Accuracy of subcategories A, B, C in BI-RADS 4 lesions by combined mammography and breast ultrasound findings

Journal of Medicine and Medical Sciences Vol. 2(3) pp. 728-733 March 2011 Available online@ http://www.interesjournals.org/JMMS Copyright ©2011 Intern...
Author: Ambrose King
5 downloads 0 Views 832KB Size
Journal of Medicine and Medical Sciences Vol. 2(3) pp. 728-733 March 2011 Available online@ http://www.interesjournals.org/JMMS Copyright ©2011 International Research Journals

Full Length Research Paper

Accuracy of subcategories A, B, C in BI-RADS 4 lesions by combined mammography and breast ultrasound findings Wanaporn Burivong1 and Ornsiri Amornvithayacharn2 1

Radiology instructor, Department of Radiology, Faculty of Medicine, Srinakharinwirot University, Ongkarak, Nakhonnayok, Thailand, Email [email protected] 2 Radiology instructor, Department of Radiology, Faculty of Medicine, Srinakharinwirot University, Ongkarak, Nakhonnayok, Thailand, Email: [email protected] Accepted 20 March, 2011

Subcategorizing BI-RADS category 4 lesions are useful to inform the level of concern for carcinoma. To determine the accuracy of BI-RADS 4 subcategories, mammographic and ultrasound images of 143 patients were independently and blindly reviewed by two radiologists and later compared with pathologic diagnosis. Sixty-eight of 143 (47%) were classified as subcategory 4A, 54 (37%) were 4B and 21 (14%) were 4C. The positive predictive value after consensus evaluation in subcategory 4A, 4B and 4C are 4.4%, 43.3% and 52.1%, respectively. Interobserver agreement for mammographic and ultrasound descriptions varies from perfect to fair agreement. We conclude that the malignancy rate in subcategories 4B and 4C was significantly higher than in subcategory 4A. Interobserver agreement varies from slightly lower to comparable with previous studies. The accuracy of mammographic and breast ultrasound interpretation can be improved by consensus evaluation in uncertain cases to reduce underestimated malignancy. Keywords: BI-RADS 4, subcategory, mammography, breast ultrasound INTRODUCTION Breast cancer is the most frequent cancer of women in industrialized and developed countries. Incidence rates are lower in Africa and Asia (Parkin et al., 2005). In Thailand, breast cancer is the second most common cancer in women. The incidence rate has been increasing in the past decade. The estimated incidence rate is now 20.5 per 100,000 women (Chaiwerawatana, 2007). Early stages of breast cancers are generally small and usually have no symptoms. Mammography has improved the detection of these lesions. However, due to significant higher breast density in Asian population, sonography following mammogram is valuable (Carmen et al., 2007). Breast sonography is also be used to evaluate masses detected by mammogram or palpable masses from clinical examination (Kim et al., 2008).

The Breast Imaging Reporting and Data System (BIRADS) developed by the American College of Radiology (ACR) is use to standardized mammographic reporting and interpretations. Details of diagnostic mammography and ultrasound findings are included in the same report with final assessment category ranging from 1 to 6 according to the degree of lesion suspicion (Orel, 1999; Balleyguier, 2007). In 2003, the fourth edition of BI-RADS includes section of ultrasound breast and magnetic resonance imaging as well as sub-categorizing category 4 lesion into three subdivisions (4A, 4B and 4C). The BIRADS category 4 is used for stratification wide range of lesions that require interventional procedures. Subcategorizing category 4 lesions are stratified to category 4A (needing intervention but low suspicion for

Wanaporn and Ornsiri 729

malignancy), category 4B (lesions with intermediate suspicion for malignancy) and category 4C (moderate concern but not classic for malignancy). Clinicians are better informed and help to understand the level of concern for carcinoma. This subdivision also encourages pathologists to aware of benign subcategory 4C lesions (American College of Radiology, 2003). The accuracy of mammographic and ultrasound interpretations depends on various factors such as characteristic of population, patients age, technical aspects, radiologist experience and previous knowledge of BI-RADS guideline (Liberman, 1998; Hong, 2005). The purpose of this study is to evaluate the reliability and variability of subcategories A, B, and C in BI-RADS category 4 lesions by determining the positive predictive value in each subcategory and evaluating interobserver agreement of mammography and breast ultrasound findings. MATERIAL AND METHOD The authors retrospectively reviewed 143 patients whose previous mammographic report were classified as BIRADS category 4 lesions and had percutaneous or surgical biopsy between August 1, 2003 and December 31, 2009 at HRH Princess Maha Chakri Sirindhorn Medical Center. Mammographic and ultrasound images were independently and blindly reviewed by two boardcertified diagnostic radiologists. Description for mammographic and ultrasound findings using the terminology bases on fourth edition of BI-RADS lexicon was provide for each observer to select the most appropriate description. Final subcategories of BI-RADS category 4 lesions were assessed based on the observer’s previous knowledge of BI-RADS guidelines and individual experience. If the final subcategory assessment is disagreeing, re-evaluation and discussion to re-subcategorize was performed. Pathologic diagnosis was determined by image-guided core needle biopsy, excisional biopsy and fine needle aspiration. Once surgical excision was performed after percutaneous core needle biopsy or fine needle aspiration, the pathologic diagnosis was based on the final surgical pathologic analysis. The positive predictive value for subcategories 4A, 4B and 4C was determined by term of positive biopsy rate (PBR) from each observer and after consensus evaluation. Interobserver agreement for the description of mammographic and ultrasound lesions as well as final subcategory assessment were calculated by means of the Cohen’s kappa (k) test.

