Local control in ductal carcinoma in situ treated by excision alone: incremental benefit of larger margins

The American Journal of Surgery 190 (2005) 521 525 George Peter s Award Winner Local control in ductal carcinoma in situ treated by excision alone: incremental benefit of larger margins Heather R. MacDonald, M.D., Melvin J. Silverstein, M.D.*, Helen Mabry, M.D., Brenda Moorthy, M.D., Wei Ye, M.S., Melinda S. Epstein, Ph.D., Dennis Holmes, M.D., Howard Silberman, M.D., Michael Lagios, M.D. USC Norris Cancer Center, Keck School of Medicine, University of Southern California, 1441 Eastlake Ave., Room 7415, Los Angeles, CA 90033-0800, USA Manuscript received June 5, 2005; revised manuscript June 10, 2005 Presented at the Sixth Annual Meeting of the American Society of Breast Surgeons, Los Angeles, California, March 16 20, 2005 Abstract Background: Margin width has been shown previously to be the most important predictor of local treatment failure after breast conservation for ductal carcinoma in situ (DCIS). Methods: Five variables thought to be associated with local recurrence were evaluated by univariate and multivariate analysis in 455 nonrandomized patients with DCIS treated with excision alone. Results: Multivariate analysis showed that margin width, age, nuclear grade, and tumor size all were independent predictors of local recurrence, with margin width as the single most important predictor. After adjusting for all other predictors the likelihood of local recurrence for patients with margins less than 10 mm was 5.39 times as much as that for patients with margins of 10 mm or more (95% confidence interval, 2.68 10.64). Conclusions: Margin width, the distance between DCIS and the closest inked margin, was the single most important predictor of local recurrence. As margin width increases, the risk for local recurrence decreases. 2005 Excerpta Medica Inc. All rights reserved. Keywords: Ductal carcinoma in situ; Margin width; Local recurrence risk; Treatment failure Approximately half of all local treatment failures after breast conservation therapy for ductal carcinoma in situ (DCIS) are invasive cancer and therefore a potential threat to life [1 4]. Preventing local recurrence is of paramount importance. Numerous factors including margin width, nuclear grade, presence of comedo necrosis, age, and treatment have been shown to influence the probability of local recurrence. In patients treated with excision alone, margin width was shown previously to exert a greater effect on risk for recurrence than nuclear grade, histologic type, patient age, or tumor size [5,6]. Previous prospective randomized trials that investigated the risk for local recurrence after conservative therapy for DCIS did not collect margin width measurements routinely in a standardized prospective fashion [7 11]. For example, the European Organization for Research and Treatment (EORTC) only recorded exact margin width in 5% of pathology reports. The National Surgical Adjuvant Breast and Bowel Project (NSABP) has not published any data regarding margin width. Margin width cannot be determined retrospectively if the specimens were not inked, completely sampled, or sequentially submitted. This study stratified margin widths in a group of patients with DCIS who were treated by excision alone, and calculated the relationship of margin width to the risk for local recurrence. The goal was to determine the effect of increasing margin width on local treatment failure. Methods * Corresponding author. Tel.: 1-323-865-3535; fax: 1-323-865-3539. E-mail address: melsilver9@aol.com From 1972 to 2004, 445 patients with pure DCIS were treated with excision alone. The surgical margins were mea- 0002-9610/05/$ see front matter 2005 Excerpta Medica Inc. All rights reserved. doi:10.1016/j.amjsurg.2005.06.005

522 H.R. MacDonald et al. / The American Journal of Surgery 190 (2005) 521 525 sured in every specimen. An effort was made to excise all suspicious lesions completely and to process all excised tissue. Needle localization, intraoperative specimen radiography, and correlation with the preoperative mammogram were performed for all cases involving nonpalpable lesions (95% of cases). The margins of the specimen were inked or dyed and serially sectioned at 2- to 3-mm intervals. Tissue sections were arranged and processed in sequence. Pathologic evaluation included histologic subtype, nuclear grade, the presence or absence of comedo necrosis, the maximum size of the lesion, and margin width. Margin width was determined by direct measurement or ocular micrometry. The closest single distance between DCIS and an inked margin was reported. Tumors were divided into groups by margin width: 0 mm (tumor transected), 0.1 to 0.9 mm, 1.0 to 1.9 mm, 2.0 to 2.9 mm, 3.0 to 5.9 mm, 6.0 to 9.9 mm, and 10 mm or greater. Patients who