Validation study of a modern t

, BaNabil Wasif, M.D., F.A.C.S. , Victor J. Pizzitola, M.D. , Marina E. Giurescu, M.D. ,

aDepartment of Surgery, bDepar

KEYWORDS:Breast cancer;Nipple discharge;Treatment algorithm;Validation study

d-

2014 Elsevier Inc. All rights reserved.

complaints. Previous studies have found a breast carcinoma there is no clear guideline on what differentiates a benign

treatment algorithm (Fig. 1) for selectively identifying pa-tients with nipple discharge who were at higher risk of ma-lignancy and should undergo excision, whereas the othergroup had low enough risk to undergo follow-up.14 The

There were no relevant financial relationships or any sources of support

in the form of grants, equipment, or drugs.

The authors declare no conflicts of interest.

* Corresponding author. Tel.: 11-480-342-2849; fax: 11-480-342-2866.

The American Journal of Surgery (2014) -, --E-mail address: gray.richard@mayo.eduincidence of 9.3% to 21.3% in women with pathologicnipple discharge,210 but most only included patients who

etiology from a malignancy based on clinical and radio-graphic assessment.

For decades, patients with pathologic nipple dischargewere counseled to undergo a duct excision.3,6,1013 Howev-er, in our previous study,14 we proposed an evidence-basedNipple discharge is a frequent presenting symptom,occurring in 3% to 10%1,2 of all women with breast-related

underwent an operative excision.3,810 While nearly 90%of women with nipple discharge have benign disease,10graphic findings and clinical examination for safe clinical follow-up. Most will have resolution avoiing a surgical procedure.Manuscript received July 23, 2013;

2013

0002-9610/$ - see front matter 2014http://dx.doi.org/10.1016/j.amjsurg.20tment of Radiology, Mayo Clinic, 5777 E. Mayo Blvd, Phoenix, AZ, USA

AbstractBACKGROUND: Nipple discharge occurs in 2% to 5% of women. We evaluated the effectiveness of a

previously proposed treatment algorithm for these patients.METHODS: Patients with pathologic nipple discharge and a negative mammogram and subareolar

ultrasound were offered follow-up from 2005 to 2011 according to the algorithm.RESULTS: A total of 192 patients, mean age 56 years, were studied. Risk of carcinoma among the

entire cohort was 5%. Breast surgeon was consulted for 142 (74%) patients: 48 (34%) underwent initialsubareolar excision and 94 (66%) were clinically followed. The rate of carcinoma was 17% (8/48) afterinitial subareolar excision, 0% (0/13) for those without imaging abnormalities, 23% (8/35) with imag-ing abnormalities, and 1% (1/94) with clinical follow-up. Of patients who underwent follow-up, 21% (n5 20) underwent subareolar excision because of imaging abnormality (n 5 1, 1%) or persistentdischarge (n 5 19, 20%). Most patients had ductal carcinoma in situ (n 5 5, 56%).

CONCLUSIONS: Patients with nipple discharge can be prospectively identified based on radio-Richard J. Gray, M.D., F.A.C.S.a,*for nipple discharge

Awais Ashfaq, M.D.a, Derek Senior, B.S.aarevised manuscript November 4,

Elsevier Inc. All rights reserved.

13.12.035reatment algorithm

rbara A. Pockaj, M.D., F.A.C.S.a,b bgoal of this study was to validate the proposed treatmentalgorithm after its implementation in our practice.

f nip

The American Journal of Surgery, Vol -, No -, - 2014Methods

Figure 1 Algorithm for the treatment o

2All patients with the presenting symptom of nippledischarge from 2005 to 2011 at a single institution wereidentified through a database of electronic medical records,breast clinic complaints, and billing records after the studywas approved by the Institutional Review Board. Theelectronic medical records system was queried for thephrases nipple discharge and breast discharge withinany clinical note and for any patients with an InternationalClassification of Diseases, Ninth Revision code diagnosisof 611.79 (discharge nipple, discharge breast) to identifythe cohort of patients. Thus, all patients treated for nippledischarge after institution of the nipple discharge manage-ment algorithm in our practice were reviewed for manage-ment compliance and outcome.

