| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Original Research |
1 Department of Radiology, Cedar Breast Clinic, McGill University Health Center,
Royal Victoria Hospital, 687 Pine Ave. West, Montreal, QC H3G 1A4,
Canada.
2 Department of Pathology, Cedar Breast Clinic, McGill University Health Center,
Royal Victoria Hospital, Montreal, QC H3G 1A4, Canada.
3 Department of Surgery, Cedar Breast Clinic, McGill University Health Center,
Royal Victoria Hospital, Montreal, QC H3G 1A4, Canada.
Received January 1, 2005;
accepted after revision February 22, 2005.
Address correspondence to B. Mesurolle.
Abstract
 Top Abstract IntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
MATERIALS AND METHODS. One hundred four lumpectomies were performed in 99 consecutive patients with 131 nonpalpable breast lesions after sonographically guided needle localization. All 104 surgical specimens were scanned on sonography, and 86 specimen radiographs were obtained. Visualization of the lesion on sonography was compared with specimen radiographs and histologic findings. Sonographic margin status was classified as negative (shortest distance between tumor and specimen margin, > 0.2 cm) or positive (shortest distance between tumor and specimen margin, 0.2 cm) and was compared with pathology results.
RESULTS. Specimen sonography showed 95.4% (125/131) of the excised abnormalities; nonfatty background and a lesion size of greater than 0.5 cm contributed significantly to the success of specimen sonography. Four of six lesions missed on sonography were identified on specimen radiography. Among 81 malignant specimens, sonography identified 38 specimens with positive margins and 43 with negative margins. Pathologic examination revealed eight false-positive and 10 false-negative results (21% false-positive rate and 23.2% false-negative rate).
CONCLUSION. Specimen sonography is an effective procedure for identifying the presence of the lesion within the specimen; however, it is of limited value in cases of small hypoechoic lesions against a fatty background. Assessment of margins is limited by both false-positive and false-negative results.
Keywords: breast • breast cancer • breast specimen • mammography • sonography
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Because of the increased number of sonographically guided needle localizations of nonpalpable masses and because of the high performance levels of recent sonography apparatus available in our practice, we found it interesting to retrospectively review our experience in evaluating specimens using sonography. The objective of this study was to clarify the following issues: What is the optimum technique with which to perform specimen sonography? Is specimen sonography reliable for visualization of an excised lesion? If so, how accurate is this technique for the assessment of gross tumor margins?
|
|
|
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
At the beginning of our experience, a subgroup of 86 specimens (containing 111 masses) were evaluated using specimen radiography in addition to sonography. Among that group, all 111 masses were preoperatively visualized on sonography and 74 on mammography. The subsequent 18 specimens were evaluated only on sonography because, after increased experience with specimen sonography, we estimated the results to be reliable enough to perform only sonography examination of the specimen when the lesion was sonographically identified.
Specimen sonography was being performed as part of routine care at our institution. Thus, our research ethics board considered specimen sonography as innovative care and did not require additional approval for its use. Permission was obtained from the hospital to review the patients' medical records.
Sonography Equipment and Wire Localization Procedure
The equipment used to perform sonography included a high-resolution scanner
with high-frequency linear-array 10-14-MHz transducers (15L8w broadband
transducer, Sequoia, Siemens Medical Solutions, Acuson; high-frequency Matrix
transducer PLT1204AX, Aplio, Toshiba). All procedures were performed in the
Cedar Breast Clinic on the morning of surgery. Needle localizations were
performed under sonography guidance with a 20-gauge curved J retractable hook
guidewire (Homer Mammalok needles, Medical Device Technologies). One needle
was used in 91 specimens, two needles in six specimens, and three needles in
five specimens, with each needle being located in a different mass. The
bracketing technique was used in the two remaining cases (four needles). All
surgeries were performed by one of the nine oncologic breast surgeons.
