missed breast carcinoma; why and how to avoid?...carcinoma detection. it was diagnostic in 138...

ABSTRACT

Introduction: Despite the advances in mammographytechniques, it still has a number of limitations. It isestimated that about 10 to 25% of lesions are overlookedin mammograms out of which about two thirds are detectedretrospectively by radiologists and oncologists. Causesof missed breast cancer on mammography can be secondaryto many factors including those related to the patient(whether inherent or acquired), the nature of the malignantmass itself, poor mammographic techniques, providerfactors or interpretive skills of radiologists and oncologists(including perception and interpretation errors).

Aim of Work: The aim of this study is to investigatethe aforementioned factors hindering early breast cancerdetection and in turn lowering mammographic sensitivityand to outline the major guidelines to overcome thesefactors aiming to an optimum mammographic examinationand interpretation by radiologists and oncologists.

Subject and Methods: We conducted this multicenterstudy over a two-year interval. We included 152 histo-pathologicaly proven breast carcinomas that were initiallymissed on mammography. The cases were subjected tomammography, complementary US, MRI and digitalmammography in some cases and all cases were histo-pathologically proven either by FNAB, CNB or openbiopsy.

Results: Revision of the pathological specimens ofthese 152 cases revealed 121 infiltrating ductal carcinomas,2 lobular, 4 mucinous, 14 inflammatory carcinomas, 6carcinomas in situ (3 of which were intracystic), 2 intra-ductal papillary carcinomas and 3 cases with Paget'sdisease of the nipple. In analyzing the causes responsiblefor misdiagnosis of these carcinomas we classified theminto 4 causative factors; patient, tumor, technical or pro-vider factors. Tumor factors were the most commonlyencountered, accounting for 44.1%, while provider factorswere the least commonly encountered in 14.5%. Carcino-

Journal of the Egyptian Nat. Cancer Inst., Vol. 19, No. 3, September: 178-194, 2007

Missed Breast Carcinoma; Why and How to Avoid?

RASHA M. KAMAL, M.D.; NAGLAA M. ABDEL RAZEK, M.D.; MOHAMED A. HASSAN, M.D. andMOHAMED A. SHAALAN, M.D.*

The Departments of Radiodiagnosis, Cairo University and General Surgery*, National Cancer Institute.

178

mas were detected using several individual or combinedcomplementary techniques. These techniques mainlyincluded double reading, additional mammography views,ultrasound and MRI examinations. Forty four carcinomaswere detected on double and re-reading by more experi-enced radiologists. Additional mammographic views wererecommended in 35 (23%) cases. Complementary ultra-sound examination was performed for all 152 cases (100%)and showed a higher sensitivity than mammography incarcinoma detection. It was diagnostic in 138 (90.8%)cases only. In the remaining 14 cases, further MRI andbiopsy were performed.

Conclusion: Why can breast carcinoma be missed?Four main factors are responsible for missing a carcinoma:(1) Patient factors (Inherently dense breasts or acquireddense breasts). (2) Tumor factors (subtle carcinoma,masked carcinoma, multifocal carcinoma and multicentriccarcinoma). (3) Technical factors (bad exposure factors,malpositioned breasts and bad processing quality). (4)Provider factors (bad perception and misinterpretation).

How to avoid missing a breast carcinoma? Reviewclinical data and use US and other adjunct techniques asMRI and biopsy to assess a palpable or mammographicallydetected mass. Be strict about positioning and technicalfactors. Try to optimize image quality. Be alert to subtlefeatures of breast cancers. Always consider the well definedcarcinoma. Compare current images with multiple priorstudies to look for subtle increases in lesion size. Lookfor other lesions when one abnormality is seen. Judge alesion by its most malignant features. Double reading andthe use of computer aided diagnosis (CAD) and finallyFFDM (Full Field Digital Mammography). Close cooper-ation between the oncologist, radiologist and pathologistis essential to avoid missing any case of breast carcinoma.

Key Words: Missed breast carcinoma – Mammography –Ultrasonography – MRI.

INTRODUCTION

The primary role of any breast imagingmodality is breast cancer early detection. Thestandard of care for breast cancer screening of

Correspondence: Dr Rasha M. Kamal, Radiology Dept.Faculty of Medicine Cairo University, [email protected]

179

asymptomatic women is mammography [1]. Itis the only available screening method provento reduce breast cancer mortality [2]. Mammog-raphy plays a major role in early detection ofbreast cancers, detecting about 75% of cancersat least a year before they can be felt. Withearly detection, intervention and postoperativetreatment, breast cancer mortality has decreased[3]. This means that breast cancer can be detect-ed early enough to make a difference in thequality of life, disease progression and mortalityrates.

Although mammography is a major weaponin the war against breast cancer; unfortunatelyit is not perfect. It is a fact that most expertsbelieve it lowers the death rate from cancer byat least 20% [4], but this will not happen if themammograms are not done and are not inter-preted correctly. Despite the advances in mam-mography techniques, it still has a number oflimitations. It is estimated that about 10 to 25%of lesions are overlooked in mammograms, outof which about two thirds are detected retro-spectively by radiologists [5].

