The investigation and management of unilateral nipple discharge

Abstract

Introduction

Between 16,000 and 48,000 women are estimated to present to UK breast clinics with nipple discharge each year. The incidence of malignancy in these women is 2.7–24.2%. Currently, there is no consensus on the best way to investigate and manage these women. The aim of this study was to assess the rate of malignancy in women presenting with unilateral nipple discharge, and to evaluate the role of examination, imaging and cytology in reliably predicting outcome.

Methods

Breast units were asked to prospectively collect data on all new patients with unilateral nipple discharge. Data collected included discharge colour, whether it was uniductal or multiductal, examination and imaging findings, cytology results and outcome.

Results

Complete datasets were submitted by 5 units on 228 patients. The incidence of malignancy was 4.4%. Clinical examination was valuable in detecting malignancy and multiductal discharge was not related to malignancy. The positive predictive value for detecting malignancy for an abnormality found on mammography was 53.5% and for ultrasonography, it was 65.2%. The role of cytology in detecting malignancy was inconclusive with positive predictive values of the presence of red blood cells and epithelial cells at 6.1% and 10.7% respectively.

Conclusions

A large number of women are investigated for nipple discharge (with huge resource implications) but there is little reliable evidence on the best way to investigate and manage these patients. A larger study is needed to evaluate the role of investigations in nipple discharge to produce guidelines on optimal management.

Introduction

In England, there were 541,260 new referrals to breast units in 2017–2018¹ and over 46,000 women had a new diagnosis of breast cancer.² Three to nine per cent of new referrals to a breast clinic are for nipple discharge,^3,4 which would equate to approximately 16,200–48,700 patients per year. Pathological causes of nipple discharge include duct ectasia and intraductal papilloma but malignancy needs to be excluded. Reports of the incidence of malignancy in patients presenting with discharge vary widely in the literature, from 2.7% to 24.2%.^3–6 However, there is no consensus among clinicians on the best way to investigate and manage nipple discharge. Although the Association of Breast Surgery updated its guidelines in 2019,⁷ these are based on poor-quality evidence.

Most clinicians use a combination of clinical examination, imaging and cytological analysis of nipple discharge to investigate these women. Enormous resources are required for this large number of patients. On average, mammography costs £92 per patient, ultrasonography costs £52 per patient and cytology costs approximately £25 for analysis of two slides. There are also the costs of outpatient clinic appointments and further treatment such as surgery (either to aid diagnosis, for symptomatic relief or as definitive cancer treatment).

A meta-analysis of eight studies, which included 3,110 women presenting with nipple discharge, found that 24.8% of women with bloody nipple discharge were diagnosed with cancer compared with 12.1% of women with non-bloody discharge.⁵ These results are echoed in a study of 848 patients that showed that red/brown colour of nipple discharge was associated with a 36.1% breast cancer rate compared with a 9.5% incidence with clear discharge and an incidence of 11.7% with other colours of discharge.⁸ These studies were based on the gross macroscopic appearance of the nipple discharge and would indicate that bloody discharge is a useful predictor of breast cancer risk. Nevertheless, if clinicians only investigated women with bloody nipple discharge, they would potentially miss the 9.5–12.1% of women with breast cancer who present with non-bloody discharge.

Several studies have looked at the role of nipple discharge cytology in predicting pathology. A cohort of 618 patients with uniductal discharge showed that cytology only detected 5 of 36 malignancies and 84 of 127 benign lesions.⁹ It was concluded that nipple discharge cytology had a very low complementary diagnostic value in both benign and malignant disease. A smaller study of 89 patients demonstrated that positive cytology correlated with the pathological findings in 68% of cases but was non-concordant in 23.6% of cases and negative cytology missed a pathological diagnosis in 62.5% of cases, all of which were benign.¹⁰

The College of American Pathologists interlaboratory comparison programme sent out nipple discharge slides to 2,506 pathologists for review.¹¹ Two hundred and twenty responses were different to the agreed diagnosis for the slides sent, with a false positive/suspicious rate of 12.8% and a false negative rate of 3.4%. This confirms the difficulty pathologists have in assessing nipple discharge cytology on a single slide with little clinical information and the paper’s authors concluded that cytology had little value in the evaluation of nipple discharge.