Postexcision Specimen Imaging Technique
All specimen sonography examinations were performed in the breast clinic by
two radiologists with experience in breast imaging; they had performed 97 and
seven specimen sonography examinations, respectively. The radiologists had
three objectives. The first was to identify the lesion; the second was to
measure the largest dimension of the lesion; and the third was to identify and
measure the shortest distance between the tumor edge and the closest surface
of the specimen. The transducer head was coated with gel and then covered with
a plastic sheath (Figs. 1A,
1B,
1C,
1D, and
1E). The same apparatus was
used to perform needle localization and specimen sonography. The excised
tissue previously oriented by the surgeon with placement of sutures or needles
was placed on a towel and oriented the same way as during the needle
localization (Figs. 1A,
1B,
1C,
1D, and
1E). The specimen was scanned
by applying gel and placing the probe on the specimen itself (Figs.
1A,
1B,
1C,
1D, and
1E). When the specimen was
being scanned, the hookwire was followed when present.
When the lesion was detected, it was scanned both longitudinally and transversely (Figs. 1A, 1B, 1C, 1D, 1E, 2A, 2B, 2C, 2D, 2E, 2F, and 2G). If the lesion appeared to be deeply located (close to the posterior surface of the specimen), the specimen could be turned over for additional visualization and better assessment of the margins (Figs. 2A, 2B, 2C, 2D, 2E, 2F, 2G, 3A, 3B, 3C, 3D, and 3E). The breast parenchyma surrounding the lesion (specimen background) was classified as follows on the basis of specimen sonography features: fatty, group A (Figs. 1A, 1B, 1C, 1D, 1E, 2A, 2B, 2C, 2D, 2E, 2F, and 2G); or nonfatty (i.e., fat and glandular tissue or mostly glandular tissue), group B (Figs. 3A, 3B, 3C, 3D, 3E, 4A, 4B, 4C, 4D, and 4E). The presence or absence of a hookwire or hookwires within the specimen was recorded. On the final sonography assessment, the margins were deemed positive or negative to compare sonography with pathologic findings. Positive margins were those where tumor was seen at the margin of the specimen or where the shortest distance between the tumor and the specimen margin was 0.2 cm or less. Negative margins were those where the shortest distance between the tumor and the specimen margin was more than 0.2 cm. We arbitrarily chose 0.2 cm because it corresponds, in practice, to the minimal distance measurable in a specimen between the tumor border and the margin of the specimen using our sonography apparatus. The overall time to perform specimen sonography was 3-6 min.
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
Pathologic Correlation
The specimen was then brought directly to the pathology department where
the pathologist inked, measured, and weighed it. All surgical specimens were
subjected to serial section examination and histologic evaluation. Specimen
interpretation was performed by pathologists experienced in breast diseases.
On the basis of the results in the pathology report, a score similar to the
sonography score was established regarding margin status. A similar
classification was used for both sonographic and pathologic margin assessment:
negative indicated no invasive carcinoma or ductal carcinoma in situ (DCIS)
within 0.2 cm of the cut surface of the resected specimen, and positive meant
that the tumor extended to the inked margin or within 0.2 cm of the inked
margin. If a lesion had both invasive and DCIS described in the pathology
report, it was considered an invasive ductal carcinoma (IDC). The grade and
intraductal component were recorded. An extensive intraductal component was
present when in situ cancer occupied 25% or more of the area encompassed by
the infiltrating tumor.
Statistical Analysis
Statistical analysis was performed using SAS software (version 8.2, SAS
Institute). Statistical comparisons of characteristics between subgroups were
performed using Wilcoxon's signed rank test or, when appropriate, the
Student's t test. A logistic regression (Wald chi-square test) was
used to investigate the relationship between specimen sonography success and
breast and lesion characteristics. A p value of 0.05 or less was used
to determine statistical significance. The sensitivity, specificity, and
positive and negative predictive values were calculated for predicting
positive histologic margins.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
|
Lesion Measurements
The longest dimensions of the lesions identified on specimen sonography
ranged from 0.4 to 2.5 cm (mean, 1 cm). Among the malignant lesions, the mean
sonography size was 1 cm and the mean pathologic size was 1.25 cm. The
difference was significant (p = 0.004, Student's t test).
