Does magnetic resonance imaging accurately predict residual disease in breast cancer?

The American Journal of Surgery (2009) 198, 547–552

The American Society of Breast Surgeons

Does magnetic resonance imaging accurately predict residual disease in breast cancer? Chee-Chee H. Stucky, M.D.a,*, Sarah A. McLaughlin, M.D.b, Amylou C. Dueck, Ph.D.c, Richard J. Gray, M.D.a, Marina E. Giurescu, M.D.d, Susanne G. Carpenter, M.D.a, Gwen M. Grimsby, M.D.a, Heidi A. Apsey, N.P.a, Barbara A. Pockaj, M.D.a a

Department of Surgery, Mayo Clinic, 5777 E. Mayo Blvd., Phoenix, AZ 85054, USA; bDepartment of Surgery, Mayo Clinic, Jacksonville, FL, USA; cDepartment of Biostatistics, Mayo Clinic, Phoenix, AZ, USA; dDepartment of Radiology, Mayo Clinic, Phoenix, AZ, USA KEYWORDS: Breast conservation therapy (BCT); magnetic resonance imaging (MRI); re-excision; Positive margins; Invasive breast cancer; Residual disease

Abstract BACKGROUND: The accuracy of magnetic resonance imaging (MRI) in identifying residual disease after breast conservation therapy (BCT) is unclear. METHOD: Review of an institutional database identified patients with positive or close (ⱕ2 mm) margins undergoing MRI before re-excision. Histopathologic correlation was performed. RESULTS: Forty-three women underwent MRI after BCT. MRI suggested residual disease in 29 patients, of whom 20 (69%) had residual carcinoma pathologically. Nine patients had false-positive MRI as seen by benign pathology findings. Fourteen MRIs indicated no residual disease, of which 6 had residual disease pathologically. The sensitivity and positive predictive value of MRI was 77% and 69%, respectively. MRI conducted within 28 days of the original surgery was 85% sensitive. MRI performed after 28 days was 69% sensitive. CONCLUSIONS: MRI is able to detect residual disease among most patients undergoing re-excision. False-positive results may be caused by inflammatory processes that resemble residual disease. © 2009 Elsevier Inc. All rights reserved.

Magnetic resonance imaging (MRI) has become a useful tool in the diagnostic work-up of breast carcinoma. There are no definitive criteria for MRI evaluation in newly diagnosed breast cancer; although there are criteria for MRI screening in a select group of high-risk patients.1,2 Commonly used criteria include determination of the extent of disease in women with complex standard imaging (mammography and ultrasound), occult breast cancer in the face of dense breast tissue or pathologically confirmed lymph* Corresponding author: Tel.: ⫹1-480-342-2000; fax: ⫹1-480-342-2866. E-mail address: [email protected] Manuscript received March 24, 2009; revised manuscript April 18, 2009

0002-9610/$ - see front matter © 2009 Elsevier Inc. All rights reserved. doi:10.1016/j.amjsurg.2009.04.008

adenopathy, and suspicion of multifocal/multicentric disease.3,4 However, these and other indications do not represent strict criteria, and therefore the use of breast MRI often is performed at the surgeon’s discretion. As the technology for breast cancer imaging has improved, so has surgical therapy. Multiple studies have shown that breast conservation therapy (BCT) is equivalent to mastectomy in appropriately selected stage I and II breast cancer patients.5–7 These trials have shown that there are no significant differences in overall or disease-free survival rates when comparing the two treatments. Proper patient selection includes the verification that negative margins can be obtained while still maintaining a good esthetic result. With or without intraoperative evaluation of margins, the

548 rate of positive margins after lumpectomy is still high.8,9 Although most BCT patients undergo radiation therapy after surgery, the presence of residual disease or close margins (defined as ⱕ2 mm) confers an increased risk of local recurrence.10 –13 Therefore, most surgeons will re-excise to obtain negative margins.14 –16 However, residual disease is not always found in the re-excision specimen, suggesting that some patients potentially could avoid additional surgery if imaging reliably can prove no residual disease exists. Our goal was to determine whether MRI improved the detection of residual disease in the lumpectomy cavity in patients with positive or close margins after BCT. Also, we sought to evaluate if the length of time between surgery and MRI influenced the ability to detect residual disease.

