Use of Predictors of Recurrence to Plan Therapy for DCIS of ......judgment was weak, and we had shifted the burden of that judgment to her."[1] That paragraph was written in 1991 about

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

Use of Predictors of Recurrence to Plan Therapy for DCIS of theBreastReview Article [1] | March 01, 1997 | Breast Cancer [2]By Melvin J. Silverstein, MD [3] and Michael D. Lagios, MD [4]

Despite the results of the National Surgical Adjuvant Breast and Bowel Project B-17, there continuesto be debate regarding the most appropriate treatment for patients with ductal carcinoma in situ(DCIS) of the breast. Numerous clinical, pathologic, and laboratory factors can aid clinicians andpatients wrestling with the difficult

Introduction

"Her agony came from the fact that mastectomy would be curative and it was hard to turn thatdown. A lesser procedure, while preserving her breast and her femininity, offered her somewhat lesschance for a complete cure--but exactly how much less was unknown. Perhaps only a small amountless. It didn't seem worth losing her breast for a few percentage points. Yet, maybe it was. It was themost difficult decision of her life. But medicine had failed her. The data upon which to base herjudgment was weak, and we had shifted the burden of that judgment to her."[1]

That paragraph was written in 1991 about a woman with ductal carcinoma in situ (DCIS) of thebreast and her arduous journey through the medical system as she searched for the "correct"treatment. There were a number of "correct" treatments then for her particular form of DCIS, buteach was flawed in one way or another, confounding her thoughts and making her decision moredifficult. But that was 1991. Today, we know much more about DCIS. But is the decision-makingprocess any easier?The results of the National Surgical Adjuvant Breast Project (NSABP) Protocol B-17 were published in1993[2] and updated in 1995[3] and 1997[3a]. This prospective, randomized clinical trial wasdesigned to resolve the controversy over the treatment of DCIS. More than 800 patients with DCISexcised with clear surgical margins were randomized to one of two treatments: excision only orexcision plus radiation therapy. At 5 years, there was a statistically significant decrease in localrecurrence of both DCIS and invasive breast cancer in patients treated with radiation therapy. Thesedata led the NSABP to recommend postexcision radiation therapy for all patients with DCIS whochose to save their breasts--a recommendation that some consider too broad.[4,5]The NSABP B-17 study was criticized for a number of reasons, including its definition of clear margins(which the NSABP defined as a tumor that is not transected), determination of size by central reviewof the pathology report, the absence of size measurements for more than 40% of cases, and,perhaps most important, the lack of pathologic subset analysis in the initial report.[4,5] In defense ofthe NSABP, the trial did exactly what it was designed to do; namely, it proved that radiation therapywas effective for patients with DCIS. It was not designed to answer the questions about patientsubgroups that we ask today.Consider the following two patients, both of whom merit radiation therapy based on the results ofNSABP B-17. The first patient is a woman with a 12-mm low-grade lesion that has been widelyexcised with a minimum of 15-mm margins in all directions. Compare her with the second patient, awoman with a 35-mm high-grade lesion with DCIS approaching to within 0.2 mm of the inked marginbut not involving it. According to the NSABP, both of these patients should be treated with radiationtherapy.At our facilities, based on data that will be presented below, the first patient would receive noadditional therapy. Rather, she would be carefully followed with physical examination every 6months and mammography every 6 to 12 months. The second patient would undergo a widereexcision prior to making a final treatment decision. Significant residual disease approaching thenew margins would earn a recommendation for mastectomy and immediate reconstruction; widelyclear new margins with little or no residual DCIS would warrant consideration for radiation therapy.Thus, despite the results of NSABP B-17, there continues to be debate regarding the DCISdecision-making process, which is not much clearer now than it was in 1991.

Page 1 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

Numerous clinical, pathologic, and laboratory factors can aid clinicians and patients wrestling withthe difficult treatment decision-making process. Our research has shown that nuclear grade, thepresence of comedo-type necrosis (coagulative necrosis), tumor size, and margin width are all keyfactors in predicting local recurrence in patients with DCIS.[6-8] By using a combination of thesefactors, it may be possible to identify subgroups of patients who do not require irradiation, if breastconservation is elected; it also may be possible to identify patients whose recurrence rate ispotentially so high, even with breast irradiation, that mastectomy is preferable.

Changing Patterns of Disease

Ductal carcinoma in situ is a biologically and histologically heterogeneous group of lesions.[9,10]With the appreciation and acceptance of this heterogeneity, DCIS has become confusing for bothpatients and physicians. Currently, it is not uncommon for DCIS patients to seek second, third, andeven fourth opinions and to receive a diverse spectrum of advice ranging from biopsy only to wideexcision, segmental resection, quadrant resection, mastectomy, or even bilateral mastectomy. As anadjunct to all these treatments except mastectomy, radiation therapy may be advised.Patients seeking treatment advice will find physicians willing to support most of these options.[1]The second opinion-givers are usually oncologists specializing in medicine, surgery, or radiationtherapy. Some patients, however, seek advice from their gynecologists, internists, or familypractitioners. Many women also turn for counsel to family, friends, and other women who have hadbreast cancer, most of whom have had invasive disease.Table 1 shows the changing nature of DCIS during the last decade. Before the widespread use ofmammography, DCIS was diagnosed infrequently, representing less than 1% of all breast cancercases.[11,12] Today, DCIS is common, accounting for approximately 12% to 15% of all newlydiagnosed cases[13] and as many as 20% to 40% of cases at institutions that effectively utilizemammography.[14,15] In 1997, more than 36,000 new cases of DCIS are expected to be diagnosedin the United States.[16]Previously, most patients with DCIS presented with clinical symptoms, such as a breast mass, bloodyor serous nipple discharge, or Paget's disease and frequently had extensive disease.[11,17-19]Today, most lesions are smaller, nonpalpable, subclinical, and detected by mammography alone.Until recently, the treatment for most patients with DCIS was mastectomy. Currently, many patientsare being treated with breast preservation. Fifteen years ago, when mastectomy was common,reconstruction was infrequent and, if performed, was generally done as a delayed procedure withimplants. Today, reconstruction for patients with DCIS treated by mastectomy is common and isusually done immediately, at the time of mastectomy, and often with autologous tissue.In the past, when a mastectomy was performed, large amounts of skin were discarded. Now, it isconsidered safe to perform a skin-sparing mastectomy for DCIS.[20-23] We must keep in mind,however, that in patients with extensive disease, recurrences may develop after mastectomy (withor without reconstruction) in the scant residual breast tissue. The thicker the skin flaps, the moreresidual breast tissue is left behind and the more likely there is to be a recurrence.In the past, there was no confusion. All breast cancers were considered the same and mastectomywas the only treatment. Today, we know that all breast cancers are different. There are manytreatments and a great deal of confusion.Factors Responsible for the Changes These changes were brought about by numerous factors. The most important of these are increasedutilization of mammography, improvements in mammographic technique, and the acceptance ofbreast-conservation therapy for invasive breast cancer.Mammography--The acceptance of mammography not only changed the way we detect DCIS, italso altered the nature of the disease that we detected by allowing us to enter the neoplasticcontinuum at an earlier time. Every institution employing mammography has witnessed a relativelylarge increase in the number of small, mammographically detected cases of DCIS. This can beappreciated by charting the impact that mammography has had on the number and type of DCIScases at one of our facilities, the Breast Center in Van Nuys, California.[24]From 1979 to 1981, the Van Nuys group treated a total of only 15 patients with DCIS, an average of5 per year. Only two lesions (13%) were nonpalpable. Two new mammography units and a full-time,experienced mammographer were added in 1982, and immediately the number of new DCIS casesincreased to more than 30 per year, most of them nonpalpable. With the addition of a thirdmammography machine in 1987, almost 40 new cases per year were diagnosed. In 1994, a fourthmammography machine and a stereotactic biopsy unit were added. Analysis of the Van Nuys series