Imaging is thought to be a vital tool in the assessment of nipple discharge. Studies show that 7–27.7% of patients with discharge have an abnormality on mammography^10,12,13 but it has a relatively poor sensitivity of only 18% in detecting cancer in patients with nipple discharge¹⁴ and negative mammography does not exclude a malignancy. Ultrasonography detects 15.1–63% of mammographically occult malignant lesions,^13,15 and certainly plays a useful role in distinguishing between benign intraductal papillomas and non-mass lesions such as inspissated secretions with a reported sensitivity of 100% and a specificity of 82.4%.¹⁴

The majority of papers in the literature regarding nipple discharge are retrospective reviews following surgical excision (either a total duct excision or microdochectomy) in relatively small numbers of patients. The incidence of breast cancer in these studies ranges from 2.4% to 23.9%.^16–19 Blood stained nipple discharge was found to be a statistically significant predictor of malignancy in one study¹⁵ but thought to be a non-statistically proven predictor of malignancy in two others.^16,17 None of the studies were able to find a statistically proven association between clinical examination, imaging findings, cytology and histological diagnosis.

The difficulty in evaluating the usefulness of these studies to aid decisions regarding the best form of investigation is that they do not include all cases of nipple discharge, usually excluding those who did not undergo surgical excision. Not all have analysed data on clinical, cytological and imaging findings to assess whether a predictor of pathological outcome can be identified.

These figures illustrate that a large number of women presenting to the breast clinic are being investigated for benign disease, in a manner that is not based on good quality evidence and at an enormous cost to the National Health Service (NHS). The aim of this study was to assess the incidence of malignancy and papilloma in women presenting to the breast clinic with unilateral nipple discharge, and to evaluate the role of imaging and cytology in making a diagnosis.

Methods

Breast units in South West England undertook a prospective analysis of all women presenting to the breast clinic with unilateral nipple discharge over a six-month period. Data were collected on type of discharge, which was classified as clear, bloody, milky, yellow or green as described by the clinician or from the history. The results of any cytology, imaging and core biopsy were analysed, and the final outcome and/or pathological diagnosis was recorded. Each unit logged the study with its local audit department and all data were anonymised.

Results

Complete datasets were submitted by 5 breast units on 228 women who had presented with unilateral nipple discharge. There was wide variation in the number of patients included in the study by each unit. The mean age of patients was 49.2 years (range: 18–95 years). The incidence of malignancy was 4.4% (n=10) and the incidence of papilloma was 5.7% (n=13). The mean age of women with malignancy was 60.9 years (range: 39–84 years).

Clinical examination

The colour of the nipple discharge was taken from the history or the clinician’s examination findings (Table 1). Approximately a third of women (n=83) presented with bloody discharge, which had a positive predictive value for malignancy of 4.9%. In seven women, the colour of the discharge had not been recorded. Malignancies were seen in women presenting with all colours of discharge except yellow but were most common in women with bloody discharge. The rate of malignancy was similar for those with bloody discharge and those with non-bloody discharge (4.8% vs 4.1% respectively). Non-bloody discharge had a negative predictive value for malignancy of 93.1%. The colour of the discharge was not related to finding a papilloma.

Table 1.

Incidence of malignancy and papilloma related to nipple discharge colour

Discharge colour	n	Malignancy	Papilloma
Bloody	83	4	4
Clear	59	3	6
Milky	34	1	0
Yellow	25	0	3
Green	19	1	0
Unknown	7	1	0

The majority (64.5%) of women presented with uniductal discharge with a positive predictive value for malignancy of only 4.1%. In 48 cases, it was unknown whether the discharge was uniductal or multiductal. This was generally because there was no discharge produced in clinic rather than owing to the data not being recorded. There were no malignancies in women presenting with multiductal nipple discharge, which had a negative predictive value for malignancy of 100%. All of the women with papillomas presented with uniductal discharge, which is to be expected.

The results for clinical examination are shown in Table 2. Nearly 75% of women had no other findings except nipple discharge on examination. All but one of those with a malignancy did have clinical signs noted at examination. However, this did not discriminate from those without a malignancy. An abnormal examination finding had a positive predictive value of detecting malignancy of 15.5%. As expected, those with a papilloma usually had a normal examination. The negative predictive value of a normal examination was 99.4%.

Table 2.

Incidence of malignancy and papilloma related to examination findings

Examination findings	n	Malignancy	Papilloma
No abnormality detected	170	1	11
Lump	21	3	1
Nodularity	12	2	1
Nipple inversion/flattening	7	2	0
Thickening	5	1	0
Erythema	3	0	0
Nodule	3	0	0
Glandular asymmetry	2	1	0
Eczema	2	0	0
Nipple swelling	1	0	0
Mammary fistula	1	0	0
Implants	1	0	0

Imaging

Almost half (40.2%) of the patients had an abnormality seen on mammography. There were 81 women who were aged <40 years and of these, only 2 underwent mammography; in these women, 1 had a cyst visible and 1 had M5 calcification. One hundred and forty-seven women were aged >40 years. Of these, 125 had mammography. All ten of the patients with a malignancy had an abnormality visible on mammography, with eight graded as M3 or above. The positive predictive value of abnormal mammography detecting malignancy was 53.3% and the negative predictive value of normal mammography was 98.2%.