Methods Patient selection A variety of databases containing prospectively collected data collected between January 2003 and June 2008 were reviewed retrospectively. Patients analyzed were treated for breast cancer with excisional biopsy or BCT with or without sentinel lymph node biopsy. Study patients included those with positive margins or close margins defined as 2 mm or less as documented on final pathology or those who had had previous BCT with concern of residual or recurrent disease at the lumpectomy cavity by either postexcision mammography or physical examination. If the patient’s original surgery was performed at an outside facility, the histology slides were reviewed at our institution to confirm the findings of close or positive margins as well as the original diagnosis. All the re-excisions were performed at 1 of 2 hospitals within the same institution. All patients underwent MRI at 1 of these 2 hospitals before returning to the operating room for re-excision. Patients with invasive cancer and/or ductal carcinoma in situ (DCIS) were included in the study.

Breast MRI technique MRI imaging was performed on a 1.5-T General Electric Healthcare (Waukesha, WI, USA) magnet. Patients were placed prone with the breasts positioned properly in a dedicated 8-channel breast coil. Fiduciary markers (vitamin E gel capsules) were applied on nipples and any areas of clinical interest. The diagnostic MR protocol began with preliminary imaging (of one or both breasts) with fast spin echo sagittal T2 with fat saturation and axial T1 sequences. It was followed by dynamic high-resolution simultaneous imaging of both breasts with the VIBRANT (General Electric Healthcare) sequence performed after intravenous administration of a contrast agent (usually gadolinium, .1 mmol/kg) and using fat saturation.

The American Journal of Surgery, Vol 198, No 4, October 2009 Three sagittal acquisitions were performed at 30 seconds, 3 minutes, and 6 minutes after the injection, followed by 1 axial T1 fast spoiled gradient (FSPGR) sequence. Postprocessing consisted of 2 series of subtraction images (30 s and 6 min). Slices were 3-mm thick, field of view was 18 to 22 cm and matrix was 256 ⫻ 256. Contrast-enhanced images were sent to the CADStream (Confirma; Bellevue, WA, USA) workstation, a MR computer-aided detection program. Maximum intensity projections, angiogenesis color maps, and time-enhancement kinetic curves can be generated to assist in interpretation. The examinations were interpreted before all re-excisions by dedicated breast radiologists who specialize in interpreting breast MRI examinations. They always were interpreted in conjunction with a recent mammogram (within 6 months) and ultrasound (the latter when available) and compared with any prior examinations. Morphologic and kinetic features of the enhancements as well as the lumpectomy cavity were described using Breast MR Lexicon terminology. Those with benign features were characterized as such and those with suspicious findings were labeled as consistent with residual disease.

Surgical methods BCT methods for nonpalpable tumors used radioactive seed-localization. The use of intraoperative frozen section (FS) assessment for margin evaluation was used at the discretion of the surgeon and pathologist. When FS was used, the surgeon achieved negative margins intraoperatively by frozen pathology. All FS margins were re-reviewed on permanent hematoxylin-eosin stains to confirm margin status. Standard practice at our institution is to achieve negative margins of at least 2 mm including both invasive tumor and DCIS components. If sentinel lymph node biopsy was performed, patients underwent dual mapping with technetium-99 –labeled sulfur colloid and isosulfan blue dye. Re-excision was performed as a separate surgery regardless of the MRI findings. Adequate tissue was removed from the margin of the lumpectomy cavity described as close or positive on original pathology. Again, intraoperative FS margin analysis was conducted at the discretion of the surgeon and pathologist to verify clear margins. If FS evaluation showed persistently close or positive margins, further re-excision was performed until margins of at least 2 mm were achieved or it was decided that BCT was not feasible.