Among the benign lesions, the mean sonography size (0.89 cm) did not
significantly differ from the mean pathologic size (0.84 cm) (p =
0.8241, Student's t test). Based on sonography findings, 63.4% of
localized lesions (83/131) measured 1 cm or less and 10.7% (14/131) measured
0.5 cm or less.
Success and Failure Rates
Specimen sonography identified 125 (95.4%) of 131 targeted lesions. Six
lesions (three IDC; two invasive lobular carcinoma; one DCIS, no extensive
intraductal component) in six patients localized with sonography (in vivo)
were not visualized within the specimen on sonography (ex vivo), which
corresponds to a failure rate of 4.6% (6/131). Information summarizing the
characteristics of negative specimen sonography is shown in
Table 2. One failure at the
beginning of our experience was probably related to the technique used because
needle localization and specimen sonography were performed by two different
operators in a case of a small, subtle lesion against a fatty background. Four
of six lesions missed on sonography were identified on specimen radiography.
Two lesions not identified on either specimen sonography or specimen
radiography corresponded to widespread DCIS and one of three multifocal
lesions. Additional tissue was removed at surgery, but the lesions were
identified in the initially obtained specimens on final pathologic
examination. Additional specimens were not imaged with specimen sonography or
specimen radiography.
|
Among the 86 specimens analyzed with both sonography and specimen radiography, specimen sonography depicted 105 of 111 localized lesions and specimen radiography revealed 70 (111 lesions visualized on preoperative breast sonography and 74 on preoperative mammography) (Figs. 1A, 1B, 1C, 1D, 1E, 3A, 3B, 3C, 3D, and 3E).
A logistic regression was used to investigate the relationship between
specimen sonography success and breast and lesions characteristics. Success or
failure of specimen sonography was significantly related to the background
type (p = 0.0164, Wald chi-square test) and to lesion size
(p = 0.0169, Wald chi-square test), but not to the grade of the
lesion, intraductal component, weight of the specimen, or presence or absence
of the hookwire within the specimen (p > 0.05). The group B
(nonfatty) background was found to contribute significantly to the success of
specimen sonography (group A vs group B: point estimate = 0.073; 95% Wald
confidence interval [CI], 0.012-0.434). If lesions were divided into two
groups according to size (small, 0.5 cm; large, > 0.5 cm), a lesion greater
than 0.5 cm was found to contribute more significantly to the success of
specimen sonography than a lesion of 0.5 cm or less (size, > 0.5 vs 
0.5 cm: point estimate = 84.5; 95% Wald CI, 8.635-822.7).
Margin Assessment
Based on final pathologic results, 44 (51.2%) of the 86 malignant specimens
had negative margins (Figs. 1A,
1B,
1C,
1D,
1E,
3A,
3B,
3C,
3D, and
3E) and 42 (48.8%) had
positive margins, of which 27 (31.4%) had close margins (Figs.
2A,
2B,
2C,
2D,
2E,
2F, and
2G) and 15 (17.4%) had
involved margins (Figs. 4A,
4B,
4C,
4D, and
4E) (seven corresponding to
the invasive component and eight to the intraductal component).
Among the 86 malignant specimens, 81 were used to compare histologic and sonography results. Five specimens with a single lesion that could not be identified on sonography were excluded. One specimen in which two of the three targeted lesions were identified on sonography (negative result) was not excluded. There were 10 false-negative (Figs. 4A, 4B, 4C, 4D, and 4E) and eight false-positive (Figs. 3A, 3B, 3C, 3D, and 3E) results in margin assessment by sonography corresponding to a 21% false-positive rate (8/38) and a 23.2% false-negative rate (10/43). The sensitivity, specificity, and negative and positive predictive values of specimen sonography for predicting histologic margins were 75%, 80.5%, 76.7%, and 79%, respectively. Correlations between the margins in specimen sonography and histology are shown in Tables 3 and 4. Logistic regression did not show any relationship between final results (false-negative, false-positive, true-negative, and true-positive) and specimen characteristics (size, weight, grade, presence of needle, intraductal component). Six (60%) of the 10 false-negative results were associated with an extensive intraductal component or exclusive DCIS (Figs. 4A, 4B, 4C, 4D, and 4E), but this finding was not statistically significant (p >0.05).