The same treatment algorithm was followed by allsurgeons practicing at our institution. Patients with patho-logic discharge (defined as spontaneous, bloody, or serousdischarge from a single duct), an otherwise benign physicalexamination, and a negative mammogram and subareolarultrasound were offered clinical follow-up according to thealgorithm.

Results

A total of 192 patients with nipple discharge wereidentified. Data on whether the discharge was unilateral orbilateral were available for 100% of the patients, whetherthe discharge was spontaneous in 94%, characteristics of

ple discharge. Derived from Gray et al.14the discharged fluid in 98%, and single versus multiple ductdischarge in 81% of patients.

Two patients were male and 190 patients were female(Table 1). The mean age of the patients was 56 years (range19 to 94). Unilateral nipple discharge was present in 77%, itwas spontaneous in 72%, the discharge was bloody in 34%,and clear/serous in 43%. Single duct discharge was presentin 69% of patients for whom this information was available.

Radiologic imaging was performed in all patients.Mammography was performed in 177 patients (92%) andin all patients aged 30 or greater. Mammographic abnor-malities (defined as a mass, indeterminate/suspiciouscalcifications, or architectural distortion) were present in13 (7%) patients. Ultrasonography was performed in 149patients (78%). Sonographic abnormalities (defined as a

Table 1 Patient demographics and characteristics ofdischarge

Variables n (%)

Median age (years) 56 (1994)Female 190 (99%)Discharge characteristics

Unilateral 148 (77%)Spontaneous 138 (72%)Single duct 132 (69%)Bloody 65 (34%)Clear/serous 83 (43%)

mass or intraductal mass(es)) were present in 46 (24%)patients. Ductography was performed in 13 patients (7%).Ductographic abnormalities (duct cutoff or filling defect)were present in 9 (64%) patients. Of the 9 patients who hadductogram abnormalities, 3 had intraductal papillomas, 3had duct ectasia, 2 had ductal hyperplasia, and 1 had ductalcarcinoma in situ. Breast magnetic resonance imaging(MRI) was performed in 8 patients (4%). Abnormalitieson MRI (defined as a mass or suspicious enhancementpattern) were present in 4 patients (50%). The single MRIthat failed to identify a carcinoma was performed at another

institution and while it was of good quality, limitedsequence imaging was provided to us for analysis. Thesensitivity and specificity of these radiologic procedures arepresented in Table 2. Each patient with an imaging abnor-mality had tissue sampling performed.

There were 142 (74%) patients who were evaluated by abreast surgeon and the remaining 50 patients were seen byprimary care physicians alone (Fig. 2). No patient who wasnot referred to a breast surgeon developed breast cancerafter mean and median follow-up of 25 and 17 months,respectively. Among patients consulting a surgeon, a biopsy

Table 2 Sensitivity and specificity of radiologic procedures in detecting carcinoma

Examination Total number of patients Sensitivity Specificity Positive predictive value

Mammography 177 2/9 (22%) 157/168 (94%) 2/13 (15%)Ultrasonography 149 8/8 (100%) 102/140 (73%) 8/46 (17%)Ductography 13 1/1 (100%) 4/15 (33%) 1/9 (11%)Magnetic resonance imaging 8 1/2 (50%) 4/7 (57%) 1/4 (25%)

A. Ashfaq et al. Treatment algorithm for nipple discharge 3Figure 2 Flow diagram of all patients reviewed and the risk of developing carcinoma.

or subareolar duct excision at the time of initial evaluationwas performed in 48 (34%) patients including 12% (13/107) of those with no imaging abnormality and 100%(35/35) of the patients with an imaging abnormality. Therates of carcinoma were 0% (0/13) for those with no imag-ing abnormality and 23% (8/35) for those with an imagingabnormality.