Data collection and analysis Data points collected included patient characteristics such as age, tumor characteristics, indication for re-excision, date of original surgery, date of sentinel lymph node biopsy (if different from date of original surgery), and date

C.-C.H. Stucky et al. Table 1

MRI and residual disease

549

Results

Patient clinical characteristics

Characteristics T stage Tis T1 T2 T3 Pathology DCIS IDC ILC Mixed IDC/ILC Other Papillary Tubular Squamous cell Tumor markers ER⫹ PR⫹ Her-2/neu overexpressed Margin status Positive Close (ⱕ2 mm) Other Location of original surgery Outside our institution Within our institution

Number of patients

Percentage

9 25 5 4

21 58 12 9

9 22 7 2

21 51 16 5

1 1 1

2 2 2

37 32 3

86 74 7

24 14 5

56 33 12

30 13

70 30

Forty-three patients were included in this study. IDC ⫽ infiltrating ductal carcinoma; ILC ⫽ infiltrating lobular carcinoma; ER ⫽ estrogen receptor; PR ⫽ progesterone receptor.

of MRI evaluation. The length of time between the original surgery (ie, the surgery resulting in positive or close margins) and the MRI evaluation was categorized as 28 days or less, or more than 28 days. MRI detection of the presence or absence of residual disease, pathology verification of the presence or absence of residual disease, and the pathology findings within the re-excision specimen also were recorded. Data analysis consisted of placing patients into 1 of 4 categories: true-positive, true-negative, false-positive, or false-negative MRI. Patients with a true-positive MRI had MRI findings consistent with residual disease that was verified on pathology. Those with a false-positive MRI had MRI findings consistent with residual disease but none found on pathology. Those in the true-negative category had MRI findings that were consistent with benign postsurgical changes verified pathologically. Those in the false-negative category had MRI findings consistent with benign postsurgical changes but were found to have residual disease pathologically. The sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), and accuracy then were determined. Patients without residual invasive or in situ disease on final pathology were classified into 1 of 5 pathologic groups: atypia, fat necrosis, normal breast tissue, proliferative breast tissue, and other.

Patient characteristics Forty-three women were included in this study. The average patient age was 59 years (range, 39 – 84 y) and the average tumor size was 1.8 cm (range, .2–7.5 cm). Nine women (21%) had tumors larger than 2 cm. Additional tumor characteristics are summarized in Table 1. Most women had an invasive breast cancer, with most having either infiltrating ductal carcinoma (51%) or infiltrating lobular carcinoma (16%). Nine women had pure DCIS. Of the 43 patients evaluated, 24 patients (56%) underwent further surgery for positive margins and 14 patients (33%) underwent re-excision for close margins of 2 mm or less. The 5 remaining patients underwent further surgery at our institution after physical examination raised concern for recurrence at the lumpectomy cavity (3 patients), or mammography was concerning for residual disease within the lumpectomy cavity (2 patients). Of note, 30 of 43 (70%) patients underwent their original excisional biopsy or BCT at an outside facility. After they were found to have positive or close margins on their final pathology evaluation, they were referred to our breast centers for definitive surgical management.

Diagnostic accuracy of MRI evaluation for residual disease Residual disease was found in 26 of 43 (60%) women undergoing re-excision. Among 43 women undergoing MRI before re-excision but after the original breast surgery, 29 had a MRI indicating residual disease. Twenty of these patients (69%) had residual disease confirmed on pathology and were classified as having a true-positive MRI (Table 2). MRI findings included rapidly enhancing and significant irregular, spiculated enhancement around the lumpectomy cavity or near the lumpectomy cavity (Fig. 1). Final histologic evaluations on the new margins were consistent with residual infiltrating ductal carcinoma, infiltrating lobular carcinoma, and DCIS (Table 3). The other 9 women (31%) with a MRI indicating residual disease were not found to have residual disease on final pathology and therefore were classified as having a falsepositive MRI (Table 2). In these women, the MRI findings

Table 2

Accuracy of MRI evaluation for residual disease

Overall analysis: 43 patients MRI indicated residual disease MRI did not indicate residual disease

Pathology showed residual disease

Pathology did not show residual disease

20

9

6

8

550

Figure 1 (arrow).