underwent re-excision, and in whom no additional DCIS was found, were scored as having 10-mm margins or greater. Tumors were divided into 3 groups by nuclear grade: nuclear grade 1 low; nuclear grade 2 intermediate; nuclear grade 3 high. Comedo necrosis was considered present for any architectural pattern of DCIS in which a central zone of necrotic debris with karyorrhexis was identified. Tumors were divided into 2 groups by the presence or absence of any amount of comedo necrosis. Age was analyzed and data were separated into 3 groups: 39 years and younger, 40 to 60 years, and 61 years and older. All patients with DCIS and known margin widths were included in the study. The outcome studied was the time to local recurrence, which was calculated from the date of diagnosis to the date of local recurrence. Because of the difficulty of accurately differentiating between a true local recurrence and a new cancer in another quadrant of the breast, all ipsilateral breast cancer events were scored as local recurrences. Patients who did not experience a local recurrence were censored at the date of their last follow-up evaluation or death. Kaplan-Meier estimates of the probabilities of remaining free of local recurrence were calculated at 5 and 8 years. The Greenwood formula was used to calculate the standard errors. Both univariate and multivariate Cox regression analyses were performed for surgical margin, age at diagnosis, tumor size, nuclear grade, and the presence of comedo necrosis. The proportional hazard assumption was checked using Schoenfeld residuals. The partial likelihood ratio test based on the Cox model was used to calculate P values (all 2-sided). Results A total of 445 patients were treated for DCIS with excision alone. Seventy-nine (17%) patients experienced local recurrences, of which 26 (33%) were invasive. The median follow-up period was 57 months. The median tumor size was 10 mm. Among patients who experienced local recurrences, the median time to local recurrence was 26 months. There were 79 ipsilateral breast cancer events after initial treatment, 72 of which were in the same quadrant as the original lesion (91%). One patient developed metastatic breast cancer and died of cancer. Twenty-three additional patients died from non breast cancer related causes. Table 1 shows a univariate analysis of local recurrence risk as affected by margin width, high nuclear grade, younger age, greater tumor size, and the presence of comedo necrosis. All were associated significantly with greater risk for developing local recurrence (P.001). Only 9 of 197 (4.6%) patients with a greater than 10-mm margin experienced local failure. For patients with 10-mm or greater margins, the probability of remaining free of local recurrence at 5 years was 93%. For patients with 1.0- to 1.9-mm margins it was 73%. For patients with transected margins it was 48%. The relative risk of developing a local recurrence stratified by surgical margin was plotted as a continuous variable (Fig. 1). A clear trend emerged: as margin width increased, the hazard ratio for local failure decreased (Fig. 1). Table 1 breaks margin width into 7 groups and, overall, the risk for developing local recurrence decreased as the margin width increased. Table 2 shows the mean tumor size for the 7 different margin widths. Patients with 3.0- to 5.9-mm margins had a higher relative risk than those with 2.0- to 2.9-mm margins, possibly owing to their larger mean tumor size (19.7 vs 9.1 mm) (Table 2) and small numbers of patients in each group. Multivariate analysis showed that the margin width had the smallest P value and the greatest hazard ratio and was thus the most significant predictor of local recurrence (Table 3). Age at diagnosis, nuclear grade, and tumor size also remained significant predictors of local recurrence by multivariate analysis although the presence of necrosis no longer was significant. The effect of margin width was independent of age at diagnosis, nuclear grade, and tumor size. After adjusting for all other predictors the likelihood of local recurrence for patients with margins less than 10 mm was 5.39 times as much as that for patients with margins of 10 mm or more (95% confidence interval, 2.68 10.64). A comparison of hazard ratios of tumor predictors also shows the dominant effect of margin width on local recurrence risk (Table 4). The hazard ratio for developing local recurrence if margins were transected was 7.69 times as great as the risk if margin widths were 10 mm or greater. The hazard ratios for other predictors were less: for age at diagnosis of less than 40 years versus older than 60 years the relative hazard ratio was 4.17. For tumor with nuclear grade III versus grades I or II the relative hazard ratio was 3.44. For tumor size at 40 mm versus 0 mm the relative hazard ratio was 2.81.