|
|
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
Technically, the procedure we used by applying gel directly to the specimen is easy to perform with fewer constraints than previously described. Some authors have placed the specimen in a cellophane bag [7] or in a container filled with a small amount of saline [6]. We do not find either step necessary. Because of the high image quality currently available, these manipulations can be avoided. High-resolution transducers used in the present study allowed adequate visualization of superficial lesions even in the first few millimeters of the transducer's near field of view. In this context, we found that rescanning the specimen after turning it over was useful to visualize with more clarity the volume of normal-appearing tissue between the lesion and the specimen margin.
Another interesting point, in our opinion, is the necessity to perform both (in vivo and ex vivo) examinations using the same sonography apparatus. We did not test this hypothesis in performing localization and specimen sonography on two different machines. However, intuitively, we assume that the different settings available on each specific apparatus can give various appearances to the lesions depending on whether compounding, harmonic, or regular settings are used. Performing both in vivo and ex vivo examinations with the same apparatus might be particularly relevant in cases of small or subtle lesions that are difficult to identify on sonography. The presence or absence of the needle within the specimen did not affect the successful outcome of the technique; however, it seemed relevant to ask the surgeons to keep the hookwire in position within the specimen because detecting the lesion is easier by following the hookwire, especially in a large specimen. In addition, the hookwire is helpful in orienting the specimen properly to evaluate the margins.
The major problem with lumpectomy after needle localization is failure to excise the target, which in modern series occurs in fewer than 3% of cases [8]. In our series, no failure of needle localization was observed. These good results might be attributed to two factors. First, only masses and no clusters of microcalcifications were localized by sonography guidance. Unsuccessful needle-localized surgical breast biopsy is more likely reported with microcalcifications [5]. Second, sonography guidance allows real-time placement of the hookwire through the mass in all cases. Mammographic guidance allows locating the needle close enough to the lesion, but without traversing it, in a significant number of cases [9].
Sonography examination of surgically excised specimens is a reliable method with which to confirm complete excision of breast lesions previously identified on sonography. In our study, specimen sonography successfully documented 95.4% of the excised lesions (98.2% success rate if specimens with fatty backgrounds were excluded). Negative results of specimen sonography were predictable: They were observed with small hypoechoic lesions against a fatty background or with entities known to be difficult to image on sonography such as extensive DCIS [10, 11]. However, specimen radiography allows visualization of the lesion in most of these cases. Our results suggest that specimen sonography can be an alternative technique of specimen imaging in cases of nonfatty breasts and for lesions measuring more than 0.5 cm. This is particularly true in the case of a lesion detected on sonography exclusively or one that is hardly visible on mammography—that is, in cases in which specimen sonography has the advantage of showing clearly the lesion.
The low success rate of specimen radiography in our study (63%) compared with that of a previous study (93%) [12] is attributed to the large number of lesions exclusively identified on sonography in that study. In our practice, specimen sonography is performed not only in cases of lesions identified on sonography but also in cases of lesions identified on both sonography and mammography. In cases of negative specimen sonography, our policy is to perform radiography of the specimen. Ideally, as suggested by Fornage et al. [6], failure to visualize the mass in the specimen should prompt the radiologist to perform intraoperative sonography and scan the area of the wound to identify the residual mass (if the lesion was clearly identified preoperatively).