The American Journal of Surgery, Vol 198, No 4, October 2009

True-positive MRI evaluation for residual disease

were read as significant irregular enhancement and clumped nodular enhancement concerning for DCIS (Fig. 2). Histologic evaluations revealed 1 patient with atypia within the new specimen, 3 patients with fat necrosis, 4 patients with both normal breast tissue and foreign body giant cell reaction, and 1 patient with giant cell granulomatous inflammation at the lumpectomy cavity (Table 3). Among 14 women with a MRI read as negative for residual disease, 8 patients (57%) had no residual disease confirmed on final pathology at the time of re-excision and

Table 3

Figure 2 (arrow).

False-positive MRI evaluation for residual disease

were categorized as true negative (Table 2). The MRI reports of these patients did describe areas of progressive enhancement, but ultimately the radiologists determined that the findings were not consistent with residual disease (Fig. 3). Final histology review found patients with atypia, fat necrosis, normal breast tissue, foreign body reaction, and fibrocystic changes (Table 3). Finally, 6 of 14 women (43%) with a MRI read as negative for residual disease were found to have residual

Histologic findings at re-excision

Histology at re-excision True-positive category Invasive carcinoma only DCIS only Mixed invasive and in situ False-positive category Atypia Fat necrosis Normal breast tissue Proliferative changes Other True-negative category Atypia Fat necrosis Normal breast tissue Proliferative changes Other (in addition to above) False-negative category Invasive carcinoma only DCIS only Mixed invasive and in situ

Number of patients 5 10 5 1 3 4 0 1 2 1 3 0 2 2 4 0

Figure 3 (arrow).

True-negative MRI evaluation for residual disease

C.-C.H. Stucky et al.

MRI and residual disease

551

Comments

Figure 4 (arrow).

False-negative MRI evaluation for residual disease

invasive carcinoma or DCIS on final pathology. These patients were classified as false negative (Table 2). In these patients, the radiologist described early enhancing lesions or nodularity and mass-like enhancement (Fig. 4), but ultimately it was determined that the nonspecific pattern of the enhancement likely represented normal breast parenchyma. Histology review of the re-excision specimens showed invasive carcinoma (both ductal and lobular) as well as DCIS in these patients (Table 3). Overall, the sensitivity of MRI detection of residual disease in the lumpectomy cavity was 77% (95% confidence interval [CI], 56%–91%) and the specificity was 47% (95% CI, 23%–72%). The PPV was 69% (95% CI, 49%– 85%) and the NPV was 57% (95% CI, 29%– 82%). The accuracy was 65% (95% CI, 49%–79%).

Effect of time period on MRI accuracy In the 22 patients who underwent a MRI within 28 days (range, 6 –28 d) of their original surgery, the sensitivity of MRI in detecting residual disease was 85% (95% CI, 55%– 98%) and the specificity was 56% (95% CI, 21%– 86%). The PPV of MRIs conducted within 28 days of the original surgery was 73% (95% CI, 45%–92%) and the NPV was 71% (95% CI, 29%–96%). The accuracy at 28 days or less was 73% (95% CI, 50%– 89%). Twenty-one patients had a MRI more than 28 days (range, 29 – 425 d) after the original surgery. In this group, the sensitivity and specificity of MRI detecting residual disease was lower at 69% (95% CI, 39%– 91%) and 38% (95% CI, 9%–76%), respectively. The PPV and NPV also were lower at 64% (95% CI, 35%– 87%) and 43% (95% CI, 10%– 82%), respectively. Accuracy after 28 days decreased to 57% (95% CI, 34%–78%).