Causes of missed breast cancer on mammog-raphy can be secondary to many factors includ-ing those related to the patient (whether inherentor acquired), the nature of the malignant massitself, poor mammographic techniques, providerfactors or interpretive skills of radiologists(including perception and interpretation errors).

Aim of work:

The aim of this study is to investigate theaforementioned factors hindering early breastcancer detection and in turn lowering mammo-graphic sensitivity and to outline the majorguidelines to overcome these factors aiming toan optimum mammographic examination andinterpretation.

SUBJECTS AND METHODS

We conducted this study over a period oftwo years between December 2004 to December2006. We included 152 histopathologicaly prov-en breast carcinomas that were missed on mam-mography and detected by double reading and/orcomplementary imaging modalities includingultrasound and/or MRI examinations.

Symptomatic patients were referred mainlyfrom the Surgical and Oncology Departmentsand Outpatient Clinics of Kasr El Aini & Na-tional Cancer Institute to the "Women ImagingUnit" in the Radiodiagnosis Department of KasrEl Aini Hospital to perform a diagnostic mam-mogram. Another group of patients were referredfrom private clinics to the "Mammography Unit"in ALFA Scan Radiology Centre.

Diagnostic Mammograms were performedfor all cases under request of their treatingphysicians after full history taking and completeclinical examination. Mammograms were inter-preted by one senior (48 mammograms) orcombined junior and senior (104 mammograms)staff radiologists. Ultrasound was performedfor all cases as a complementary confirmatoryexamination tool to exclude or detect any un-derlying pathologies missed on mammograms.In selected cases, MRI examination was alsoperformed when ultrasound results were stillinconclusive.

Inclusion criteria included cases with falsenegative mammograms for malignancy initiallyassigned BIRADS 0 to BIRADS 3 score withnon concomitant ultrasound, MRI and conse-quently histopathological results. BIRADS 4and 5 cases were included only if multifocal,multicentric or contralateral carcinomas weremissed on mammograms.

Mammography examination:1- Mammography machines:

• Siemens Mammomat Mammography Unit.• ALPHA RT GE Mammography Unit.• Full Field Digital Mammography GE 2000

Unit.

2- Mammographic views:• Conventional views (Craniocaudal (CC)

and mediolateral oblique (MLO) views.• Additional views (extended CC and MLO,

spot compression, magnification Views)whenever indicated.

3- Exposure factors:• Compression is applied and respiration is

suspended on exposure.• 25-28KVP.• 30-60MAS.

Missed Breast Carcinoma

Rasha M. Kamal, et al.180

4- Mammography interpretation and report:

• Films were reviewed in a mirror imagepattern to compare similar areas.

• Well illuminated view boxes were used.

• Old films when available were reviewed.

• Parenchymal density was assessed.

• Skin thickness, nipple areola complex andassociated axillary lymph nodes were as-sessed and reported.

• Areas of parenchymal distortion and asym-metrical densities were considered.

• Mass lesions and calcifications were ana-lyzed and a BIRADS score was assignedfor detected lesions.

Breast imaging reporting and data system (BI-RADS) category [6,7]:

0- Assessment is incomplete. Need to reviewprior studies and/or complete additionalimaging studies.

1- Negative. Continue routine screening.

2- Benign finding. Continue routine screening.

3- Probably benign finding. Short term follow-up mammogram after 6 months, then every6 to 12 months for 1-2 years.

4- Suspicious abnormality, perform biopsy,preferably needle biopsy.

5- Highly suspicious of malignancy, appropriateaction should be taken. Biopsy and treatmentare necessary.

Ultrasound examination:

1- Ultrasound machines:

• Ultramark Philips HDI 5000.• Siemens Elegra.• General Electric Logic 7.

2- Technique of examination:

• Patient lied supine with the examined sideelevated.

• Examination extended from the clavicledown to the inframammary fold and fromthe sternum medially to the mid axillay linelaterally.

• The axillae were carefully scrutinized forlymph nodes enlargement.

MRI examination:

1- MRI machine:

• Philips Intera (Netherlands) 1.5 tesla Ma-chine.

2- Technique of examination:

• Patient lies prone.

• Breasts are placed within special designedbreast coils.

• Both breasts were examined in axial andsagittal planes.

• T1 and T2 WI and post contrast T1 fatsuppressed dynamic sequences were ob-tained.

• Time signal intensity curves were plottedfor detected enhancing lesions.

• Morphology and dynamics of detected le-sions were assessed.

RESULTS

The study included 152 women with histo-pathologicaly proven mammographicaly missedbreast carcinomas diagnosed by double reading,U/S and/or MRI examinations. The mean ageof the patients was 53.5 years (range 32 to 75years).

Ninety-one ladies were referred from theSurgical and Oncology Departments and Out-patient Clinics to the "Women Imaging Unit"in the Radiodiagnosis Department of Kasr ElAini Hospital to perform a diagnostic mammo-gram. Sixty one ladies were referred from pri-vate clinics to the "Mammography Unit" inALFA Scan Radiology Centre.

Revising the pathological specimens of these152 cases yielded 121 infiltrating ductal carci-nomas, 2 lobular, 4 mucinous, 14 inflammatorycarcinomas, 6 carcinomas in situ (3 of whichwere intracystic), 2 intraductal papillary carci-nomas and 3 cases with Paget's disease of thenipple.