Eighty-three percent of patients in our study were operated on for malignant lesions. This high rate, compared with those of previous studies (20-30% malignancy rate), is attributed to the dramatic reduction in the number of surgical biopsies [13] because all patients underwent preoperative sonographically guided biopsy. Percutaneous diagnosis facilitates preoperative planning and enables breast cancer surgical treatment—that is, breast-conservation surgery—in one operation [4, 14, 15]. This surgical act must balance the goals of maintaining a good cosmetic result and achieving adequate excision. Indeed, there is widespread agreement in the literature that tumor cells from in situ or invasive lesions at the margin of a lumpectomy are associated with a higher rate of cancer recurrence than when histologically negative margins are obtained [16-21]. Clear histologic margins of resection were obtained in 82.6% of specimens in our study, which is similar to a previously reported rate of 92% [13, 22]. The selection of cancers in our series manifesting only as masses likely partly contributed to this high rate because residual tumors have been found in cancers manifesting as calcifications more often in those seen as masses on mammography (69% vs 44%, respectively) [23].
Specimen radiography is the reference imaging technique with which to visualize localized lesions, particularly in cases of microcalcifications [2, 3, 5, 9]. However, specimen radiography has a limited ability to adequately define the surgical excision margins. Specimen radiography is accurate only if radiography shows a tumor at the margins of the specimen (high positive predictive value) [23, 24]. We did not evaluate the value of specimen radiography in margin assessment in our study, but previous studies have shown that the negative predictive value for the assessment of the margins of a specimen is low (55.7% and 32%) with a false-negative rate of up to 44% [23, 24]. The reason is attributed to the limited ability to verify the radiologic margins on the basis of a 2D view of the specimen. A margin can be found to be involved only if an appropriate tangential view is obtained.
Compared with specimen radiography, specimen sonography shows a better negative predictive value (76.9% vs 32%, respectively), but a more limited positive predictive value (79% vs 98%, respectively) [24]. Visualization of the lesion in 3D allows a better assessment of its location within the specimen. Scanning in two perpendicular planes with the ability to turn over the specimen allows measurements of the distance between all sonographically visible lesions' margins and the specimen.
Specimen sonography shows fewer false-negative results than specimen radiography, but underestimates margin involvement in 23% of negative results. The first factor, which might explain false-negative results, is attributed to the intraductal component. As in a study by Graham et al. [24], these false-negative results were more frequent in cases of exclusive or extensive intraductal component (60% of the false-negative results) even if not significant. Despite recent technical improvements, the intraductal component is known to be poorly visualized on sonography. Sonography features are often vague and exact lesion delineation is uncertain, decreasing the negative predictive value [10, 11]. As suggested by Mai et al. [25], appropriate surgical management of patients whose core biopsy showed DCIS only might include wider resection margins than would otherwise be taken. The second factor that radiologists should keep in mind is the propensity of sonography to significantly underestimate the size of a lesion, which can explain an inadequate margin assessment with underestimation of margin involvement. Indeed, as in previous studies [26], we observed that sonography underestimated the size of malignant lesions in the present study.
Interestingly, and contrary to specimen radiography, specimen sonography can overestimate margin involvement, which explains the relatively low positive predictive value compared with specimen radiography. This can be attributed to several factors. Despite the fact that no significant variable was statistically identified, an important, but not measurable, factor related to the sonography technique used to scan the specimen must be discussed: Applying the transducer directly on the specimen to create enough surface friction might induce flattening of the specimen (similar to the "pancake phenomenon" described with specimen radiography [27]), contributing to underestimation of the normal-appearing tissue volume located between the transducer (specimen surface) superficially and tumor margin deeply, and might cause the tumor to appear artificially close to the specimen surface. This factor is particularly relevant if the lesion is not perfectly centrally located in the specimen, if the lesion is close to one edge, and if the specimen itself is small. From a histologic point of view, invasive lobular carcinoma was associated with false-positive margins in two cases. In those particular cases, invasive lobular carcinoma showed a classic sonography appearance with ill-defined margins associated with posterior attenuation, obscuring the posterior aspect of an ill-defined lesion.