Page 2 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

through June 1996 (more than 500 patients) revealed that 81% of lesions were nonpalpable. If weconsider only those lesions that were diagnosed after 1991, 92% were nonpalpable.Breast Conservation--The second factor that affected how we think about DCIS was theacceptance of breast-conservation therapy (lumpectomy, axillary node dissection, and radiationtherapy) for patients with invasive breast cancer. Until 1980, the treatment for most patients withany form of breast cancer was mastectomy. Since then, numerous prospective randomized trialshave revealed that survival in patients with invasive breast cancer treated with lumpectomy andradiation therapy is equivalent to that in women who undergo mastectomy.[25-32] Based on theseresults, it was difficult to continue treating noninvasive disease with mastectomy while treating moreaggressive invasive breast cancer with breast preservation.Patients often ask the question, "You mean if I waited until my cancer was invasive, I could havesaved my breast?" The answer is not that simple. Although there is clearly a relationship betweenDCIS and invasive breast cancer, the two entities are different heterogeneous groups of diseaseswith some overlap. It is extremely common to see both DCIS and invasive breast cancer within asingle specimen. Authorities agree that DCIS is an obligate precursor to invasive breast cancer, butthere is speculation that DCIS may be less amenable to control with irradiation. Thus, while patientswith invasive breast cancers that are 4 cm or smaller and have little or no intraductal component canreadily be treated by lumpectomy and radiation therapy, the same may not be true for patients withpure DCIS or for those with invasive breast cancer with an extensive intraductal component.Nevertheless, current data suggest that many patients with DCIS can be successfully treated withbreast preservation (with or without radiation therapy). In the sections that follow, we will show howeasily available data can be used to predict which patients are more likely to suffer a recurrenceafter breast conservation. Knowing the probability of local recurrence can help simplify the complextreatment selection process.

Classifying DCIS

There is no single universally accepted histopathologic classification for ductal carcinoma in situ.Tumor Architecture Pathologists traditionally divide DCIS into five or six architectural subtypes (papillary, micropapillary,cribriform, clinging, solid, and comedo), often grouping the first five together as noncomedo lesionsand comparing them with comedo DCIS.[14,33,34] Comedo DCIS is frequently associated with highnuclear grade,[14,33-35] aneuploidy,[36] a higher proliferation rate,[37] HER-2/neu (c-erb-B2) geneamplification or protein overexpression,[38-43] and more clinically aggressive behavior.[8,44-46]Noncomedo lesions tend to have the opposite features. However, such a division by architecture,comedo vs noncomedo, is an oversimplification and has not been shown to correlate with outcome inall cases.Any architectural subtype may present with any nuclear grade with or without comedo-type necrosis.It is not uncommon for high-nuclear-grade noncomedo lesions to express markers similar tohigh-grade comedo lesions. Such lesions may require more aggressive treatment.Furthermore, mixtures of various architectural subtypes within a single biopsy specimen arecommon. In our series, 68% of all lesions had significant amounts of two or more architecturalsubtypes (Figure 1). Adding to the confusion, there is no uniform agreement among pathologists onexactly how much comedo DCIS needs to be present to consider the lesion a comedo DCIS.Architecture, without a strict set of universally agreed upon criteria (which currently do not exist), isa poor way to classify DCIS.Nuclear Grade Nuclear grade is a more dependable biologic indicator than architecture and has emerged as a keyhistopathologic factor for identifying aggressive tumor behavior.[8,11,33,34,46-49] In a multivariateanalysis of the Van Nuys series that was limited to DCIS patients treated with excision plus radiationtherapy, nuclear grade was the only significant factor that predicted for local recurrence of both DCISand invasive breast cancer.[7]This result led us to perform a more detailed analysis of all patients in our series who underwentbreast preservation (excision alone or excision plus radiation therapy).[6,50] We analyzed 15prognostic factors by univariate analysis (log-rank test). All statistically significant predictors of localrecurrence by univariate analysis were then evaluated using a Cox multivariate regression analysiswith backward elimination.As shown in Table 2, six factors--nuclear grade, margin width, tumor size, presence of necrosis,comedo architecture, and HER-2/neu overexpression--were significant predictors of local recurrence