The majority of women (n=171) underwent ultrasonography. Just under half (43.3%) had an abnormality but only one malignancy did not show on this modality. Six of the thirteen papillomas were visible on ultrasonography although four women with a papilloma did not have ultrasonography. The positive predictive value for malignancy of abnormal ultrasonography was 65.2% and the negative predictive value for normal ultrasonography was 99.3%. The results for the mammography and ultrasonography findings are shown in Table 3.

Table 3.

Incidence of malignancy and papilloma related to mammography and ultrasonography findings

Imaging findings	n	Malignancy	Papilloma
Mammography
M1	77	0	5
M2	36	2	4
M3	7	2	1
M4	3	1	1
M5	5	5	0
Not performed	101	0	2
Ultrasonography
U1	97	1	3
U2	51	0	1
U3	17	2	5
U4	1	1	0
U5	5	5	0
Not performed	57	1	4

Table 4.

Positive and negative predictors for malignancy

Predictor		Malignancy	No malignancy	Predictive value
Examination	Abnormal	9	49	PPV 15.5% NPV 99.4%
Normal	1	169
Mammography	Abnormal (M3, M4, M5)	8	7	PPV 53.3% NPV 98.2%
Normal (M1, M2)	2	111
Ultrasonography	Abnormal (U3, U4, U5)	8	15	PPV 65.2% NPV 99.3%
Normal (U1, U2)	1	147
Discharge colour	Bloody	4	78	PPV 4.9% NPV 93.1%
Non-bloody	6	81
Ductal source	Uniductal	6	141	PPV 4.1% NPV 100%
Multiductal	0	33
Red blood cells	Present	2	31	PPV 6.1% NPV 96.9%
Absent	4	127
Epithelial cells	Present	3	25	PPV 10.7% NPV 97.1%
Absent	3	100

NPV = negative predictive value; PPV = positive predictive value

Cytology

One unit did not routinely send samples of nipple discharge for cytology although it did in occasional cases. This unit contributed data on 57 women. Some units sent samples for cytology on all women and others performed this variably. The presence of red blood cells and epithelial cells are thought to be most likely to indicate the presence of pathology, and one or both of these were present in the five women who had cytological analysis and were found to have a malignancy and in the nine women who had a papilloma. However, 47 women had 1 of these 2 cell types without any pathology being identified.

Twenty-six patients had a second sample sent for cytology. The results for the first and second sample were the same in five patients. Nine women had red blood cells with/without epithelial cells in the first sample but of these, only four had the same results in the second sample. Twelve patients had either macrophages, squames or an acellular sample at first assessment but repeated cytology found red blood cells with/without epithelial cells in the second sample of five women. Of these, three had microdochectomies in which two papillomas were found.

The unit that did not routinely send samples for nipple discharge cytology had the same incidence of malignancy and papilloma as those units that did. The positive predictive values for detecting malignancy for red blood cells and epithelial cells were 6.1% and 10.7% respectively.

Patients without discharge present at clinic visit

There were 40 women seen in clinic with a history of nipple discharge but no discharge on examination. Malignancies were found in three of these patients and a papilloma in one. On examination, while 32 had a normal examination, 2 of the patients had nipple changes and 1 had glandular asymmetry; all 3 had abnormalities on imaging. Only five women did not undergo any form of imaging. The rest had either mammography, ultrasonography or both and were then discharged or offered further follow-up appointments.

Outcome

One hundred and fifty women were reassured and discharged after clinical examination and investigation although it is not known whether this was after one or more clinic visits. Thirty-four women underwent either major duct excision or microdochectomy. Of these, 1 had a malignancy, 12 had a papilloma and the remaining 21 had benign disease, most commonly ectatic ducts and periductal mastitis. The ten women with a malignancy had formal cancer resection; two of these had initial surgery to make the diagnosis (1 microdochectomy, 1 excision biopsy). At the time of data collection, 5 women were awaiting ductal surgery, 1 had declined ductal surgery, 3 had been referred elsewhere, 2 had undergone vacuum assisted biopsy, 5 had been given self-referral clinic appointments for ongoing symptoms and 17 had declined or had failed to attend follow-up visits. One patient was found to have a contralateral cancer.

Discussion

Most clinicians probably feel that they know how they should investigate nipple discharge but many might feel uncomfortable discharging patients without a diagnosis and may offer repeated follow-up appointments for such patients. This could be construed as a waste of valuable clinic time, and an evidence-based algorithm for the diagnosis and management of these patients could reduce these unnecessary appointments. Although this study has concentrated on incidence of malignancy and papilloma, it should be remembered that for some women, nipple discharge is a distressing and difficult symptom to manage, and there is a role for surgery for symptomatic relief. This study did not look at the decision-making process in clinic in terms of who is offered surgery for symptomatic relief and there are no reported data in the literature on this aspect but it is an important factor to consider when studying this group of patients.