H.R. MacDonald et al. / The American Journal of Surgery 190 (2005) 521 525 523 Table 1 Univariate analysis of predictors of local recurrence of DCIS Total N/no. of events Relative hazard ratio (95% confidence interval) P value Probability of remaining free of local recurrence at 5 years ( SE) Probability of remaining free of local recurrence at 8 years ( SE) Margin.001 0 (transected) 32/15 1.00 48% 11% 39% 12%.1.9 53/18.61 (.31 1.20) 63% 8% 58% 9% 1.0 1.9 20/7.58 (.23 1.42) 73% 12% 49% 16% 2.0 2.9 82/20.21 (.10.42) 81% 5% 78% 5% 3.0 5.9 39/8.35 (.15.83) 64% 11% 64% 11% 6.0 9.9 22/2.20 (.05.87) 91% 9% 61% 25% 10 197/9.07 (.03.15) 93% 3% 91% 3% Age (y).001 40 28/12 1.00 60% 11% 47% 12% 40 60 306/59.46 (.25.86) 80% 3% 74% 3% 60 111/8.18 (.07.43) 90% 4% 90% 4% Nuclear grade.001 I/II 258/20 1.00 93% 2% 87% 3% III 187/59 4.85 (2.92 8.06) 64% 4% 59% 5% Tumor size (mm).001 5 111/11 1.00 90% 4% 86% 5% 5 10 114/17 1.41 (.67 2.99) 89% 3% 80% 5% 10 15 67/12 2.05 (.91 4.61) 78% 7% 78% 7% 15 20 55/13 3.88 (1.73 8.70) 63% 9% 63% 9% 20 30 44/10 2.83 (1.20 6.69) 73% 8% 68% 9% 30 50 35/10 4.62 (1.95 10.93) 66% 11% 55% 14% 50 19/6 7.92 (2.81 22.28) 49% 17% Necrosis.001 No 157/13 1.00 92% 3% 89% 3% Yes 288/66 3.81 (2.10 6.93) 73% 3% 66% 4% Comments The treatment of DCIS with excision alone has psychologic and physical advantages for the patient over mastectomy or excision plus radiotherapy. However, excision alone is accompanied by an increase in risk for local treatment failure [1,8,12]. In other studies, approximately 50% of local recurrences are invasive cancer [1 4]. In our dataset, 33% of local recurrences were invasive cancer. Although no randomized trial has shown any survival benefit for more aggressive treatment, preventing a local invasive recurrence is of great importance. When an invasive recurrence occurs an almost completely curable noninvasive lesion has advanced to a potentially less curable form with a higher potential mortality rate. An analysis of our data show that as the margin width increased, the probability of local recurrence decreased. This trend was not affected by patient age or nuclear grade. Tumor size may exert a confounding effect on the local recurrence risk for patients with 2.0- to 2.9-mm and 3.0- to 5.9-mm margin widths because the former group had smaller tumors and lower hazard ratios than the latter. This represents a reversal of the overall trend of increasing margin width and decreasing local recurrence risk. Multivariate analysis provides a mathematic model that establishes both tumor size and margin width as independent predictors of local recurrence risk. However, in vivo, neither factor exists alone. Both factors exert their influence in all patients. Small numbers of patients in each margin-width stratification are another possible confounding factor. If wide margins are obtained, regardless of other factors, the probability of local recurrence remains small. With greater than 10-mm margins, the probability of remaining recurrence free at 8 years is greater than 90% without postoperative radiotherapy. With less than 1-mm margins the probability of remaining recurrence free at 8 years with excision alone decreases to 58%. Simply defining margins as nontransected allows a broad range of margin widths to be considered acceptable or clear. Clear but very close margins are associated with