Page 3 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

on univariate analysis, but only three--nuclear grade, tumor size, and margin width--remainedsignificant on multivariate analysis. The presence of comedo-type necrosis also approachedsignificance on multivariate analysis (P = .09). The remaining nine factors (estrogen receptor,progesterone receptor, S-phase, ploidy, p53, microcalcifications, palpability, year of diagnosis, andage) were not significant predictors of local recurrence by univariate analysis.Van Nuys Pathologic Classification With this in mind, in May 1995, the Van Nuys group introduced a new pathologic classification ofDCIS, the Van Nuys Pathologic Classification, based on two statistically important predictors of localrecurrence: the presence or absence of high nuclear grade and comedo-type necrosis.[49] Webelieve that both of these factors reflect tumor biology.To use the Van Nuys Pathologic Classification (Figure 2), all high-grade lesions, regardless of thepresence or absence of comedo-type necrosis, are placed into the worst prognostic group (group 3)(Figures 3a and 3b). The remaining non-high grade lesions (nuclear grades 1 or 2) are thenseparated by the presence of any amount of comedo-type necrosis (group 2; Figures 4a-4c) or theabsence of comedo necrosis (group 1; Figures 5a and 5b). This results in three easily identifiablegroups with significantly different outcomes as measured by local tumor recurrence (Figure 6).High nuclear grade was chosen as the most important factor in the Van Nuys classification becauseof the results of our multivariate analysis (Table 2) and because there is general agreement that therecurrence rate after breast conservation is higher in patients with high-grade lesions than in thosepatients with low-grade lesions.[8,11,30,34,47,39,51] Comedo-type necrosis was chosen because itspresence also suggests a poor prognosis.[14,44-46,52,53] Both high nuclear grade and comedo-typenecrosis are easily and consistently recognized by practicing pathologists.[54]In our pathologic classification, no requirement is made for a minimum or specific amount ofhigh-grade DCIS or comedo-type necrosis. Occasional desquamated or individually necrotic cells areignored and are not scored as comedo-type necrosis. Since publication, the reproducibility of the VanNuys Pathologic Classification, as compared with other proposed classification systems, has beenindependently confirmed by Gupta and colleagues.[55]The most difficult part of nuclear grading is the intermediate-grade lesion. The subtleties of theintermediate-grade lesion are not important to our classification; only nuclear grade 3 needs to berecognized. This is a fairly straightforward task for most pathologists.[49] The cells must be largeand pleomorphic, lack architectural differentiation and polarity, have prominent nucleoli and coarseclumped chromatin, and generally show mitoses[14,33,34,52] (Figures 3a and 3b). This pathologicclassification, when combined with tumor size and margin width, is an integral part of the Van NuysPrognostic Index, a system that will be explained in detail below.

Which Treatment Option?

For most patients who have DCIS, there is no single correct treatment approach. Patients generallyhave to make a choice among several options--a decision that, although seemingly simple, is not. Asthe treatment alternatives increase and become more complicated, frustration will increase for bothpatient and physician.[1]There is no easy way to tell a patient that she has breast cancer. But is DCIS really breast cancer?When we think of cancer, we think of a disease that, if untreated, can run an unrelenting coursetoward death. That is certainly not the case with DCIS. The cancer phenotype consists of at least fivefactors: unlimited growth, genomic elasticity (resistance to treatment), angiogenesis, invasion, andmetastasis.[56,57] Ductal carcinoma in situ lacks the latter two characteristics. In all likelihood, whenwe understand why some DCIS lesions develop the ability to invade and metastasize and why othersdo not, we will have opened the door to a far better understanding of the neoplastic process.When counseling a patient with DCIS, one must emphasize that she has a borderline, or"preinvasive," lesion, which, at this time, is not a threat to her life. In the Van Nuys series of 504patients with DCIS, absolute mortality is 0.6% (there have been three breast cancer-related deaths).The 10-year actuarial breast cancer-specific mortality is 0% for mastectomy patients, 1% for allpatients, and 2% for breast-preservation patients. Numerous other DCIS series[2,7,14,17,58-63]confirm the extremely low mortality of DCIS.One of the most common concerns expressed by patients once a diagnosis of cancer has been madeis the fear that the cancer has spread. The DCIS patient can be assured that no invasion was seenmicroscopically and that the likelihood of systemic spread is minimal.The patient with DCIS should be educated that the term "breast cancer" encompasses a wide varietyof lesions with a wide range of aggressiveness and potential lethality. She needs to be reassured

Page 4 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

that she has a minimal lesion and that she may require some additional treatment, which mayinclude further surgery and/or radiation therapy. The patient needs to know that she will not requirechemotherapy, that her hair will not fall out, and that it is highly unlikely that she will die from thislesion. She will, of course, also need careful clinical follow-up.