Analyzing the causes responsible for misdi-agnosis of these carcinomas, we classified theminto 4 causative factors; patient, tumor, technicalor provider factors as shown in Table (1). Tumorfactors were the most commonly encountered,accounting for 44.1% (67 cases) while provider

181

factors were the least commonly encounteredin 14.5% (22 cases).

An initial BI-RADS score was assigned forthese cases as shown in Table (2). BIRADS 0(71 cases) included 34 inherently or acquireddense breasts (Figs. 1,2), 14 cases with missedsubtle areas of asymmetrical densities and pa-renchymal distortion (Fig. 3), 15 cases withdiffuse edema pattern, 3 cases with Paget'sdisease (Fig. 11) and 5 cases missed due totechnical factors (Figs. 12,15). BIRADS 1 scorewas assigned for 34 cases including 4 smallundetected subtle carcinomas, 21 malpositionedbreasts with deeply seated carcinomas (Fig. 13)and 9 missed carcinomas due to bad perceptionby unskilled radiologists. BIRADS 2 (23 cases)included 14 subtle well defined, 2 masked (Fig.5) and 7 misinterpreted carcinomas. BIRADS3 (9 cases) included 6 misinterpreted carcinomasand 3 badly processed mammograms (Figs.4,14). BIRADS 4 and 5 (15 cases) included 7multicentric and 3 contralateral carcinomas(Figs. 6,7) and 5 breasts with diffuse edemapattern with pathologicaly enlarged lymph nodes(Fig. 8).

Carcinomas were detected using severalindividual or combined complementary tech-niques in reference to Table (3). These tech-niques mainly included double reading, addi-tional mammographic views and postprocessingcapabilities available on digital mammographylike using the magnifying lens and the invertedimages (Figs. 9,16,17), ultrasound and MRIexaminations. Forty four carcinomas were de-tected on double and re-reading by more expe-rienced radiologists. Additional mammographicviews were recommended in 35 (23%) cases.Complementary ultrasound examination wasperformed for all 152 cases (100%) and showeda higher sensitivity than mammography in car-cinoma detection. It was diagnostic in 138(90.8%) cases only. In the remaining 14 cases,further MRI and biopsy were performed (Fig.10). These cases included 2 lobular, 2 mucinous,2 well defined infiltrating ductal carcinomas,3 papillary carcinomas and 3 cases with Paget'sdisease of the nipple.

The different mammographic and ultrasoundfeatures of missed carcinomas are described inTable (4).


Table (2): Initially assigned BIRADS score for 152 per-formed mammograms.

7134239510

152

No. of cases

0-1-2-3-4-5-

Total

Birads score

Table (3): Complementary techniques used in carcinomadetection.

68.42310011.8

Percentage%

Double & re-readingAddit. mammographic views *UltrasoundMRI

Imaging technique

1043515218

No.of cases

* Spot compression views were performed for 13 cases, postprocessing magnification using the digital machine lens for 7cases and extended CC and MLO views for 15 cases.

Table (1): Different factors responsible for missed carci-nomas and number of cases included undereach factor.

* Subtle carcinomas include 4 small (<3mm) carcinomas, 14well defined carcinomas and 10 carcinomas presenting withfocal asymmetrical densities and 4 with areas of parenchymaldistortion.

** Diffuse edema pattern was detected in 6 cases of infiltratingduct carcinoma and 14 cases of inflammatory carcinomas.

22.415.86.62.64

44.1211.34.6

213.22

193.213.82

14.55.98.6

100

Percentage%

34241046

673227

3203

295213

22913

152

No.of cases

Patient factors:• Inherent dense breast• Aquired dense breast

Hormonal therapyPost treatment breasts

Tumor factors:• Subtle carcinomas *• Masked Carcinomas• Multicentric and multifocal

carcinomas• Contralateral breast carcinoma• Diffuse edema pattern **• Paget's disease

Technical factors:• Bad exposure factors• Malpositioned breasts• Bad processing quality

Provider factors:• Bad perception• Misinterpretation

Total

Cause of missedcarcinoma


Fig. (1): A 51-year-old female with an inherently dense breastparenchyma showing an area of focal increased density withinthe left upper outer breast quadrant (A). An underlying small10x13mm carcinoma was detected on ultrasound examination(B). The message in this case is ''complementary US is essentialin dense breasts''.

Fig. (2): A 41-year-old patient under hormonal replacementtherapy with an acquired dense breast parenchyma on mammog-raphy (A). A giant left intramammary mass lesion (B) withinfiltrated axillary nodes (C) were detected on complementaryultrasound examination.

Table (4): Mammography and ultrasound features ofmissed carcinomas.

1111422-203

152

Ultrasound

Typical mass *Atypical mass **No mass detectedParenchymal distortionFocal asymmetric densitiesDiffuse edema patternSkin thickening

Total

Feature

301453528203

152

Mammography

* Typical malignant mass lesions are considered when massesshowed microlobulated or spiculated outlines, with markedsurrounding distortion +/- clustered microcalcification andpathologically enlarged lymph node.