Local control is the mainstay of breast surgery. Patients who develop local recurrence have been shown to have worse overall survival.10 The presence of a positive margin presents a vexing problem for surgeons and patients alike. Unfortunately, rates of positive or close margins for invasive and/or in situ disease are reported to range from 20% to 60% of BCT cases8,9 and are decreased only minimally by various types of intraoperative margin assessment. Furthermore, re-excision lumpectomy or mastectomy identifies residual disease in only 30% to 63% of cases.8,17 Therefore, it is relatively common that no residual disease is found at re-excision and that the surgery was not necessary. Currently, no definitive criteria or tests exist to determine which patient might be spared a re-excision. Many studies have been published trying to determine preoperative and pathologic characteristics to identify which patients will have residual disease but none have been able to determine which patients can be spared surgery.17–19 The use of postlumpectomy mammography for patients whose disease presented with calcifications has been shown to be beneficial in determining if all the calcifications have been removed.20,21 If residual calcifications were noted on the mammogram, 64% to 75% were noted to have residual disease. Although the definitive method for determining the presence of residual disease at the lumpectomy cavity is histologic evaluation on re-excision, MRI may prove to be a useful tool in determining which patients may not need to undergo further surgery. Few existing studies have determined the utility of MRI in evaluating patients with positive or close margins after BCT or excisional biopsy.22–25 One study, published in 2003, found that of 82 patients undergoing re-excision for positive or close margins, 60% had residual in situ or invasive cancer at histologic examination.22 Similarly, we showed that 67% of patients in our study had residual disease. In the study by Lee et al,22 the sensitivity for MRI detection of residual disease was 61%, whereas our sensitivity calculated approximately 6 years later was 77%. This may be explained by improvements in MRI technique and more experience in interpreting these images. Lee et al22 also reported a specificity of nearly 70%, which was significantly higher than our specificity of only 47%. Our lower specificity may be the result of bias on the part of our radiologists who knew at the time of MRI that the patient had positive or close margins from the original surgery. The PPV reported in the 2003 study22 and in our study were similar at 75% and 69%, respectively. The NPV from 200322 and our current report also were similar at 55% and 57%, respectively. This precision rate is encouraging because it verifies the usefulness of MRI in diagnosing residual disease and it is noteworthy that it is consistent over time and among different institutions. However, our data show that routine MRI in the setting of close or positive margins must be used cautiously. There is still an almost 30% rate of false-positive radiologic findings in the lumpectomy site and a significant false-negative

552 rate primarily owing to residual DCIS. This suggests that significant overlap of imaging findings between benign postoperative changes and malignant disease exists. Therefore, a re-excision lumpectomy, regardless of MRI findings, still is recommended for close or positive margins. One study published in 200023 found that the greatest specificity of MRI evaluation of residual disease occurred when MRI was performed between the 28th and 35th postoperative days, whereas the accuracy of the postoperative MRI was significantly worse if performed before that time period. The investigators suggested that at least 28 days should elapse before a MRI should be attempted to determine if there is residual disease. In contrast to this study, our accuracy appeared higher if MRI was performed earlier. Because no time period has been shown to improve MRI accuracy consistently, we do not advocate performing MRI at a specific time after excision. One caveat to this would be in the premenopausal patient in whom timing of the menstrual cycle significantly impacts the accuracy of MRI; MRI should be performed in the first half of the menstrual cycle to maximize accuracy.26 Therefore, even though MRI may be helpful to assess the extent of residual disease, it cannot predict reliably which patients must undergo re-excision and those who may not. By contrast, if a large amount of residual disease is suspected, MRI is an excellent tool to confirm those suspicions with the caveat that pathologic confirmation should be performed before recommending mastectomy as the next treatment option. Unfortunately, further surgery still is required for those patients who are found to have a close or positive margin after lumpectomy.

References 1. Saslow D, Boetes C, Burke W, et al. American Cancer Society guidelines for breast screening with MRI as an adjunct to mammography. CA Cancer J Clin 2007;57:75– 89. 2. Singh V, Saunders C, Wylie L, et al. New diagnostic techniques for breast cancer detection. Futures Oncol 2008;4:501–13. 3. Sinha S, Sinha U. Recent advances in breast MRI and MRS. NMR Biomed 2009;22:3–16. 4. Cooney CS, Khouri NF, Tsangaris TN. The role of breast MRI in the management of patients with breast disease. Adv Surg 2008;42:299 – 312. 5. Fisher B, Jeong JH, Anderson S, et al. Twenty-five-year follow-up of a randomized trial comparing radical mastectomy, total mastectomy, and total mastectomy followed by irradiation. N Engl J Med 2002; 347:567–75. 6. Poggi MM, Danforth DN, Sciuto LC, et al. Eighteen-year results in the treatment of early breast carcinoma with mastectomy versus breast conservation therapy: the National Cancer Institute randomized trial. Cancer 2003;98:697–702.