The Van Nuys Prognostic Index

It is clear that breast irradiation reduces the local recurrence rate at 5 years by about half (fromabout 20% to about 10%). Does this mean that all conservatively treated patients with DCIS shouldreceive postoperative radiation therapy? Series with longer follow-up[64,65] suggest that as timepasses, recurrences continue to accrue in patients treated with radiation therapy. This raisesspeculation that, at least in some patients, radiation merely delays, rather than prevents, aninevitable recurrence.The median time for local recurrence for excision-only patients in our series was 1.7 years, whereasfor patients treated with excision plus radiation therapy, it was 4.7 years. There is now sufficient,easily available, information that can aid clinicians in differentiating patients who may benefit frompostexcision radiation therapy from those who may not. These same data can identify patients whoare better served by mastectomy because their recurrence rates with breast conservation areunacceptably high with or without radiation therapy.Our research[6-8,14,49,51,66], as well as that of other investigators,[44,46,52,53,67,68] has shownthat various combinations of nuclear grade, the presence of comedo-type necrosis, tumor size, andmargin status are all important predictors of local recurrence in conservatively treated patients withDCIS. We believe that it is possible, by using a combination of these prognostic factors, to distinguishsubgroups of patients who do not require radiation therapy in addition to complete excision and alsopatients whose recurrence rate is theoretically so high, even with breast irradiation, thatmastectomy is preferable.The first two of these prognostic factors (nuclear grade and necrosis) were used to develop the VanNuys Pathologic Classification[49] described above (which henceforth will be referred to aspathologic classification). Nuclear grade and comedo-type necrosis reflect the biology of the lesion,but neither is adequate as the sole guideline for treatment decision-making. Tumor size and marginwidth reflect the extent of disease, the adequacy of surgical treatment, and the likelihood of residualdisease and therefore, are extremely important. The results of the multivariate analysis confirm thecritical importance of these variables.How Tumors Are Scored The Van Nuys Prognostic Index (VNPI)[50,69,70] was devised by combining these three statisticallysignificant predictors of local tumor recurrence in patients with DCIS: tumor size, margin width, andpathologic classification. A score, ranging from 1 for lesions with the best prognosis to 3 for lesionswith the worst prognosis, was assigned for each of the three predictors. The objective with all threepredictors was to create three statistically different subgroups for each, using local recurrence as themarker of treatment failure. Cut-off points (for example, what size or margin width constitutes low,intermediate, or high risk of local recurrence) were determined by statistical modeling, using thelog-rank test with an optimum P-value approach.Size Score--A score of 1 was given for small tumors ( 15 mm or less in diameter); a score of 2, forintermediate-sized tumors (16 to 40 mm); and a score of 3, large tumors ( 41 mm or more).Margin Score--A score of 1 was used to denote widely clear tumor-free margins (10 mm or more).This was most commonly achieved by reexcision with the finding of no residual DCIS or only focalresidual DCIS in the biopsy cavity. A score of 2 was used for intermediate margins (1 to 9 mm) and ascore of 3, for involved or close margins (less than 1 mm).Pathologic Classification Score--A score of 3 was given for tumors classified as group 3 (allhigh-grade lesions); a score of 2, for tumors classified as group 2 (non-high-grade lesion withcomedo-type necrosis); and a score of 1, for tumors classified as group 1 (non-high-grade lesionwithout comedo-type necrosis).[49]Calculating the VNPI The VNPI formula was determined by using the beta values, obtained from the initial multivariateanalysis.[6,70] The beta values reflect the relative contribution of each factor to the estimation ofthe likelihood of local recurrence[7,70,71]; these values were similar for all three factors. Thismethod yielded a relatively user-unfriendly formula with 27 possible VNPI scores. The 27 subgroupsnaturally divided into three prognostic subgroups with low, intermediate, and high risks,respectively, of recurrence.

Page 5 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

Because the beta values for the three factors were similar, additional analyses revealed that theformula could be simplified, without compromising validity, by omitting the beta weighting,suggested by the multivariate analysis, and by readjusting the numerical range for each of the threesubgroups. Thus, the final formula for the VNPI became:VNPI = Pathologic classification score + Margin score + Size score This formula yielded a total of seven groups with whole-number scores ranging from 3 to 9. The bestpossible VNPI score was 3, a score of 1 for each predictor (eg, a 5-mm low-grade lesion with widelyclear margins [10 mm or more] would earn this score). The worst possible score was 9, a score of 3for each predictor (eg, a 50-mm high-grade lesion with close or involved margins [less than 1 mm]would warrant such a score). Table 3 summarizes the scoring for the VNPI. When patients weresubdivided into those with scores of 3 or 4, those with scores of 5, 6, or 7, and those with scores of 8or 9, the results were identical to the more complicated beta-weighted version and the VNPI wasmuch easier to use.Validating the VNPI The VNPI was initially tested on 254 breast-conservation patients from The Breast Center, and theseresults were presented at the annual meeting of the American Society of Clinical Oncology (ASCO) inApril 1995.[6] The index was subsequently validated by analyzing a series of patients treated withexcision-only compiled by Lagios et al.[8,51,66] The validation yielded almost identical localdisease-free survival curves for all VNPI subgroups when 83 Breast Center excision-only patients,who met the strict criteria of Lagios et al,[51] were compared with the 79 patients of Lagios etal.[8,66] The results of the combined series were presented at the 18th Annual San Antonio BreastCancer Symposium[50] in December 1995.The VNPI was next tested in a group of 42 DCIS patients treated by the Breast Center's radiationoncologists (Western Tumor Radiation Therapy Group) but not operated upon by the center'ssurgeons. All pathology was centrally reviewed by the Breast Center's pathologist. The results againrevealed almost identical outcomes when the two groups of radiation therapy patients werecompared by VNPI scores.With two outside groups of patients validating the VNPI, we feel comfortable presenting data throughmid-1996 by combining all three groups for a total of 394 patients (273 breast-preservation patientsfrom the Breast Center, 79 excision-only patients from Lagios et al, and 42 excision-plus-radiationtherapy patients from the Western Tumor Radiation Therapy Group).Results of an Analysis Using the VNPI The local recurrence-free survival for all 394 patients is shown by pathologic classification in Figure6, by tumor size in Figure 7, and by margin width in Figure 8. The differences among the threesurvival curves for each of the three predictors that constitute the VNPI are statistically significant.Figure 9 shows the local recurrence-free survival rate for each of the seven VNPI groups, 3 through9. Figure 10 groups patients with low (VNPI = 3 or 4), intermediate (VNPI = 5, 6, or 7), or high (VNPI= 8 or 9) recurrence rates together. Each of these three groups is statistically different from theother two with respect to local recurrence-free survival.As shown in Figure 11, breast irradiation does not significantly decrease local recurrence in patientswith low VNPI scores (3 or 4). Patients with intermediate VNPI scores (5, 6, or 7) derive some benefitfrom irradiation (Figure 12). Among this group, there is a 13% average decrease in local recurrencerate in irradiated patients, which is statistically significant compared to those treated by excisionalone (P = .027). Patients with high VNPI scores (8 or 9) benefit the most from the addition ofradiation therapy to their treatment regimen (Figure 13). However, even though the differencebetween the two groups is significant (P = .025), patients with a VNPI score of 8 or 9 have anextremely high rate of recurrence, regardless of whether of not they receive radiation therapy.