** Atypical mass lesions included 9 well defined and 3 intracysticand 2 intraductal carcinomas.

(A)

(B)

(A)

(B)

(C)

183Missed Breast Carcinoma

Fig. (3): Subtle left intra mammary areas of parenchymal distortion (A). On ultrasound examination underlyingmultifocal, small, few mm carcinomas were detected.

Fig. (4): A 46-year-old female with a well defined right retroareolar mass lesion. Her mammogram (A) wasassigned a BI-RADS 3 score. On ultrasound examination an intracystic fungating mass lesion was detected.

(A) (B)

(A) (B)

Fig. (5): A calcified fibroadenoma masked an adjacent carcinoma(arrow). The message is ''one should avoid the satisfaction of search''.


Fig. (6): A young 28-year-old female complaining of mastalgia with multiple palpable left intramammary mass lesions. Her initialmammogram (A) was assigned a BI-RADS 3 score. On ultrasound (B) examination a deeply seated carcinoma was detected. Re-ray ofthe patient after adequate breast pulling (C) was performed and the US detected carcinoma was detected (arrow). Biopsy proved themalignant nature of the lesions and the case was upgraded to BI-RADS 5 with multicentric carcinoma.

Fig. (7): A large right retroareolar well mass lesion detected on the mammogram (A) of a 70-year-old female. On ultrasoundexamination (B) a large intra-ductal papillary carcinoma was detected. Clustered microcalific foci (arrow) were missed on the contralateralleft breast mammogram (C) seen only in the post processing magnification view (D). An underlying carcinoma was detected on ultrasoundexamination (E). Again, the message is the radiologist should avoid the satisfaction of search.

(A) (B) (C)

(C) (D) (E)

(A) (B)


Fig. (8): A case of inflammatory carcinoma showing diffuse edema pattern of the right breast (A). Onultrasound examination, marked underlying interstitial edema was detected with overlying skin thickening (B)associated with right axillary pathological lymphadenopathy (C).

Fig. (9): A 40-year-old female with a palpable left breast mass. Her initial mammogram was assigned a BI-RADS 3 score. The spiculated outline of the mass (missed by a junior radiologist) was more evident on the postprocessing magnification view of the digital machine unit and complementary ultrasound examination.

(A)

(B) (C)

(B)(A)


Fig. (10): A well defined macrolobulated outlined mass lesion misinterpreted in both mammography andultrasound as benign. A BI-RADS 3 score was assigned for the patient. Excisional biopsy of this mass revealedan underlying infiltrating ductal carcinoma.

Fig. (11): (A) Negative mammogram in a 44-year-old presented with nipple ulcer. The ulcer was biopsiedand proved to be Paget's disease of the nipple. The patient was referred for US (B&C) and a small hypoechoiclesion fixed to the areola is seen. The dermal defect corresponding to the ulcer is appreciated in the spectralhigh resolution ultrasonography (arrow). The message is complementary US as well as the patient's complaintand clinical findings are very important to reach a correct diagnosis.

(B)(A)

(B)(A)

(C)


Fig. (12): Right axillary tail irregular mass not seen in the CC views and only projected in the MLO (A&B). The case was missedin the initial mammogram that was done without adequate pulling in the MLO views. Targeted US (C) confirmed the presence of a solidhypoechoic malignant looking mass (BIRADS 5) in the axillary tail of the left breast.

Fig. (13): A 43 -year- old lady with a palpable lesionat the right upper outer quadrant (A) multiformat CRfilm combining the CC & MLO views of the right breastand the associated complementary US images showingnegative findings (This was done outside the centers ofthe study and the case was referred to one of the authorsfor revision). The author decided to repeat the mammo-gram because the posterior third of the breast is notshown in the images. Bilateral CC & MLO (B & C)views after good positioning showing two irregularspiculated lesions in the right upper outer quadrant. Highresolution superficial complementary US images (D &E) showing one of the two spiculated masses as well asthe pathologically enlarged lymph nodes. The case wasdiagnosed as multifocal invasive duct carcinoma wasmanaged by radical mastectomy.

(A) (B) (C)

(A) (B)

(C) (D)

(E)


Fig. (14): A small well defined lesion in the right upper outer quadrant was discovered on screening mammography (A) and confirmedon the targeted US image (B). It was classified as BIRADS 3. Unfortunately, the lesion was diagnosed by histopathology after openbiopsy as medullary carcinoma. The message is that ''some malignant breast lesions present as a well defined nodule and so biopsy isrecommended in all suspicious cases''.

Fig. (15): A 37-year-old female with positive family history. Repeated CC mammographic view(A) after adjusting the exposure factors resulting in clear demonstration of an irregular spiculated mass(BIRADS 5) that was masked in the first examination (B) with low exposure factors. ComplementaryUS image (C) revealed the solid hypoechoic nature of the lesion that was pathologically proved to beinvasive duct carcinoma.

(A) (B)

(C)

(A) (B)


Fig. (16): CC & MLO digital mammographic views showing such a small malignant looking lesion with pathological microcalcificationonly appreciated by the use of magnifying lens (A), a post processing advantage available on digital mammography.

Fig. (17): (A) Inverted digital mammographic MLO of both breasts showing a small left breast irregular lesion appreciated in theinverted image (arrow) and in the spot magnified image (B).