Of the 94 patients undergoing close clinical follow-up by asurgeon, 1 (1%) was subsequently found to have ductalcarcinoma in situ (DCIS) at her 6-month follow-up visit. Thispatient had not undergone a subareolar ultrasound at initialevaluation, as called for on the algorithm, and whenperformed at the 6-month follow-up appointment, thisdemonstrated intraductal abnormalities prompting the sub-areolar duct excision that provided the diagnosis. The 9carcinomas diagnosed (Table 3) includedDCIS (n5 5, 56%),invasive ductal carcinoma (n 5 2, 22%), invasive papillarycarcinoma (n 5 1, 11%), and invasive lobular carcinoma (n5 1, 11%). The stages at diagnosis were stage 0 (n 5 5,56%), stage I (n5 3, 33%), and stage III (n5 1, 11%). Me-dian follow-up for those with cancer was 46 months and allpatients remained cancer-free. The single patient with more

Table 3 Characteristics of patients with carcinoma

Variables n (%)

Total no. of patients with carcinoma 9 (4.6%)Histology

Ductal carcinoma in situ 5 (56%)Invasive ductal carcinoma 2 (22%)Invasive papillary carcinoma 1 (11%)Invasive lobular carcinoma 1 (11%)

Stage at diagnosisStage 0 5 (56%)Stage I 3 (33%)Stage II 0 (0%)Stage III 1 (11%)

Median follow-up (months) 46 (1102)Disease-free survival 9 (100%)

4advanced disease (stage III invasive ductal carcinoma) wasa 74-year-old woman who had a normal mammogram but amarkedly abnormal physical examination and subareolar ul-trasound. These findings allowed an immediate diagnosis bypercutaneous core needle biopsy and the institution of multi-modality treatment. She remains free of disease at 49 monthsof follow-up.

Of the patients undergoing close clinical follow-up by asurgeon, 21% (20/94) underwent subsequent subareolarduct excision because of developing an imaging abnormal-ity (1/94, 1%) or for bothersome, persistent discharge (19/94, 20%). The median follow-up was 28 months for the 74patients not undergoing subareolar duct excision and 81%of these patients had resolution of their discharge at lastfollow-up. None of these patients developed carcinoma.

The risk of carcinoma among the entire cohort was 5%,including 4% among women and 50% (1/2) among men.The incidence of carcinoma in the bloody discharge group

lactation. Periductal mastitis is another frequent cause

of a multicolored, sticky nipple discharge.17 Mammaryduct ectasia is often associated with chronic duct inflam-mation (periductal mastitis). This results in a multicolorednipple discharge (green, yellow, white, brown, gray, or red-dish brown). Usually the discharge originates from multi-ple ducts and is often bilateral. All these entities can besafely managed with routine screening imaging and phys-ical examination. In our study, the risk of carcinoma forthese patients was 0%.

For patients with pathologic discharge, we performeddiagnostic mammography (for those 30 years and older)was 11% (7/65) versus 2% (2/127) in the nonbloody group.All patients with carcinoma had an imaging abnormality(although 1 patient did not have that abnormality identifieduntil the 6-month follow-up evaluation as noted above)including 2 with an abnormal mammogram (sensitivity22%, specificity 15%), 8 with an abnormal subareolarultrasound (sensitivity 100%, specificity 17%), 1 with anabnormal ductogram (sensitivity 100%, specificity 11%),and 1 with an abnormal breast MRI (sensitivity 50%,specificity 25%) (Table 2).

Comments

Nipple discharge may be caused by benign conditionssuch as galactorrhea, physiologic stimulation, apocrineglandular secretion, or intraductal papilloma, but is gener-ally idiopathic. Only a small proportion of patients withnipple discharge are found to have a malignancy. However,because it has been felt that the risk of carcinoma cannot beexcluded without surgical duct excision, this operation hasbeen widely recommended for all patients with pathologicnipple discharge.3,6,1013 In our previous retrospective re-view,14 we found that the subset of patients with pathologicnipple discharge, a benign physical examination, a negativemammogram, and a negative subareolar ultrasound has alow risk of underlying carcinoma and proposed followingthese patients clinically. The current cohort allows the re-porting of outcomes of patients evaluated and treated ac-cording to that systematic approach.

