary node dissection, fewer cases of positive lymph nodes occurred in those with stellate lesions. Among tumors smaller than 1 cm, positive nodes were found in only 3.2% (1/31) of lesions in the stellate group compared with 9.1% (4/44) of lesions with the other patterns combined. Among the tumors measuring 1.0-1.4 cm, positive nodes were found in 19.6% (9/46) of those with the stellate pattern compared with 23% (14/61) of tumors with the other patterns.
Table 4 shows treatments given according to stellate pattern and all other patterns. Patients with stellate tumors were less likely to undergo mastectomy and more likely to undergo lumpectomy with radiation therapy. Stellate tumors and tumors in all other pattern groups were managed with locoregional and systemic therapies at nearly identical frequencies.
|
Survival trends according to prognostic category pattern were compared according to frequency of occurrence within both size groups and overall. The frequency distributions for mammographic categories and for breast cancer-specific deaths occurring within those categories are depicted by bar graph in Figure 6. Despite the predominance of stellate tumors overall and in both size categories, there were few breast cancer deaths among patients with the stellate pattern of lesions. There were no cancer-specific deaths among the 37 patients with 0.1- to 0.9-cm stellate tumors (survival rate, 100%). Overall among the 89 patients with stellate tumors and known follow-up results, the disease-specific survival rate was 98.9% (95% confidence interval [CI], 96.7-100%).
|
The overall incidence of casting calcifications in our series was 15.4%; this group accounted for 22.2% (2/9) of the breast cancer deaths. Casting calcifications were 3.1 times more likely (95% CI, 1.4-6.9 times) to occur in 0.1- to 0.9-cm tumors (20/83 [24.1%]) than in those larger than 1 cm (11/118 [9.3%]). Most cases (27/31 [87.1%]) of the casting pattern reflected DCIS in association with the invasive component. Of the 27 cases with DCIS and casting calcifications, 25 were described by the pathologist as high-nuclear grade DCIS; the other two were labeled low nuclear grade. The four cases in which DCIS was not found were reviewed both by the pathologist and the radiologist. These cases had only one or two rod-shaped calcifications with a preponderance of crushed-stone forms. These calcifications were located within the tumors (n = 2) and represented tumoral necrosis or incidental fibrocystic changes in the adjacent breast tissue (n = 1). Calcifications were unaccounted for in the fourth case. In our series only two patients had a large area of involvement (at least one quadrant of the breast) that has been reported as carrying a poor long-term prognosis. One of these patients died of breast cancer 4 years after the initial diagnosis. The other was alive and disease free after 10 years of follow-up. The only patient in our series known to have systemic metastatic disease at the time of this writing presented initially with the casting microcalcification pattern. Excluding the one patient in this pattern group lost to follow-up, the disease-specific survival rate was 93.3% (28/30) (95% CI, 84.4-100%).
The overall incidence of crushed-stone calcifications in our series was 5.5%; 11.1% (1/9) of breast cancer deaths occurred in this pattern group. Crushed-stone microcalcifications reflected high- or intermediate-nuclear grade DCIS in association with a small invasive tumor in 7 of 11 cases (63.6%). As in the casting group, the other four cases represented either fibrocystic changes (n = 2) or tumoral necrosis (n = 2). Like casting calcifications, crushed-stone microcalcifications were more common in tumors smaller than 1.0 cm than in larger tumors (7.2% vs 4.2%). The overall breast cancer-specific survival rate among patients with crushed-stone calcifications was 90.9% (10/11) (95% CI, 73.9-100%).
Combining cases of crushed-stone and casting microcalcifications that represent high-nuclear grade DCIS, as other authors have done [10], would have accounted for 12.9% of patients (26/201) and 33.3% (3/9) of deaths. The survival rate for this combined DCIS calcification group would have been 91.9% (34/37) (95% CI, 93.1-100%).
Only a small minority of the overall study population had 0.1- to 1.4-cm invasive tumors with the other-calcifications pattern. Among the eight cases in this category there was no associated DCIS and there were no deaths. Caution is in order when analyzing this group because of its limited size.
The survival curves based on log-rank analysis comparing stellate survival versus survival for all other groups are displayed in Figure 7. We removed non-breast cancer deaths from this analysis to compare rate of only breast cancer deaths with survival rate. If the one patient with an unknown cause of death, who was in the stellate group, is excluded, the curves show significantly better survival in the stellate group (p = 0.01). In an alternative, worst-case scenario, if the unknown cause of death among the stellate group is considered breast cancer, the p value becomes 0.04, still suggesting the prognosis for stellate tumors is better than for other types.
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Thurfjell et al. [10] reported on 96 invasive tumors smaller than 10 mm classified by pattern and correlated with survival. The median follow-up time was 7.0 years. Most (58.3% [n = 56]) of those small tumors were stellate on mammograms. There were two (3.6%) node-positive stellate tumors and two deaths in this size range. The survival rate of 95% associated with stellate tumors was the highest among the mammographic patterns. In our study, the statistically significant 98.9% survival rate for 0.1- to 1.4-cm stellate tumors in comparison with the other patterns, even though there were no differences in treatment between the pattern groups, suggests the stellate pattern was an important prognostic factor and that the effect of adjuvant therapy on these small tumors may be small. Although the survival prognosis for the other groups remains good, the high frequency of stellate tumors and exceptionally good survival outcome of the stellate tumors are largely responsible for the excellent survival estimates for small tumors as a whole.