(A) (B)

DISCUSSION

Since mammography was introduced as apart of preventive care, breast cancer is mostoften diagnosed at its early asymptomatic stageas a non-palpable suspicious lesion on mam-mography [8]. Early detection of breast canceris vital to decrease the mortality and morbiditycaused by the disease. Although currently mam-mography is the best screening modality avail-able, unfortunately it is both underused andover-rated [8].

It is estimated in previous studies that only40% of high-risk women undergo regular screen-ing in the United States, despite the fact that itreduces mortality by 30% in women 50 yearsand older [9-11]. Moreover, despite a high sen-

sitivity and specificity, mammography can missdeadly tumors [8]. According to the data fromthe Breast Cancer Detection DemonstrationProject, the false negative rate for screeningmammography may be as high as 8-10% [12,13].Since then, other authors have suggested aneven higher rate (up to 25%) of missed breastcancers during screening mammography [14].

Screening mammography is used to detectclinically occult breast carcinomas, while diag-nostic mammography is performed to assesssymptomatic patients or to further evaluate anabnormality detected on screening mammogra-phy [15]. Up to date, there are no validatedresults to the national screening program inEgypt and thus we conducted this study ondiagnostic mammograms of symptomatic ladies

(A) (B)


complaining of palpable breast masses, nippledischarge, mastalgia, skin changes or signs dueto distant metastases.

Causes of missed breast cancers can relateto a variety of factors, including those relatedto the patient, to the nature of the tumor itself,to technical factors and to provider factors. Wedefined a false negative mammogram as thatassigned an initial BI-RADS score of 0 to 3with non concomitant MRI and ultrasound find-ings or those assigned as Bi-RADS 4 and 5 withundetected multicentric, multifocal or contralat-eral carcinomas on mammography. Histopatho-logical examination of these masses was con-sidered as our mark reference.

To improve the detection of breast cancer,various imaging modalities are being used asadjuncts when a mammogram has abnormal orinconclusive results [15]. Complementary ultra-sound was performed for all patients (100%)on the same day of mammography examination.We performed MRI examination for 18 ladies(11.8%) with inconclusive ultrasound examina-tion yet highly suspicious clinical complaint.Additional mammographic views were per-formed for 35 cases (23%). Carcinomas weredetected on double or re-reading by more expe-rienced senior staff radiologists in 104 cases(68.4%).

'Patient factors' accounted for 22.4% of cases(34 patients), mainly attributed to increasedbreast parenchyma density whether inherent oracquired following surgery, hormonal, radio orchemotherapy. A nearly equal percentage (24%)was recorded in previous studies performed byBird et al. [14] in a screening population. Fajardo,et al. [16] demonstrated that the radiologist'scertainty of interpretation of mammograms isinversely proportion to the breast density andcomplexity of images. Dense breasts may ob-scure breast tumors and make interpretationmore difficult, resulting in decreased sensitivityand specificity of mammography [17,18]. Missedcarcinomas in dense breasts are also usuallynon calcified and non distorting [19]. Malignantlesions could sometimes be missed in densebreasts due to technical factors, including re-duced image contrast and unsharpness [20]. Formammographic dense breasts we strongly rec-ommend that examinations are performed ondigital machines. Adequate compression should

be applied. Thorough comparative examinationof both breasts should be done on well illumi-nated view boxes on which images of bothbreast are viewed as mirror image. Enough timeshould be spent looking for subtle areas ofarchitectural distortion and clustered microcal-cification. Complementary ultrasound examina-tion being ultimate in carcinoma detection inmammographic dense breasts should be consid-ered a standard complementary technique torule out any underlying pathology. Contrastenhanced MRI examination may be requestedfor selected cases. Furthermore, women at riskreceiving adjuvant therapy should be madeaware of the lower sensitivity of mammographyand should be offered alternative proceduresfor screening as recommended in previous stud-ies [21].

'Tumor factor' was the most common factorresponsible for missed breast carcinoma in ourstudy accounting for 44.1% of cases (67 pa-tients). According to Majid et al. [19], the cancersthat are most challenging to diagnose are thosewith subtle or indistinct features of malignancy.These features include areas of architecturaldistortion, small groups of amorphous or punc-tate microcalcification, focal asymmetrical den-sities, dilated ducts and well circumscribedmasses. Georgen et al. [22], found that missedcancers were statistically significant of lowerdensity and were seen on only one of two viewsmore often than were detected cancers. Weencountered 32 (21%) subtle carcinomas thatwere missed in mammography including: 4small (<3mm) infiltrating duct carcinomas and14 well defined carcinomas (5 infiltrating duct,4 mucinous, 3 intracystic and 2 intraductalpapillary carcinomas). Others included 10 car-cinomas presenting with subtle focal areas ofasymmetrical densities and 4 cases with areasof parenchymal distortion 2 of which turnedout to be infiltrating lobular carcinomas.