The number of patients included in the study was much lower than expected and the study population was too small to perform any meaningful statistical analysis. It is obvious that not all women presenting to each unit with unilateral nipple discharge were included in the study as the number of patients from some units was very low. However, from the data available, we have tried to identify patterns and evaluate the role of the different modalities in investigating nipple discharge.

Clinical examination is vitally important. Not only does this give information on the colour of the discharge and whether it is uniductal or multiductal but all apart from one of the patients with a malignancy had physical signs of either a lump, nipple change, nodularity or glandular asymmetry. Any of these patients who had presented to the breast clinic without nipple discharge would have undergone standard triple assessment of these findings. Nevertheless, clinical examination was not useful in the detection of papillomas.

There were no patients who presented with multiductal discharge who had either malignancy or a papilloma and in such patients, it may be possible to forego further investigation. There is little recorded in the literature on the value of clinical examination findings in relation to the investigation of nipple discharge. Investigators were asked to classify discharge that was red, brown or black as bloody. This may overestimate the number of patients with bloody nipple discharge but from history and clinical examination, it is difficult to be accurate regarding the presence of blood. Future study could test for blood in a clinic room using urine dipsticks. This should be addressed in further work and could be a confounding factor. The present study did not look at the duration of nipple discharge and prolonged symptoms may be an indicator of malignancy or papilloma.

Chen et al’s meta-analysis showed a rate of malignancy in cases of bloody nipple discharge of 24.8% compared with 12.1% in non-bloody discharge.⁵ Our results do not echo this as the two groups had roughly equal rates of malignancy at approximately 4%. However, our data were taken from both the clinicians’ observation and the patient history when no discharge was seen in clinic, which may affect their accuracy.

In the present study, 40.2% of women had an abnormality on mammography and 43.3% had an abnormality on ultrasonography, which is much higher than in the literature.^9,13,14 All malignancies were seen on either mammography or ultrasonography. Unsurprisingly, ultrasonography was much more sensitive at detecting papilloma than mammography. This illustrates the importance of these investigations in women with unilateral nipple discharge.

There is wide discordance in the literature regarding the value of nipple discharge cytology.^9–11 Our results show that it is unreliable in predicting the presence of malignancy or papilloma. The lack of agreement between first and second cytology samples demonstrates the unreliability of discharge cytology. At the cost of nearly £25 for analysis of two slides, omitting this in the 16,000–48,000 women presenting to breast clinics across the UK could save the NHS between £400,000 and £1.2 million. As the unit that did not routinely send samples for cytology had equivalent rates of malignancy and papilloma to those that did, it would seem that it would be safe to omit this investigation.

It is difficult to know what to do with those patients who present in clinic with a history of nipple discharge but do not have any on examination. There were three malignancies in this group of patients, all with abnormalities on examination and imaging. None of the patients with a normal examination had a malignancy or papilloma, which leads to the question: In those patients with a normal clinical examination, would it be safe to omit imaging? Long-term follow-up of these patients is needed to answer this (which has implications for resources) so a one-stop clinic visit with imaging may negate the need for extra clinic visits and detect the occasional papilloma.

All of the women in the study presented with nipple discharge but it is difficult to assess whether the discharge is caused by malignancy or papilloma, or whether it is an unrelated symptom. However, one of the aims of this study was to attempt to identify women in whom it is safe to not perform further investigations. The small numbers in the study have not allowed this but there are indications that women with multiductal discharge may not require further investigation.

Conclusions

Our study shows an incidence of malignancy of 4.4%, which is lower than incidences published in the literature. Nevertheless, few reported studies have included all women presenting with nipple discharge. Although the numbers in this study were small and it is therefore not possible to form any meaningful conclusions, it does highlight some valid points:

• The current evidence for the value of different types of investigation of nipple discharge is limited.

• Clinical examination is important and other signs of malignancy are often present in this group of patients.

• Multiductal discharge does not appear to be a sign of malignancy or papilloma and it may be possible to omit investigation in this group of patients.

• Imaging is good at detecting abnormalities and should be performed in all patients with unilateral, uniductal nipple discharge.

• Cytology may not add much to imaging and examination findings.

Further work is needed in this area. The NUND (National Uniductal Nipple Discharge) study launched in October 2019 with the aim of producing evidence-based guidelines for the investigation and management of unilateral nipple discharge. It is important to emphasise that until good quality evidence becomes available, clinicians should continue to perform standard investigations for nipple discharge.