Discussion

Although mastectomy is curative for approximately 98% to 99% of patients withDCIS,[7,17,58,59,62,63,72-74] mastectomy represents overtreatment for most cases detected bycurrent methods. When breast conservation is elected, radiation therapy statistically decreases thelikelihood of local recurrence when compared with excision alone.[2,3] However, radiation therapy,like mastectomy, may also represent overtreatment for a significant number of patients who opt forbreast preservation.Subsets of patients who are not likely to derive a significant benefit from radiation therapy can beidentified, eg, those with VNPI scores of 3 or 4 in the series presented here, low-grade lesions in theseries of Lagios et al,[8,66] small noncomedo lesions with uninvolved margins in the series of

Page 6 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

Schwartz, et al,[44] or the well-differentiated lesions of Zafrani et al.[68] Such patients may accountfor 30% to 40% of all currently diagnosed patients with DCIS.[8,44,49,66,68]The recommendation by the NSABP that radiation therapy is appropriate for all patients with DCISwho are treated with breast preservation does not take into account the heterogeneity of DCIS or thesignificant differences in subsets demonstrated by our data,[7,8,14,49,51,66] the data ofothers,[3,44,46,52,53,67,68] and, more recently, the NSABP's own data.[3]Radiation therapy, like all treatments, has side effects. Radiation changes the texture of the breastand may make subsequent mammography more difficult to interpret. Perhaps most importantly, itsuse precludes additional radiation therapy and breast conservation should invasive breast cancerdevelop at a later date. The risks and benefits of radiation therapy must be carefully weighed andcompared to those of other options prior to making a treatment recommendation. The use ofradiation therapy should then be advocated only for those patients likely to obtain a benefit.Treatment Recommendations Based on VNPI Scores Patients in this series with VNPI scores of 8 or 9 pose a special problem. While these patients derivethe greatest relative benefit from postexcision radiation therapy, their local recurrence ratecontinues to be unacceptably high, more than 60% at 5 years. Therefore, a recommendation formastectomy should be considered.Treatment recommendations for the intermediate group (patients with VNPI scores of 5, 6, or 7) arethe most difficult to make. For patients with intermediate VNPI scores and margin scores of 2 or 3,reexcision may decrease the risk of local recurrence by lowering their VNPI score ("downscoring"). Ifthe score remains intermediate after reexcision, radiation therapy should be considered. However,some patients with scores of 7 may be better treated with mastectomy (eg, a patient with a largegrade 2 lesion without necrosis that shows involved margins after reexcision). Some patients withscores of 5 may elect no further treatment (eg, a patient with widely clear margins, small tumor size,but high nuclear grade).These are independent judgments that must be made by the patient and her physician. We arehopeful that the VNPI will be a helpful adjunct as these difficult decisions are discussed. If the VNPI,or some similar system, is not used, there are few data in the literature to aid clinicians in thecomplex DCIS treatment selection process.Potential Flaws and Benefits of the VNPI System There is a strong selection or treatment bias among the patients used to develop the VNPI. In allcases, the treatment was selected by the individual patient and her physician, not by randomassignment. However, treatment selection does not bias the results since the VNPI comparespatients with different scores, not different treatments. Although the patient and her clinician controltreatment selection, neither can control final margins, tumor size, or pathologic classification. Thefact that some patients opted for suboptimal treatments that were not recommended (eg, 33patients with VNPI scores of 8 or 9 who selected breast conservation were all advised to undergomastectomy) was helpful in developing and evaluating the VNPI.Counseling patients with DCIS in a rational manner can be extremely difficult when the range oftreatment options is extreme. The VNPI permits the physician to carry out a scientifically baseddiscussion with the patient, using the parameters of the lesion obtained after an initial excision.Thus, in some cases, a patient can choose reexcision, in an effort to downscore her lesion.Successful downscoring of a patient with a VNPI score of 8 or 9 could result in a substantial reductionin the risk of local recurrence, perhaps changing a recommendation from mastectomy to breastpreservation with radiation therapy. Similarly, patients with close or involved margins who have VNPIscores of 5 or 6 after initial excision could opt for reexcision. Successful downscoring by achievingwidely clear margins could result in a final VNPI score of 3 or 4, leading to a recommendation forcareful clinical follow-up without radiation therapy.Downscoring can be accomplished only by reexcising lesions in patients with margin scores of 2 or 3.Reexcision will not lower the pathologic classification score or reduce the size of the tumor. In somecases, reexcision will upscore the tumor, increasing the VNPI score by revealing a larger tumor size,a higher nuclear grade, the presence of previously undetected comedo necrosis, or an involvedmargin.The VNPI is the first attempt to quantify known important prognostic factors in DCIS, making themclinically useful in the treatment decision-making process. It may be useful to clinicians because itdivides DCIS into three groups with different risks for local recurrence after breast-conservationtherapy. Although there is an obvious treatment choice for each group (Table 4)--excision only forpatients with scores of 3 or 4, excision plus radiation therapy for patients with scores of 5, 6, or 7,and mastectomy for patients with scores of 8 or 9--the VNPI is offered only as a guideline, a starting

Page 7 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

place for discussions with patients. The case histories below illustrate how the VNPI was used to aidthe treatment selection process.