an increased probability of local recurrence. Additionally, the risk for experiencing a local recurrence is 7 times greater for patients with transected margins than for those with greater than 10-mm margins. The most likely cause of local recurrence after excision alone for DCIS is inadequate surgery resulting in residual disease. Most local recurrences after breast conservation therapy for DCIS occur at or near the primary lesion. In our series, 91% of local recurrences after excision alone were in the same quadrant as the original lesion. Residual disease appears to be best predicted by margin width. Wide margin width increases the probability of complete excision. In previously reported data from 181 patients with DCIS treated with wide excision, 43% of patients with clear margins ( 1 mm from the inked margin) had residual disease

524 H.R. MacDonald et al. / The American Journal of Surgery 190 (2005) 521 525 Fig. 1. Hazard ratio of local recurrence versus margin width. The solid line is the relative hazard ratio function calculated using the Cox regression model based on log transformed (margin width 1). The open circles are the categorized relative hazard ratios calculated using the Cox regression model based on categorized margin-width groups. Both used transected margin (margin width 0 mm) as the reference. The categorized relative hazard ratios are plotted against the mean margin width in each margin-width group using the equation hazard ratio (margin width 1) 1.02. Table 2 Tumor size stratified by margin width in patients with DCIS Margin (mm) N Mean tumor size (mm) 0 (transected) 32 32.3.1.9 53 23.1 1.0 1.9 20 12.9 2.0 2.9 82 9.1 3.0 5.9 39 19.7 6.0 9.9 22 25.3 10 197 12.9 on re-excision or mastectomy [13]. These patients would have had clear margins if judged by NSABP standards (tumor not transected). Approximately 76% of patients with DCIS less than 1 mm from the inked original margins had residual disease on re-excision or mastectomy. In 2002, Neuschatz et al [14] analyzed re-excision specimens of 253 patients treated with lumpectomy for DCIS and re-excised for margins of 2 mm or less per institutional policy. Sixtythree percent of patients with transected margins had residual DCIS at re-excision. Forty-one percent of patients with greater than 0- to 1-mm margins had residual DCIS at re-excision. Thirty-one patients with greater than 1- to 2-mm margins had residual DCIS at re-excision. The NSABP B-17 trial did not require marking of margins, complete tissue processing, specimen radiography, or margin-width measurement. In their initial report, 40% of the patients did not have a size measurement [7]. The lack of these data makes it impossible for the NSABP study to determine size and margin width criteria for optimal excision. The EORTC and UK DCIS trials also were not de- Table 3 Multivariate analysis of predictors of local recurrence of DCIS Variable Relative hazard ratio Chi-square Degree of freedom P value Log transformed.42 (.32.56)* 40.3 1.00001 margin Age (y) 13.9 2.00096 40 1.00 40 60.71 (.38 1.35) 60.24 (.09.58) Nuclear grade 15.8 1.0007 I/II 1.00 III 3.44 (1.74 6.79) Log transformed 1.21 (1.10 1.34) 3.8 1.05 tumor size Necrosis.12 1.72 No 1.00 Yes 1.16 (.52 2.59) * Hazard ratio for each unit increase in natural log (margin width in mm 1). Hazard ratio for increase in tumor size by 100%.