The American Journal of Surgery, Vol 198, No 4, October 2009 7. Veronesi U, Cascinelli N, Mariani L, et al. Twenty-year follow-up of a randomized study comparing breast-conserving surgery with radical mastectomy for early breast cancer. N Engl J Med 2002;347:1227–32. 8. Wright MJ, Park J, Fey JV, et al. Perpendicular inked versus tangential shaved margins in breast-conserving surgery: does the method matter? J Am Coll Surg 2007;204:541–9. 9. Cabioglu N, Hunt KK, Sahin AA, et al. Role of intraoperative margin assessment in patients undergoing breast-conserving surgery. Ann Surg Oncol 2007;14:1458 –71. 10. Punglia RS, Morrow M, Winer EP, et al. Local therapy and survival in breast cancer. N Engl J Med 2007;356:2399 – 405. 11. Smitt MC, Nowels K, Carlson RW, et al. Predictors of reexcision findings and recurrence after breast conservation. Int J Radiat Oncol Biol Phys 2003;57:979 – 85. 12. Tartter PI, Kaplan J, Bleweiss I, et al. Lumpectomy margins, reexcision, and local recurrence of breast cancer. Am J Surg 2000;179:81–5. 13. Wazer DE, Schmidt-Ulrich RK, Schmid CH, et al. The value of breast lumpectomy margin assessment as a predictor of residual tumor burden. Int J Radiat Oncol Biol Phys 1997;38:291–9. 14. Singletary SE. Surgical margins in patients with early-stage breast cancer treated with breast conserving therapy. Am J Surg 2002;184: 383–93. 15. Sabel MS, Rogers K, Griffith K, et al. Residual disease after reexcision lumpectomy for close margins. J Surg Oncol 2009;99:99 – 103. 16. Aziz D, Rawlinson E, Narod SA, et al. The role of reexcision for positive margins in optimizing local disease control after breast-conserving surgery for cancer. Breast J 2006;12:331–7. 17. Jardines L, Fowble B, Schultz D, et al. Factors associated with a positive reexcision after excisional biopsy for invasive breast cancer. Surgery 1995;118:803–9. 18. Schnall MD, Blume J, Bluemke DA, et al. Diagnostic architectural and dynamic features at breast MR imaging: multicenter study. Radiology 2006;238:2–53. 19. Preda L, Villa G, Rizzo S, et al. Magnetic resonance mammography in the evaluation of recurrence at the prior lumpectomy site after conservative surgery and radiotherapy. Breast Cancer Res 2006;8:R53. 20. Aref A, Youssef E, Washington T, et al. The value of postlumpectomy mammogram in the management of breast cancer patients presenting with suspicious microcalcifications. Cancer J Sci Am 2000;6:25–7. 21. Waddell BE, Stomper PC, DeFazio JL, et al. Postexcision mammography is indicated after resection of ductal carcinoma in-situ of the breast. Ann Surg Oncol 2000;7:665– 8. 22. Lee JM, Orel SG, Czerniecki BJ, et al. MRI before reexcision surgery in patients with breast cancer. AJR Am J Roentgenol 2004;182: 473– 80. 23. Frei KA, Kinkel K, Bonel HM, et al. MR imaging of the breast in patients with positive margins after lumpectomy: influence of the time interval between lumpectomy and MR imaging. AJR Am J Roentgenol 2000;175:1577– 84. 24. Soderstrom CE, Harms SE, Farrell RS, et al. Detection with MR imaging of residual tumor in the breast soon after surgery. AJR Am J Roentgenol 1997;168:485– 8. 25. Orel SG, Reynolds C, Schnall MD, et al. Breast carcinoma: MR imaging before re-excisional biopsy. Radiology 1997;205:429 –36. 26. Delille JP, Slanetz PJ, Yeh ED, et al. Physiologic changes in breast magnetic resonance imaging during the menstrual cycle: perfusion imaging, signal enhancement, and influence of the T1 relaxation time of breast tissue. Breast J 2005;11:236 – 41.