Conclusions

Based on the research to date, including our studies, the following general conclusions can be madeabout DCIS and the role of the VNPI in helping the physician and patient navigate the complextreatment decision-making process:1. DCIS is relatively common and its frequency is increasing. Most of this increase is due to bettermammographic detection and greater familiarity with minimal lesions among pathologists.2. Not all microscopic DCIS will progress to clinical cancer, but if a patient has DCIS and is nottreated, she is far more likely to develop an ipsilateral invasive breast cancer than is a womanwithout DCIS.3. The separation of DCIS into two groups by architecture (comedo vs noncomedo) is anoversimplification and does not reflect the biologic heterogeneity of the disease. This conventionalclassification may mask the biologic potential of the lesion, which may be better recognized by usinga classification based on nuclear grade and comedo-type necrosis.4. High-grade DCIS is more aggressive and malignant in its histologic appearance and is more likelyto be associated with subsequent invasive cancer than is non-high-grade DCIS. High-grade DCIS ismore likely than non-high-grade DCIS to have a high S-phase, overexpress HER-2/neu (c-erb-B2) ,and show increased thymidine labeling. Also, when treated conservatively, high-grade DCIS is morelikely to recur locally than is non-high-grade DCIS.5. Most cases of DCIS detected today will be nonpalpable. Such cases will be detected bymammography, with microcalcifications as the most common finding. It is not uncommon for DCIS tobe larger than expected by mammography, to involve more than a quadrant of the breast, and to beunifocal (rather than multicentric) in its distribution.6. Preoperative evaluation should include film-screen mammography with compressionmagnification. The surgeon and radiologist should plan the excision procedure carefully. The firstattempt at excision is the best chance for achieving complete excision with a good cosmetic result.Reexcisions are more likely to yield poor cosmetic results. In patients with suspicious mammographiclesions, consideration should be given to stereotactic core biopsy to make a definitive diagnosis priorto initial wide excision.7. The successful treatment of a patient with DCIS, whether by breast conservation or mastectomy,presupposes that a competent and interested pathologist will be part of the team. It is best to planthe resection with the pathologist, radiologist, and surgeon all having input into the discussion.The data presented in this paper are based on complete, sequential tissue processing andmammographic/pathologic correlation. It is important that any facility in which DCIS patients aretreated have this capability.8. Following the establishment of the diagnosis, the patient should be apprised of the risks andadvantages of all alternative procedures. If she wants to preserve her breast, the surgeon andradiologist should plan the procedure carefully, using multiple wires to map out the extent of thelesion. Once the multiple-wire-directed excisional biopsy has been done, two factors can beevaluated: cosmesis and histopathology. If the cosmetic result is acceptable and the margins areclear, breast conservation can proceed. If the patient has a VNPI score of 3 or 4, the option of nofurther therapy can be considered.If the patient's VNPI score is 5, 6, or 7, reexcision may be possible. If reexcision cannot be done or ifthe postexcision score remains in the mid-range, breast irradiation should be considered. Occasionalpatients with VNPI scores of 5 may be treated with excision alone, while some patients with VNPIscores of 7 may be better served by mastectomy. Individual judgment is required for all patientswith intermediate scores.For patients with VNPI scores of 8 or 9, a mastectomy with immediate reconstruction is usuallyrecommended.9. If the initial biopsy margins are involved, consideration can be given to a reexcision procedure, butit may yield a poor cosmetic result, and margins may continue to be involved. At this point,mastectomy (with or without immediate reconstruction) should be considered. Reconstruction can beaccomplished with a variety of techniques, including an expander, an implant, and a transrectusabdominus myeocutaneous (TRAM) flap. In general, immediate reconstruction in combination with askin-sparing mastectomy is preferable. It eliminates at least one future surgical procedure andusually results in a happier patient with a better cosmetic result.

Page 8 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

10. For women with larger lesions (relative to breast size) that cannot be totally excised,mastectomy remains the treatment of choice. However, the NSABP is conducting a prospective study(Protocol B-24) that is randomizing patients to one of two treatments: excision plus breast irradiationwith or without tamoxifen (Nolvadex). In this study, positive margins and residual calcifications areallowed. The results of this study are eagerly awaited.

Future Prospects

Clearly, the validity of the VNPI must be independently confirmed by other groups with large seriesof DCIS patients and sufficient data to complete the subset analysis as outlined here. Prospectiveconfirmation, however, would be better.In the future, other factors, such as molecular markers, may be integrated into the VNPI or otherprognostic indices when they are shown to statistically influence the likelihood of local recurrenceafter breast-conservation therapy.Our knowledge of DCIS genetics and molecular biology is increasing at a remarkably rapid rate.Future studies may identify markers that will allow us to distinguish DCIS with an invasive potentialfrom DCIS that is merely a microscopic finding. When it becomes possible to make such a distinction,the treatment selection process will become much simpler. References: 1. Silverstein MJ: Intraductal breast carcinoma: Two decades of progress? Am J ClinOncol 14(6):534-537, 1991.

2. Fisher B, Costantino J, Redmond C, et al: Lumpectomy compared with lumpectomy and radiationtherapy for the treatment of intraductal breast cancer. N Engl J Med 328:1581-1586, 1993.

3. Fisher ER, Constantino J, Fisher B, et al: Pathologic findings from the National Surgical AdjuvantBreast Project (NSABP) Protocol B-17. Cancer 75:1310-1319, 1995.

3a. Mamounas E, Fisher B, Dignam J, et al: Effect of breast irradiation following lumpectomy inintraductal breast cancer (DCIS): Update results from NSABP B-17. Proc Soc Surg Oncol 50:7, 1997.

4. Lagios MD, Page DL: Radiation therapy for in situ or localized breast cancer (letter). N Engl J Med21:1577-1578, 1993.

5. Page DL, Lagios MD: Pathologic analysis of the NSABP-B17 trial: Unanswered questions remainingunanswered considering current concepts of ductal carcinoma in situ. Cancer 75:1219-1222, 1995.

6. Silverstein MJ, Barth A, Waisman JR, et al: Predicting local recurrence in patients with intraductalbreast carcinoma (DCIS). Proc Am Soc Clin Oncol 14:117, 1995.

7. Silverstein MJ, Barth A, Poller DN, et al: Ten-year results comparing mastectomy to excision andradiation therapy for ductal carcinoma in situ of the breast. Eur J Cancer 31:1425-1427, 1995.

8. Lagios NM, Margolin FR, Westdahl PR, et al: Mammographically detected duct carcinoma in situ:Frequency of local recurrence following tylectomy and prognostic effect of nuclear grade on localrecurrence. Cancer 63:619-624, 1989.

9. Lennington WJ, Jensen RA, Dalton LW, et al: Ductal carcinoma in situ of the breast: Heterogeneityof individual lesions. Cancer 73:118-124, 1994.

10. Patchefsky AS, Schwartz GF, Finkelstein SD, et al: Heterogeneity of intraductal carcinoma of thebreast. Cancer 63:731-741, 1989.

11. Morrow M, Schnitt SJ, Harris JR: Ductal carcinoma in situ, in Harris JR, Lippman MC, Morrow M, etal (eds): Diseases of the Breast, pp 355-368. Philadelphia-New York, Lippincott-Raven, 1995.

12. Nemoto T, Vana J, Bedwani RN, et al: Management and survival of female breast cancer: Resultsof a national survey by The American College of Surgeons. Cancer 45:2917-2924, 1980.

Page 9 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

13. SEER Cancer Statistics Review: 1973-1990. National Cancer Institute. NIH Pub No. 93-2789, 1993.

14. Lagios MD: Duct carcinoma in situ: Pathology and treatment. Surg Clin North Am 70:853-871,1990.

15. Silverstein MJ, Cohlan B, Gierson ED, et al: Duct carcinoma in situ: 227 cases withoutmicroinvasion. Eur J Cancer 28(2/3):630-634, 1992.