Margin assessment with sonography does not replace final pathologic assessment and should be carefully interpreted. An assessment of the margins can give the surgeon relevant information similar to intraoperative gross pathologic evaluation of tumor margins if the lesion is well identified and clearly delineated. Even if no definite factor can explain the over- or underestimation of margin involvement, we consider specimen sonography to be of value if the lesion appears centrally located in the specimen and if the intraductal component is limited. However, specimen sonography is not reliable enough to be used alone for determining the presence or absence of residual breast cancer after the initial excision of nonpalpable breast lesions because it can overestimate or underestimate margin involvement.
This technique shows some weaknesses identified in our study. Obviously, specimen sonography is not indicated for identification and localization of isolated microcalcifications without an associated mass. It is also not indicated for lesions barely visible on sonography (subtle and hypoechoic lesions against a fatty background). Mixed lesions with a fluid component (papillomas) may be difficult to detect on specimen sonography as well because the fluid component present in vivo at the time of needle localization usually disappears ex vivo in the specimen. However, in such cases, identification of the needle allows the radiologist to find the solid component.
Another limitation is related to the size of the specimen. A large specimen can be difficult to scan, and visualization of a small lesion can be time consuming if the hookwire previously traversing the lesion has been removed. When the specimen is small, adequate examination and identification of several lesions is difficult, responsible for one failure in identifying one of three lesions in a specimen. In such cases, the immersion technique with the specimen placed in a container filled with a small amount of saline [6] might be useful for complete visualization of the specimen.
The duration of the procedure itself is short. However, this parameter is probably not of practical significance and is definitively underestimated. In fact, real procedure duration should include the time from the surgical suite to the radiology department and then to the pathology department. These parameters can be extremely variable, depending mostly on the availability of radiologists, and can have an impact on operating room and anesthesia times [28, 29]. Furthermore, in the case of specimen sonography, the surgeon's confidence in the radiologist must be total. Introducing a new step in the process may complicate the procedure and fuel the debate [29]. It is for this reason that we think that the radiologist performing the needle localization and performing specimen sonography should be the same person.
The last limitation that is common to all sonography examinations is the operator dependence of the technique. Most of the examinations of this study were performed by one radiologist, which represents a significant bias in the evaluation of the technique.
In conclusion, concerning the objectives of the study, the following statements can be made: First, the results of this study suggest that postexcision specimen sonography is a simple and easy technique to perform and is a reliable method with which to confirm complete excision of a nonpalpable breast mass previously identified in vivo with sonography. Second, the value of specimen sonography is high in cases of nonfatty background and multiple lesions. Third, visualization is limited in cases of small, subtle, and hypoechoic lesions, especially those against a fatty background. Fourth, gross assessment of the margins is possible but is limited to a preliminary description of the exact position of the lesion within the specimen related to the margins. Specimen sonography is of limited value in cases of associated extensive intraductal component and in cases in which the lesion is not centrally located (under- and overestimation of margin involvement).
Acknowledgments
We thank Amina Barhdadi for assistance in the statistical analysis.
|
TopAbstractIntroductionMaterials and MethodsResultsDiscussionReferences |
|---|
CiteULike 
Complore 
Connotea 
Del.icio.us 
Digg 
Reddit 
Technorati What's this?
This article has been cited by other articles:
|
|
 |
|
  C. L. Mercado, A. A. Guth, H. K. Toth, L. Moy, D. Axelrod, and J. Cangiarella Sonographically Guided Marker Placement for Confirmation of Removal of Mammographically Occult Lesions After Localization Am. J. Roentgenol., October 1, 2008; 191(4): 1216 - 1219. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 T. G. ODLE Breast Ultrasound Radiol. Technol., January 1, 2007; 78(3): 222M - 242M. [Abstract] [Full Text] [PDF]
|
|
| |||||||||||||||||||||||