Asymmetric breast densities are frequentlyseen at mammography. According to Majid etal. [19], these findings in isolation have a lowpositive predictive value for malignancy; how-ever when they are associated with microcalci-fications or architectural distortion, the risk ofmalignancy is increased. They recommendedthat the workup of areas of asymmetric densityshould include clinical examination, additionalmammography views and ultrasound examina-

191

tion. In our study, we mainly depended onclinical assessment, ultrasound examination andpost processing manipulation (magnification,inversion and density control) when examina-tions are performed on digital machines. Mag-nification and spot compression views wereperformed for mammograms that were per-formed on conventional machines. FFDM issuperior to conventional mammography becauseof many reasons, FFDM is the system of advan-tages. It is superior in cancer detection. It leadsto reduction in examination time and retakes.It results in improved Image quality and lesiondetection, perfect analysis of microcalcification,X-ray dose reduction and improved workflowwith centricity solutions [19].

Any worrisome feature detected on ultra-sound was respected and further biopsy wasresorted to. MRI was also performed and wasdiagnostic in 5 cases. Spot compression andmagnification views were particularly usefulto evaluate the margins of masses, areas ofasymmetry and characteristics of microcalcifi-cation. Comparison of similar breast portionsand of previous mammograms is mandatory todetect any newly developed densities. The iden-tification of a focal density on one view shouldprompt a search in the same arc (measured fromthe nipple) on the other view as suggested in aprevious study [13]. A well circumscribed car-cinoma should always be considered in peri orpost menopausal women who present with acircumscribed solid mass, since fibroadenomasare not common at this age [19].

The observation of an obvious finding (be-nign or malignant) may cause the "happy eyesyndrome," misleading the radiologist into notlooking carefully for other lesions [13]. Twocarcinomas were missed due to adjacent strikingbenign lesions, one of which showed subtlemicrocalcific cluster seen on inversion imagesof digital mammography. Multicentric and mul-tifocal carcinomas were missed in 4.6% of cases(7 patients) and contralateral carcinomas in 3other patients (2% of cases). Multifocal diseaseis defined as multiple discrete, discontinuoustumor foci within 4cm in one breast while mul-ticentric disease was defined as two or moremalignant foci separated by 4 or more cm apartin one breast [20]. The detection of multicentricand multifocal carcinomas has a major impacton patient management. Multicentricity of foci

and multiplicity of histologic types may portenda worse prognosis when the clinical and patho-logic stage is otherwise specified [20]. Multi-centric carcinomas are also a contraindicationto breast conservative therapy [13]. Contralateralsynchronous breast cancer occurs in 0.19 to2.0% so careful attention must also be paid tothe opposite breast. To avoid this pitfall werecommend that when any striking feature isencountered on mammography, our searchshould not come to a stop, and we should lookcarefully in both breasts to exclude any othersynchronous lesions. Complementary ultrasoundexamination has proved itself in this study aswell as in a previous study [20] to be particularlyuseful in identifying additional foci of infiltrat-ing carcinoma.

Diffuse edema pattern found in 20 (13.2%)cases, although hindered the detection of under-lying breast masses and although not a specificsign of malignancy, yet it pointed to the presenceof an underlying pathological process. Furtherultrasound examination detected underlyinginfiltrating duct carcinoma in 6 cases. Ultra-sound together with mammography raised sus-picion of inflammatory carcinoma in 14 othercases. Mammograms of these cases showedsigns of isolated acute or subacute inflammatorychanges in the form of extensive edema, markedskin thickening with stromal coarsening asso-ciated with pathological lymph nodes. On ultra-sound examination, breasts showed increasedparenchymal echogenicity with diffusely edem-atous fat and intervening interstitial edema,overlying skin thickening with pathologicallyenlarged lymph nodes. The hallmark of thesecases is the detection of dermal lymphatic in-vasion on pathological revision of biopsy spec-imens. We reported a higher incidence of in-flammatory carcinomas in our study (9.2%),compared to a frequency ranging from 1 to 4%reported in a previous study [23]. We thus stressthe fact that any acute inflammatory process innon lactating women should be taken seriouslyand followed-up. Non resolution is an absoluteindication for immediate biopsy.

Patients with mammary Paget's disease (PD)present with a relatively long history of aneczematous skin lesion or persistent dermatitisin the nipple and adjacent areas. By thoroughhistologic examination, documented intra epi-dermal extension of malignant ductal epithelial



cells from underlying breast tissue into theepidermis was detected. Only about 50-70% ofpatients with biopsy-proven mammary PD showpositive findings on mammography includingsubareolar microcalcification which are helpfulin evaluating and locating clinically occult, nonpalpable underlying breast carcinoma, architec-tural distortion and thickening of the nipple andthe areola (reflecting edema) [24,25]. Our studyincluded 3 (2%) cases of Paget's disease of thenipple that were missed on mammography.These cases were suspected on US examination.MRI was performed for 2 cases showing mark-edly thickened enhancing areolar skin. Finaldiagnosis of these patients was mainly basedon revision of biopsy specimens.