16. Parker SL, Tong T, Bolden S, et al: Cancer statistics, 1997. CA Cancer J Clin 47(1):5-27, 1997.

17. Ashikari R, Hadju SI, Robbins GF: Intraductal carcinoma of the breast. Cancer 28:1182-1187,1971.

18. Barth A, Brenner J, Giuliano AE: Current management of ductal carcinoma in situ. Western J Med163:360-366, 1995.

19. Stockdale AD, Brierley JD, Whire WF, et al: Radiotherapy for Paget's disease of the nipple: Aconservative alternative. Lancet 2 (8664):664-666, 1989.

20. Jensen JA, Handel N, Silverstein MJ: Glandular replacement therapy (GRT) for intraductal breastcarcinoma (DCIS). Proc Am Soc Clin Oncol 14:138, 1995.

21. Jensen JA, Handel N, Silverstein MJ: Glandular Replacement Therapy: An argument for acombined surgical approach in the treatment of noninvasive breast cancer. The Breast Journal2:121-123, 1996.

22. Kroll SS, Ames F, Singletary SE, et al: The oncologic risks of skin preservation at mastectomywhen combined with immediate reconstruction of the breast. Surg Gynecol Obstet 172:17-20, 1991.

23. Singletary ES: Skin-sparing mastectomy with immediate breast reconstruction: Is it safe? BreastDiseases: A Yearbook Quarterly 6:259-260, 1995.

24. Silverstein MJ, Handel N, Hoffman RS, et al: The breast center--a multidisciplinary model, inPaterson AHG, Lees AW (eds): Fundamental Problems in Breast Cancer, pp 47-58. Boston, MartinusNijhoff, 1987.

25. Blichert-Toft M, Brincker H, Andersen J, et al: A Danish randomized trial comparing breastpreserving therapy with mastectomy in mammary carcinoma. Acta Oncol 27:671, 1988.

26. Blichert-Toft M, Rose C, Andersen J, et al: Danish randomized trial comparing breast conservativetreatment with mastectomy: Six years of life table analysis. J Natl Cancer Inst Monogr11:19, 1992.

27. Fisher B, Redmond C, Poisson R, et al: Eight-year results of a randomized clinical trial comparingtotal mastectomy and lumpectomy with or without irradiation in the treatment of breast cancer. NEngl J Med 320:822-828, 1989.

28. Fisher B, Anderson S, Redmond CK, et al: Reanalysis and results after 12 years of follow-up in arandomized clinical trial comparing total mastectomy with lumpectomy with or without irradiation inthe treatment of breast cancer. N Engl J Med 333:1456-1461, 1995.

29. Lichter A, Lippman M, Danforth D, et al: Mastectomy versus breast conserving therapy in thetreatment of stage I and II carcinoma of the breast: a randomized trial at The National CancerInstitute. J Clin Oncol 10:976, 1992.

30. Van Dongen JA, Bartelink H, Fentiman IS, et al: Randomized clinical trial to assess the value ofbreast-conserving therapy in stage I and II breast cancer, EORTC 10801 trial. Monogr Natl CancerInst 11:15-18, 1992.

Page 10 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

31. Veronesi U, Saccozzi R, Del Vecchio M, et al: Comparing radical mastectomy withquadrantectomy, axillary dissection and radiotherapy in patients with small cancers of the breast. NEngl J Med 305:6, 1981.

32. Veronesi U, Banfi A, Salvadori B, et al: Breast conservation is the treatment of choice in smallbreast cancer: Long-term results of a randomized trial. Eur J Cancer 26:668-670, 1990.

33. Page DL, Anderson TJ: Intraductal Barcinoma, in Page DL, Anderson TJ. (eds): DiagnosticHistopathology of the Breast, pp 157-174. New York, Churchill Livingstone, 1987.

34. Tavassoli FA: Intraductal carcinoma, in Tavassoli FA (ed): Pathology of the Breast, pp 229-261.Norwalk, Appleton & Lange, 1992.

35. Rosen PP, Oberman HA: Intraepithelial (Preinvasive or in situ) carcinoma, in Rosen PP, ObermanHA (eds): Atlas of Tumor Pathology--Tumors of the Mammary Gland, pp 119-156. Washington, DC,Armed Forces Institute of Pathology, 1993.

36. Aasmundstad TA, Haugen OA: DNA Ploidy in intraductal breast carcinomas. Eur J Cancer26:956-959, 1992.

37. Meyer J: Cell kinetics of histologic variants of in situ breast carcinoma. Breast Cancer Res Treat7:171-180, 1986.

38. Allred DC, Clark GM, Molin R, et al: Overexpression of progression of in situ to invasive breastcancer. Hum Pathol 23:974-979, 1992.

39. Barnes DM, Meyer JS, Gonzalez JG, et al: Relationship between c-erbB-2 immunoreactivity andthymidine labelling index in breast carcinoma in situ. Breast Cancer Res Treat 18:11-17, 1991.

40. Bartkova J, Barnes DM, Millis RR, et al: Immunohistochemical demonstration of c-erbB-2 proteinin mammary ductal carcinoma in situ. Hum Pathol 21:1164-1167, 1990.

41. Bobrow LG, Happerfield LC, Gregory WM, et al: The classification of ductal carcinoma in situ andits association with biological markers. Semin Diagn Pathol 11:199-207, 1994.

42. Liu E, Thor A, He M, et al: The HER2 (c-erbB-2) oncogene is frequently amplified in in situcarcinomas of the breast. Oncogene 7:1027-1032, 1992.

43. Van de Vijver MJ, Peterse JL, Mooi WJ, et al: Neu-protein overexpression in breast cancer:association with comedo-type ductal carcinoma in situ and limited prognostic value in stage II breastcancer. N Engl J Med 319:1239-1245, 1988.

44. Schwartz GF: The role of excision and surveillance alone in subclinical DCIS of the breast.Oncology 8(2):21-26, 1994.

45. Silverstein MJ, Waisman JR, Gierson ED, et al: Radiation therapy for intraductal carcinoma: Is it anequal alternative? Arch Surg 126:424-428, 1991.