H.R. MacDonald et al. / The American Journal of Surgery 190 (2005) 521 525 525 Table 4 Relative hazard ratios of local recurrence of predictors Variable Margin 0 vs 10 mm 7.69 Age 40 vs 60 y 4.17 Nuclear grade III vs I/II 3.44 Tumor size 40 vs 1 mm 2.81 signed to identify subgroups at high risk for local treatment failure [9 11]. The study designs of all the previously randomized trials proved the benefit of postoperative radiotherapy across a wide range of patients but they cannot be used to select groups of patients who are at minimal risk for local recurrence without postoperative radiation therapy. Further studies such as the one presented here are needed to refine surgical goals for optimally excising DCIS when treated conservatively. Margin width is the single most important factor in predicting local recurrence after excision alone for DCIS. As margin width increases, the probability of local recurrence decreases. A high rate of local control can be achieved when the margin width is 10 mm or greater, regardless of high nuclear grade, large tumor size, or young patient age. References Hazard ratio [1] Silverstein MJ, Barth A, Poller DN, et al. Ten-year results comparing mastectomy to excision and radiation therapy for ductal carcinoma in situ of the breast. Eur J Cancer 1995;31:1425 7. [2] Silverstein MJ, Lagios MD, Martino S, et al. Outcome after local recurrence in patients with ductal carcinoma in situ of the breast. J Clin Oncol 1998;16:1367 73. [3] Lagios MD, Margolin FR, Westdahl PR, et al. Mammographically detected ductal carcinoma in situ. Frequency of local recurrence following tylectomy and prognostic effect of nuclear grade on local recurrence. Cancer 1989;63:618 24. [4] Solin LJ, Kurtz J, Fourquet A, et al. Fifteen-year results of breast conserving surgery and definitive breast irradiation for the treatment of ductal carcinoma in situ of the breast. J Clin Oncol 1996;14:754 63. [5] Silverstein MJ, Lagios MD, Groshen S, et al. The influence of margin width on local control of ductal carcinoma in situ of the breast. N Engl J Med 1999;340:1455 61. [6] Boland GP, Chan KC, Knox WF, et al. Value of the Van Nuys Prognostic Index in prediction of recurrence of ductal carcinoma in situ after breast-conserving surgery. Br J Surg 2003;90:426 32. [7] Fisher B, Constantino J, Redmond C, et al. Lumpectomy compared with lumpectomy and radiation therapy for the treatment of intraductal breast cancer. N Engl J Med 1993;328:1581 6. [8] Fisher ER, Dignam J, Tan-Chiu E, et al. Pathologic finding from the National Surgical Adjuvant Breast Project (NSABP) eight-year update of protocol B-17. Cancer 1999;86:429 38. [9] Julien JP, Bijker N, Fentiman I, et al. Radiotherapy in breast conserving treatment for ductal carcinoma in situ: first results of EORTC randomized phase III trial 10853. Lancet 2000;355:528 33. [10] Bijker N, Peterse JL, Duchateau L, et al. Risk factors for recurrence and metastasis after breast conserving therapy for ductal carcinoma in situ: analysis of European Organization for Research and Treatment of Cancer Trial 10853. J Clin Oncol 2001;19:2263 71. [11] Houghton J, George WD, Cuzick J, et al. Radiotherapy and tamoxifen in women with completely excised ductal carcinoma in situ of the breast in the UK, Australia, and New Zealand: randomized controlled trial. Lancet 2003;362:95 102. [12] Silverstein MJ, Poller DN, Waisman JR, et al. Prognostic classification of breast ductal carcinoma in situ. Lancet 1995;345:1154 7. [13] Silverstein MJ, Gierson ED, Colburn WJ, et al. Can intraductal breast carcinoma be excised completely by local excision. Cancer 1994;73: 2985 9. [14] Neuschatz AC, DiPetrillo T, Safaii H, et al. The value of breast lumpectomy margin assessment as a predictor of residual tumor burden in ductal carcinoma in situ of the breast. Cancer 2002;94: 1917 24.