RESULTS From August 2003 to December 2009, 143 mammographic studies with additional ultrasound were categorized in BI-RADS 4 category and underwent intervention procedure. The mean age of patients was 48.6 years, with a range of 20 to 89 years. Of these patients, 60.8% had palpable breast mass, 15.3% of the lesions were asymptomatic and diagnosed from screening, 13.9% were diagnosed during interval mammographic follow up, 8.3% of patients presented with other breast symptoms such as mastalgia and nipple discharge and 1.4% of lesions were found during surveillance of women on hormone replacement therapy. Sixty-eight of 143 patients (47.5%) were classified as subcategory 4A, 54 (37.7%) were 4B and 21 (14.6%) were 4C. There were 39 malignant lesions in the present study. The final pathology of lesions in BI-RADS 4 subcategories were as follows: 52.1% (12/23) of subcategory 4C lesions and 43.4% (23/53) of subcategory 4B lesions were malignant (Figure 1 and figure 2). Premalignant lesions (atypical ductal hyperplasia and atypical lobular hyperplasia) was found in 8.7% of subcategory 4C lesions, 5.6% of subcategory 4B lesions and 4.4% of subcategory 4A lesions. Fortyfive percent (31/68) and 20.5% (14/68) of subcategory 4A lesions were fibrocystic disease and fibroadenomas, respectively (Figure 3). Only 5.8% (4/68) of subcategory 4A lesions were malignant (Figure 4). The positive predictive values are summarized in table 1. Interobserver agreement for BI-RADS category assessesment is moderate (K-value 0.49). Interobserver agreement for mammographic descriptions is summarized in table 2. The evaluation of breast density was fair. Substantial agreement was found for the presence of mass. Agreement was moderate for evaluation of mass shape and margin. Slight agreement was obtained for evaluation of mass density. About calcification assessment, substantial agreement was found for the presence of calcification. Agreement was nearly perfect when describing type of calcification. The distribution of calcification shows moderation agreement. Associated findings from mammogram also show moderate agreement. Interobserver agreement for ultrasound descriptions is summarized in table 3. Substantial agreement was achieved for presence of mass and orientation. Moderate agreement was obtained for shape of mass and presence of calcification. The margin of mass, lesion boundary, echogenicity and posterior acoustic shadow show fair agreement.

730 J. Med. Med. Sci.

Figure 1. BI-RADS category 4B lesion. A 54-year-old female presented with right breast mass. (A) Mammogram of right breast shows a lobulated shape with indistinct border lesion at lower inner quadrant. (B) Ultrasonography shows a hypoechoic lesion with partially indistinct border. The final pathology is mucinous carcinoma

DISCUSSION The present study shows a wide range of pathologic findings from BI-RADS category 4 lesions. They vary from benign lesions such as fibroadenomas, fibrocystic disease, and premalignant lesions to malignant cancer. The reliability of BI-RADS category 4 interpretations was assessed by determining the accuracy and positive predictive value (PPV). In the present study, the PPVs for BI-RADS subcategory 4A, 4B, and 4C from observer A are closed the previous studies from Wiratkapun et al (2010), Eradat and Basset (2007), and Lazarus et al (2006) which are 6% in subcategory 4A, 40.4% in

Figure 2. BI-RADS category 4C lesion. A 71-year-old female presented with left breast mass. (A) Mammogram of left breast shows a spiculated mass at lower inner quadrant. (B) Ultrasonography reveals a hypoechoic lesion with irregular margin. Final pathology is invasive ductal carcinoma.