46. Solin LJ, Yet IT, Kurtz J, et al: Ductal carcinoma in situ (intraductal carcinoma) of the breasttreated with breast-conserving surgery and definitive irradiation. Correlation of pathologicparameters with outcome of treatment. Cancer 71:2532-2542, 1993.

47. Holland R, Peterse JL, Millis R et al: Ductal carcinoma in situ: A proposal for a new classification.Semin Diag Pathol 11(3):167-180, 1994.

48. Kuske RR, Bean JM, Garcia DM, et al: Breast conservation therapy for intraductal carcinoma ofthe breast. Int J Radiat Oncol Biol Phys 26:391-396, 1993.

49. Silverstein MJ, Poller DN, Waisman JR: Prognostic classification of breast ductal carcinoma in situ.

Page 11 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

Lancet 345:1154-1557, 1995.

50. Silverstein MJ, Poller DN, Craig PH, et al: A prognostic index for breast ductal carcinoma in situ(abstract). Breast Cancer Res Treat 37(suppl):34, 1996.

51. Lagios MD, Westdahl PR, Margolin FR, et al: Duct Carcinoma in situ: Relationship of extent ofnoninvasive disease to the frequency of occult invasion, multicentricity, lymph node metastases, andshort-term treatment failures. Cancer 50:1309-1314, 1982.

52. Poller DN, Silverstein MJ, Galea M, et al: Ductal carcinoma in situ of the breast: A proposal for anew simplified histological classification association between cellular proliferation and c-erbB-2protein expression. Mod Pathol 7:257-262, 1994.

53. Bellamy COC, McDonald C, Salter DM, et al: Noninvasive ductal carcinoma of the breast: Therelevance of histologic categorization. Hum Pathol 24:16-23, 1993.

54. Sloane JP, Ellman R, Anderson TJ, et al: Consistency of histopathological reporting of breastlesions detected by breast screening: Findings of the UK national external quality assessment (EQA)scheme. Eur J Cancer 30:1414-1419, 1994.

55. Douglas-Jones AG, Gupta SK, Attanoos RL, et al: A critical appraisal of six modern classificaitonsof ductal carcinoma in situ of the breast (DCIS): Correlation with grade of associated invasivedisease. Histopathology 29:397-409, 1996.

56. Dickson RB, Lippman ME: Growth factors in breast cancer. Endocrine Reviews 16(5):559-589,1995.

57. Lippman ME: The rational development of biological therapies for breast cancer. Science259:631-632, 1993.

58. Archer SG, Kemp BL, Gadd M, et al: Ductal carcinoma in situ of the breast: Comedo versusnoncomedo subtype nonpredictive of recurrence of contralateral new breast primary. Breast Dis7:353-360, 1994.

59. Arnesson LG, Smeed S, Fagerberg G, et al: Follow-up of two treatment modalities for ductalcarcinoma in situ of the breast. Br J Surg 76:672-675, 1989.

60. Carter D, Smith RRL: Carcinoma in situ of the breast. Cancer 40:1189-1193, 1977.

61. Ciatto S, Bonardi R, Cataliotti L, et al: Intraductal breast carcinoma. Review of a multicenterseries of 350 cases. Tumori 76:552-554, 1990.

62. Farrow JH: Current concepts in the detection and treatment of the earliest of the breast cancers.Cancer 25:468-477, 1970.

63. Fentiman IS, Fagg N, Millis RR, et al: In situ ductal carcinoma of the breast: Implications ofdisease pattern and treatment. Eur J Surg Oncol 12:261-266, 1986.

64. Haffty BG, Peschel RE, Papadopoulos D, et al: Radiation therapy for ductal carcinoma in situ ofthe breast. Connecticut Medicine 54:482-484, 1990.

65. Solin LJ, Kurtz J, Fourquet A, et al: Fifteen-year results of breast-conserving surgery and definitiveirradiation for the treatment of ductal carcinoma in situ of the breast. J Clin Oncol 14:754-763, 1996.

66. Lagios MD: Ductal carcinoma in situ: Controversies in diagnosis, biology, and treatment. TheBreast Journal 1:68-78, 1995.

67. Ottesen GL, Graversen HP, Blichert-Toft M, et al: Ductal carcinoma in situ of the female breast:Short-term results of a prospective nationwide study. Am J Surg Pathol 16:1183-1196, 1992.

Page 12 of 13

Use of Predictors of Recurrence to Plan Therapy for DCIS of the BreastPublished on Cancer Network (http://www.cancernetwork.com)

68. Zafrani B, Leroyer A, Fourquet A, et al: Mammographically-detected ductal carcinoma in situ ofthe breast analysed with a new classification: A study of 127 cases: Correlation with estrogen andprogesterone receptors, p53 and c-erbB-2 proteins and proliferative activity. Semin Diag Pathol11(3):208-213, 1994.

69. Silverstein MJ, Lagios MD, Craig PH, et al: The Van Nuys Prognostic Index for ductal carcinoma insitu. The Breast Journal 2:38-40, 1996.

70. Silverstein MJ, Lagios MD, Craig PH, et al: A prognostic index for ductal carcinoma in situ of thebreast. Cancer 77:2267-2274, 1996.

71. Galea MH, Blamey RW, Elston CE, et al: The Nottingham Prognostic Index in primary breastcancer. Breast Cancer Res Treat 22:207-219, 1992.

72. Bradley SJ, Weaver DW, Bouwman DL: Alternatives in the surgical management of in situ breastcancer. Am Surg 56:428-432, 1990.

73. Fisher ER, Leiming ER, Anderson S, et al: Conservative management of intraductal carcinoma(DCIS) of the breast. J Surg Oncol 47:139-147, 1991.

74. Kinne DW, Petrek JA, Osborne MP, et al: Breast carcinoma in situ. Arch Surg 124:33-36, 1989. Source URL: http://www.cancernetwork.com/printpdf/use-predictors-recurrence-plan-therapy-dcis-breast-0/page/0/1

Links:[1] http://www.cancernetwork.com/review-article[2] http://www.cancernetwork.com/breast-cancer[3] http://www.cancernetwork.com/authors/melvin-j-silverstein-md[4] http://www.cancernetwork.com/authors/michael-d-lagios-md

Page 13 of 13