High quality mammography images enhancea radiologist's ability to interpret mammogramsbecause they have greater sensitivity and spec-ificity. Images should be free from artifacts andshould be performed with adequate exposure,high contrast, high resolution, proper compres-sion and inclusion of the maximum amount ofbreast tissue [15,26]. 'Technical factors' wereresponsible for 29(19%) missed carcinomas,due to bad exposure factors in 5 (3.2%), mal-positioned breasts in 21 (13.8%) and bad pro-cessing quality in 3 (2%). Correct positioningof a woman for a mammogram is needed toensure a high quality image [4]. Technologiststoday must learn to break away from old habits.They need to apply new techniques with cre-ativity. They should understand that the goal ofimaging is to maximize the amount of breasttissue seen on the film [22]. In order to reachperfect performance, special care should begiven to technologists' training and they shouldwork under continuous supervision. Many prob-lems of exclusion can be eliminated if the breastis positioned in a careful and complete way.Traditionally this view calls for placement ofthe bucky at the neutral inframammary fold butthis fails to take advantage of the breast's naturalmobility and results in blind areas [26]. We thusstrongly recommend that in craniocaudal viewsthe breasts should be elevated as much as itsnatural mobility allows and pulled as much aspossible until the pectoralis muscle appears.The nipple should be profiled and the medialand lateral fibroglandular tissue should be in-cluded. The mediolateral oblique view is thesingle most useful mammography view as itincludes the breast from high up in the axilla

down to the inferior mammary fold. The gantryangle ranges from 30 to 60% following the ruleproposed by Heinlein [27], "the higher the per-son, the higher the angle and vice versa". Thepectoral muscle should be convex in shapereaching down to the level of the nipple. In astudy performed by Buist et al. [2], they foundthat what is sometimes interpreted as a mam-mography dense breast may also be a reflectionof poor mammography quality. Therefore tech-nologists should optimize image contrast toavoid over or under penetration. Careful atten-tion to daily processor quality is also necessary[13].

The accuracy of mammography interpreta-tion among individual radiologists varies widely[18,28]. One study showed a 40% disparityamong radiologist screening sensitivity and a45% range in the rates at which women withoutbreast cancer are recommended for biopsy. Asindicated by the receiver operating characteris-tics analyses, the ability of radiologists to detectcancer varies by as much as 11% [29]. The sourceof radiologist variability has not been completelyexplained. In our study, as well as in a previousstudy [30], the experience and training of radi-ologists were the main contributing factors.

Provider factors were responsible for 22(14.5%) of missed carcinomas in our study.These cases were considered when carcinomaswere detected on double reading by a moreexperienced senior radiologist. Double readingof mammograms has been proven to increasethe detection rate of carcinomas by up to 15%.Radiologists errors were mainly attributed totwo major factors, namely, perception and in-terpretation problems. In perception errors,accounting for 4.5% of cases of missed carci-nomas in our study (9 patients), the lesions wereincluded within the field of view and was evi-dent, yet the radiologists failed to interpret it.Subtle, small, non calcified and non distortingcarcinomas are responsible for these cases andthe same precautions described above to enhancetheir detection should be followed. Interpretationerrors by the radiologists occur when a lesionwith suspicious features is mis-interpreted as abenign or probably benign lesion (BI-RADS 2and 3). Perception and interpretation are causedby several factors including; deficient training,lack of experience, subtle features of malignan-cy, presence of an obvious finding, fatigue,

193

inattention, haste, poor view conditions anddistractions [15]. We strongly support the resultsof a previous study [18], that the radiologists'accuracy of mammography interpretation isstrongly related to the years of experience andtraining. In their study, radiologists' years ofexperience had the strongest association withperformance, such that radiologists with feweryears in practice had high sensitivity but lowerspecificity. We thus, in agreement, hypothesizethat training prior to starting and during practiceis a very important component of accuracy ofmammography interpretation. Direct feedbackof performance characteristics coupled withtraining may be more helpful than experiencewithout feedback. Open discussion of misjudgedmammograms may be useful [30]. We also rec-ommend that mammography interpretationshould be carried out under optimum conditions.Optimum viewing conditions include view box-es with adequate luminance, reduced extraneouslight and low ambient room light. Distractionsshould be minimized. Computer aided diagnosis(CAD System) could increase the sensitivity ofmammography interpretation [15].

Conclusion:

Why can a breast carcinoma be missed?

Four main factors are responsible for missinga carcinoma: (1) Patient factors (Inherentlydense breasts or acquired dense breasts). (2)Tumor factors (subtle carcinoma, masked car-cinoma, multifocal carcinoma and multicentriccarcinoma). (3) Technical factors (Bad exposurefactors, malpositioned breasts and bad process-ing quality). (4) Provider factors (bad perceptionand misinterpretation).

How to avoid missing a breast carcinoma?

Review clinical data and use US and otheradjunct techniques as MRI and biopsy to assessa palpable or mammography detected mass. Bestrict about positioning and technical factors.Try to optimize image quality. Be alert to subtlefeatures of breast cancers. Always consider thedefined carcinoma. Compare current imageswith multiple prior studies to look for subtleincreases in lesion size. Look for other lesionswhen one abnormality is seen. Judge a lesionby its most malignant features. Double Readingand use of CAD and finally FFDM (full fielddigital mammography).

REFERENCES1- Morris EA, Liberman L, Ballon DJ, Robson M,

Abramson AF, Heerdet A, et al. MRI of Occult BreastCarcinoma in a High Risk Population. AJR. 2003,181: 619-26.

2- Buist DS, Porter PL, Lehman C, Taplin SH, White E.Factors Contributing to Mammography Failure inWomen aged 40-49 Years. J Natl Cancer Inst. 2004,96 (19): 1432-40.