In this study, we used the previously proposed treatmentalgorithm14 to evaluate patients with nipple discharge. His-tory and physical examination are an important first step inidentifying benign nipple discharge including galactor-rhea, physiologic, pregnancy-associated, periductalmastitis-associated, and ductal ectasia-associated.12 Galac-torrhea is milky, from multiple ducts and/or bilateral. It ismost commonly observed after pregnancy and can last for1 to 2 years.15 Physiologic nipple discharge is not sponta-neous and is generally bilateral, serous, and arises frommultiple ducts. Pregnancy-associated nipple dischargemay be unilateral or bilateral, bloody, without significantunderlying breast pathology, usually in the 2nd trimester,and can continue as long as 2 years after pregnancy and

16

The American Journal of Surgery, Vol -, No -, - 2014and subareolar ultrasound. All patients with an imaging

in this cohort remained cancer-free. So among patients withabnormality underwent a subareolar duct excision orpercutaneous biopsy and 23% of these were found tohave a carcinoma. Among the 13 patients with no physicalexamination or imaging abnormality who underwent exci-sion because of personal preference or concern of cancerrisk, 0% had malignancy. The risk of carcinoma for thosefollowed clinically was only 1% and this patient wouldhave likely undergone initial duct excision had thealgorithm been applied appropriately. This patients sub-areolar ultrasound was not performed until the 6-monthfollow-up visit and it was abnormal at that time asexplained above. Together these findings demonstrate thatthe algorithm can prospectively identify the patients athighest risk for underlying malignancy.

The excellent sensitivity of this evaluation process, andin particular of subareolar ultrasound, has nearly eliminatedductography from the diagnostic evaluation of patients withnipple discharge in our practice. Only 13 ductograms wereperformed among this cohort and all were performed atanother institution before presentation at our facility. Thesensitivity and specificity of subareolar ultrasound wereequal or superior to that of ductography in this cohort(Table 2) and as previously reported.36 Therefore, if reli-able breast sonographers are available, we do not recom-mend routine diagnostic ductography because it is a muchmore uncomfortable procedure. It is important to notethat patients in this series were evaluated by experiencedbreast imagers, which may limit the reproducibility of theseresults among less experienced radiologists, especially inthe user-dependent application of subareolar ultrasound.Breast MRI was used sparingly in this cohort, eitherbecause of mammographically occult disease or becauseit had been obtained at another institution. There weretoo few MRIs among this cohort to draw any conclusionsabout its usefulness in nipple discharge. Barring substantialadditional data, it would be hard to justify the addition ofsuch an expensive examination into a diagnostic algorithmthat already performs well.

Patients who made an informed choice for close clinicalfollow-up underwent physical examination and subareolarsonography and/or mammography every 6 months for 1 to2 years or until the discharge resolved, whichever camefirst. Examining the results of this and the previous study14

cohorts, we recommend that the follow-up imaging includesubareolar ultrasound every 6 months as this is more sensi-tive than mammography, despite its lower specificity.Mammography can be done yearly according to screeningguidelines.

Despite the low risk for subsequent detection ofcarcinoma in the patients followed clinically over a medianperiod of 28 months, a substantial number of patients (20%)chose to undergo subareolar duct excision for symptomaticrelief. There are no clear findings we could identify thatcould predict patients who would subsequently choosesubareolar excision. On the other hand, among the patientswho never underwent operation, 81% had documented

A. Ashfaq et al. Treatment algorithm for nipple dischargeresolution of their discharge. Thus, the majority of low-riskcarcinoma as the cause for their nipple discharge, mosthave early stage, lower risk disease and they can expect agood outcome. The one patient with advanced disease wasable to be immediately identified through physical exam-ination and ultrasound findings, despite a normal mammo-gram, and has had a good outcome to date as well.