The favorable prognostic influences of small size, lymph node negativity, and well-differentiated grade probably were strong influences within the stellate mammographic group. Certainly the smallest stellate tumors tended to be well differentiated and have low risk of lymph node metastasis. However, even patients with large stellate tumors with positive lymph nodes or poor histologic grade did well clinically. Perhaps the stellate tumors grow more slowly, incite a more aggressive host immune response, or have biologic characteristics that make them particularly vulnerable to the host's immune system. The resulting irregular fibroelastotic tissue that gives these tumors their mammographic appearance may thus mirror an enhanced effective immune response.
Several authors have reported that detection of small stellate or circular tumors without calcifications is more difficult for radiologists but is more important for the patient's survival probability than is detection of lesions with microcalcifications [21-23]. Computer-assisted detection of stellate tumors is less sensitive than is that of calcified lesions [24]. Training of radiologists with special emphasis on the subtle signs of noncalcified lesions is crucial for decreasing the rate of false-negative interpretations and improving survival statistics for women with small breast tumors.
Clear, concise, standardized definitions of mammographic prognostic categories are needed. Our definition of stellate was based on Swedish researchers' descriptions and is broader than the BI-RADS nomenclature [6, 7, 10, 25]. Our definition effectively includes all noncalcified tumors that are not rounded or oval and that are mammographically visible. Sonography is not a reliable aid in pattern assignment in our experience. In our series, both radiologists were trained similarly and practiced in the same breast imaging department. These factors may have been responsible for the high level of agreement between the two radiologists' category assignments. The disagreements in calcification categories stemmed from the strict definition of the casting group wherein visualization of only one fine, linear, or rod-shaped calcification was needed. Judging the predominant appearance may be more realistic. Assignment of casting and crushed-stone categories when calcifications were within a visible tumor mass helped account for the casting and crushed-stone cases in which no DCIS was found by the pathologist. We suggest including in the calcification categories only cases of calcifications occurring outside any visible tumor mass and representing DCIS. Sectioning sample error with the standard bread loaf technique may have caused pathologists to miss small components of DCIS. Finally, in the past pathologists did not always consider DCIS important to mention when there was a predominant invasive component (Biesemier, K.W.; personal communication). A high level of reproducibility among general radiologists and those who specialize in mammography is essential before mammographic categories can be used as a clinical tool.
Consistent detection of very small stellate breast tumors and decreased intensity of their treatment hold the potential for huge cost and morbidity savings. Because the incidence of lymph node involvement is less than 5%, several authors assert axillary lymph node dissection should not be performed routinely [26, 27]. Chemotherapy may not improve survival in the case of small tumors [28] and may have serious, life-threatening side effects. Radiation therapy after lumpectomy commonly results in morbidity and costs in excess of $20,000 per patient but does not improve survival [29, 30]. Nonetheless, if margin status is not considered, radiation therapy does significantly diminish the risk of recurrence in the tumor bed [29, 30], and the costs of treatment may be offset by the lack of need for repeated surgery and treatment. Justification is warranted if the risks of these adjuvant therapies for small stellate tumors are not balanced by the benefits. The results of the large study in which Tabar et al. [20] linked mammographic category and adjuvant treatments with long-term disease-specific survival suggest adjuvant therapy for stellate and other low-risk breast tumors is unnecessary.
In summary, in this sample of 201 small invasive breast tumors, our results indicated that small stellate malignant tumors of the breast have a significantly better survival curve than other mammographic prognostic categories and that radiologists who specialize in breast imaging can have excellent agreement in pattern assignment. Additional reproducibility studies and establishment of clear determinants for inclusion in each prognostic group are needed. Increasing the numbers and proportion of stellate tumors in the small size ranges, reevaluating adjuvant therapies according to prognostic category of mammographic presentation, and perhaps decreasing the intensity of therapy for tumors that are highly unlikely to be fatal may improve prognosis and quality of life among women with small breast tumors.
Acknowledgments
We thank James L. Lynde for initial pattern assignment and Linda Crist,
Sandra Angel, Cindy Cyrus, and Alisa Blankenship for invaluable research
assistance. We also acknowledge the statistical and programming assistance of
Ahinee Amamoo and Brian Neelon, of the Lineberger Comprehensive Cancer Center
at the University of North Carolina at Chapel Hill.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  A. A. Luck, A. J. Evans, J. J. James, E. A. Rakha, E. C. Paish, A. R. Green, and I. O. Ellis Breast Carcinoma with Basal Phenotype: Mammographic Findings Am. J. Roentgenol., August 1, 2008; 191(2): 346 - 351. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 Y. Wang, D. M. Ikeda, B. Narasimhan, T. A. Longacre, R. J. Bleicher, S. Pal, R. J. Jackman, and S. S. Jeffrey Estrogen Receptor-Negative Invasive Breast Cancer: Imaging Features of Tumors with and without Human Epidermal Growth Factor Receptor Type 2 Overexpression Radiology, February 1, 2008; 246(2): 367 - 375. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
C. J. D'Orsi and F. M. Hall BI-RADS Lexicon Reemphasized Am. J. Roentgenol., November 1, 2006; 187(5): W557 - W557. [Full Text] [PDF]
|
|
|
|
|
|
M. C. Alexander and B. C. Yankaskas Reply Am. J. Roentgenol., November 1, 2006; 187(5): W559 - W559. [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