3- Luigi S, Tommasco L. Breast Cancer, Mammography,Emedicine. 2005.

4- Taplin SH, Rutter CM, Elmore JG, Seger D, WhiteD, Brenner J. Accuracy of Screening MammographyUsing Single Versus Independent Double Interpreta-tion. AJR Am J Roentegenol. 2000, 174: 1257-62.

5- Sivaramakrishna R. Imaging Techniques Alternativeto Mammography for Early Detection of Breast cancer.Technol Cancer Res Treat. 2005, 4 (1): 1-4.

6- Eberl MM, Fox CH, Edge SB, Carter CA, MahoneyMC. BI-RADS Classification for Management ofAbnormal Mammograms. J Am Board Fam Med.2006, 19: 161-4.

7- American College of Radiology. The ACR BreastImaging Reporting Data System (BI-RADS), Availablefrom: http://www.acr.org/departments/stand_accred/birads/contents.html. Accessed February. 2004, 27.

8- Kaul K. Early Detection of Breast Cancer: Is Mam-mography Enough. Breast Cancer. 2002, p: 49-54.

9- McCarthy EP, Burns RB, Freund KL, Ash AS, ShwartzM, Marwill SL, et al. Mammography use, BreastCancer Stage at Diagnosis and Survival Among OlderWomen. J Am Geriatr Soc. 2000, 48: 1226-33.

10- Overmoyer B. Breast cancer Screening. Med ClinNorth Am. 1999, 83: 1443-66.

11- Sirovich BE, Sox HC. Breast Cancer Screening. SurgClin north Am. 1999, 79: 961-90.

12- Baker LH. Breast Cancer Detection DemonstartionProject: Five year summary report. CA Cancer J Clin.1982, 32: 194-225.

13- Shaw E. Missed Breast Cancer; avoiding this pitfall.Appl Radiol. 2000, 29 (10): 15-30.

14- Bird RE, Wallace TW, Yankaskas BC. Analysis ofCancers Missed at Screening Mammography. Radiol-ogy. 1992, 184: 613-7.

15- Muttarak M, Pojchamarnwiputh S, Chaiwan B. BreastCarcinomas; why are they missed. Singapore Med J.2006, 47 (10): 851-7.

16- Fajardo LL, Hillman BJ, Frey C. Correlation BetweenBreast Parenchyma Patterns and Mammographer'sCertainty of Diagnosis. Invest Radiol. 1988, 23: 505-8.

17- Laya MB, Larson EB, Taplin SH, White E. Effect ofEstrogen Replacement Therapy on the Specificity andSensitivity of Screening Mammography. J Natl CancerInst. 1996, 88: 643-9.



18- Barlow WE, Chi C, Carney PA, Taplin S, D’Orsi C,Cutter G, et al. Accuracy of Screening MammographyInterpretation by Characteristics of Radiologists. JNatl Cancer Inst. 2004, 96 (24): 1840-50.

19- Majid AS, Shaw SP, Richard DD, Sharma NR, SalvadorX. Missed Breast Carcinoma: Pitfalls and Pearls.Radiographics. 2003, 23: 881-95.

20- Wendie AB, Patricia LG. Multicentric and MultifocalCancer.Whole Breast Ultrasound in PreoperativeEvaluation. Radiology. 2000, 214: 59-66.

21- Chiarelli AM, Kirsch VA, Klar NS, Shumak R, JongR, Fishell E, et al. Influence of Patterns of HormoneReplacement Therapy Use and Mammographic Den-sity on Breast Cancer Detection. Cancer Epidemiol.Biomarkers Prev. 2006, 15: 1856-62.

22- Goergen SK, Evans J, Colen GPB, MacMillan JH.Characteristics of Breast Carcinomas Missed byScreening Radiologists. Radiology. 1997, 204: 131-5.

23- Isil GB, Esin EU, Aysenar M. Inflammatory BreastCarcinoma: Mammographic, Ultrasonographic. Clin-

ical and Pathologic Findings in 142 cases. Radiology.2002, 223: 829-38.

24- Grace FK. Paget's Disease, Mammary, www.emedicine.com/derm/topic 305. htm. 2006.

25. Valdes EK, Feldman SM. Paget's disease of the breast.Breast J 2006, 12 (1): 83.

26- Chaloeykittin L, Muttarak M. Artifacts in Mammog-raphy; ways to identify and overcome them. SingaporeMed J. 2006, 47 (7): 634-40.

27- Heinlein RT. Breast Positioning-New thinking, newtechnique. www.agfa.com./mammographer_archive/mammographer. 2006.

28- Anderson ED, Muir BB, Walsh JS, Kirkpatrick AE.The Efficacy of Double Reading Mammograms inBreast Screening. Clin Radiol. 1994, 49: 248-51.

29- Beam CA, Conant EF, Sickles EA. Factors AffectingRadiologist Inconsistency in Screening Mammography. Acad Radiol. 2002, 9: 531-40.

30- Jackson VP, Hendrick RE, Feig SA, Kopans DB.Imaging of the radiographically dense breast. Radiol-ogy. 1993, 188: 297-300.

missed breast carcinoma; why and how to avoid?...carcinoma detection. it was diagnostic in 138...

Documents