In conclusion, the implementation of an evidence-basedevaluation and treatment algorithm for patients with nippledischarge successfully stratified patients into the low-riskand high-risk categories. Low-risk patients were safelyfollowed clinically, whereas high-risk patients had a 23%rate of cancer on subareolar duct excision. By using thisapproach, one can not only select out the high-risk patientsfor intervention but also spare 66% of patients with nippledischarge an unnecessary operation.

References

1. Gulay H, Bora S, Kilicturgay S, et al. Management of nipple

discharge. J Am Coll Surg 1994;178:4714.

2. Newman HF, Klein M, Northrup JD, et al. Nipple discharge.

Frequency and pathogenesis in an ambulatory population. N Y State

J Med 1983;83:92833.

3. Adepoju LJ, Chun J, El-Tamer M, et al. The value of clinical charac-

teristics and breast-imaging studies in predicting a histopathologic

diagnosis of cancer or high-risk lesion in patients with spontaneous

nipple discharge. Am J Surg 2005;190:6446.

4. Dawes LG, Bowen C, Venta LA, et al. Ductography for nipple

discharge: no replacement for ductal excision. Surgery 1998;124:

68591.

5. Gioffre Florio M, Manganaro T, Pollicino A, et al. Surgical approach

to nipple discharge: a ten-year experience. J Surg Oncol 1999;71:

2358.

6. King TA, Carter KM, Bolton JS, et al. A simple approach to nipple

discharge. Am Surg 2000;66:9605; discussion, 9656.

7. Locker AP, Galea MH, Ellis IO, et al. Microdochectomy for single-

duct discharge from the nipple. Br J Surg 1988;75:7001.

8. Murad TM, Contesso G, Mouriesse H. Nipple discharge from thepatients benefit from avoiding initial operation becausetheir discharge resolves or never is symptomatic enough towarrant intervention.

For patient who undergo subareolar duct excision, eitherfor definitive diagnosis or for treatment of symptomaticdischarge, we prefer major duct excision. Major subareolarduct excision has been shown to detect a higher percentageof occult carcinoma than microdochotomy9 (excision of asingle ductal system with preservation of remaining ductalsystems in continuity with the nipple), result in fewer pa-tients requiring repeat duct excision,9 and to be associatedwith a 0% rate of breast cancer diagnosis over a subsequent5 years.11 The exception to this is for women planningpossible future breast feeding in whom microdochotomyis preferred.

The risk of carcinoma among the entire cohort was 5%,including 4% among women and 50% among 2 men. Amajority of patients with malignancy had DCIS. After amedian follow-up of 46 months, all patients with carcinoma

5breast. Ann Surg 1982;195:25964.

9. Sharma R, Dietz J, Wright H, et al. Comparative analysis of minimally

invasive microductectomy versus major duct excision in patients with

pathologic nipple discharge. Surgery 2005;138:5916; discussion,

5967.

10. Simmons R, Adamovich T, Brennan M, et al. Nonsurgical evaluation

of pathologic nipple discharge. Ann Surg Oncol 2003;10:1136.

11. Nelson RS, Hoehn JL. Twenty-year outcome following central duct

resection for bloody nipple discharge. Ann Surg 2006;243:5224.

12. Sakorafas GH. Nipple discharge: current diagnostic and therapeutic

approaches. Cancer Treat Rev 2001;27:27582.

13. Sauter ER, Schlatter L, Lininger J, et al. The association of bloody

nipple discharge with breast pathology. Surgery 2004;136:7805.

14. Gray RJ, Pockaj BA, Karstaedt PJ. Navigating murky waters: a mod-

ern treatment algorithm for nipple discharge. Am J Surg 2007;194:

8504; discussion, 8545.

15. Fiorica J. Nipple discharge. Obstet Gynecol Clin North Am 1994;21:

45360.

16. Isaacs J. Other nipple discharge. Clin Obstet Gynecol 1994;37:898902.

17. Peters F, Schuth W. Hyperprolactinemia and non-puerperal mastitis

(duct ectasia). JAMA 1989;262:161823.

6 The American Journal of Surgery, Vol -, No -, - 2014

Validation study of a modern treatment algorithm for nipple dischargeMethodsResultsCommentsReferences