subcategory 4B and 57% in subcategory 4C, respectively. The PPVs according to subcategories 4A and 4B from observer B are closed to the PPV reported by Eradat and Basset (2007). which are 10.5% and 40.4%, respectively. The PPV for subcategory 4C from observer B in this study is slightly lower as were in prior studies, suggesting underestimation of malignancy in this subcategory (Eradat and Bassett, 2007; Lazarus, 2006; Wiratkapan, 2010). However, after consensus evaluation and re-sub categorization, the final PPV assessment for BI-RADS subcategory 4A is lower but still comparable to the study by Wiratkapun et al (2010). Subcategory 4B shows slightly higher PPV which is more closed to the study reported by Eradat and Bassett (2007). The PPV for subcategory 4C is also more comparable to studies from Wiratkapun et al (2010) and Lazarus et al (2006)

Wanaporn and Ornsiri 731

Figure 3. BI-RADS category 4A lesion. A 38-year-old female presented with left breast mass. (A) Mammogram of left breast shows an oval-shape with circumscribed border lesion at upper outer quadrant. (B) Ultrasonography reveals a hypoechoic lesion with well-circumsribed margin and posterior acoustic enhancement. Final pathology is fibroadenoma

which range from 53-57%. This suggests that consensus evaluation helps improve accuracy from underestimation of BI-RADS subcategories. The final PPVs according to subcategories 4A, 4B and 4C were 4.4%, 43% and 52%, respectively. The malignancy rate in subcategories 4B and 4C was ten and thirteen times higher than in subcategory 4A. In subcategory 4A lesions, the major pathologic finding is

Figure 4. BI-RADS category 4A lesion. A 55-year-old female presented with left breast mass. (A) Mammogram of left breast shows a circular-shape with circumscribed border lesion at lower inner quadrant. (B) Ultrasonography reveals a mixed hypohyperechoic lesion with well-circumscribed margin. Final pathology is invasive ductal carcinoma.

benign. Their imaging findings show circular or oval shape, well circumscribed or lobulated margin and parallel orientation from sonography which is considered benign characteristics. Harvey et al. reveals that in palpable breast lesions with benign imaging findings, short-term follow-up is also a reasonable alternative management compared to immediate biopsy (Harvey et al., 2009).

732 J. Med. Med. Sci.

Table 1. Positive predictive value (positive biopsy rate)

BI-RADS subcategory 4A 4B 4C

Observer A 6.6 38.6 60.8

Positive predictive value (%) Observer B 14.0 36.5 42.8

Final assessment 4.4 43.4 50.0

Table 2. Interobserver variability in the description of mammographic lesions

Description of mammographic lesions Breast density Presence of mass Shape of mass Margin of mass Density of mass Presence of calcification Type of calcification Distribution of calcification Associated finding

K-value 0.34 0.68 0.42 0.54 0.20 0.72 0.82 0.49 0.50

Table 3. Interobserver variability in the description of ultrasound lesions

Description of ultrasound lesions Presence of mass Shape of mass Margin of mass Orientation Lesion boundary Echogenicity Posterior acoustic shadow Surrounding tissue change Calcification

The variability of BI-RADS category 4 was assessed by analysis of interobserver agreement. The agreement for BI-RADS category assessment in this study (K=0.49) was slightly higher as compared to the prior study (K=0.27) (Nascimento et al. 2010). About agreement for mass description on mammograms, the presence of mass was better than agreement for shape and margin of mass. Breast density and density of mass shows lower rate of agreement. The

K-value 0.65 0.45 0.27 0.57 0.25 0.37 0.23 0.66 0.58

agreement in shape, margin and density of mass are comparable to previous studies by Nascimento et al (2010) and Lazarus et al (2006), while agreement in presence of mass and breast density are slightly lower. The agreement for type of calcification on mammograms is higher than the agreement rate for presence of calcification and distribution of calcification. As compared to prior studies, distribution of calcification shows close agreement rate. Agreement for type of

Wanaporn and Ornsiri 733

calcification in this study is slightly higher while the agreement for presence of calcification is slightly lower (Lazarus, 2006; Nascimento, 2010). In ultrasound descriptions, the agreement for presence of mass was better than agreement for shape and orientation of mass. The lower rate of agreement for margin of mass, lesion boundary, mass echogenicity and posterior acoustic shadow suggest that observers had different interpretations in these findings. However, the agreement in margin of mass, mass echogenicity and posterior acoustic shadow were not different from the previous study by Lazarus et al. Agreement for presence of mass, shape, orientation and echogenicity of mass was lower in the present study (Lazarus , 2006; Nascimento, 2010). The limitation of this study is that all lesions were referred for biopsy which Antonio et al. described that studies without negative cases had a trend of lower kappa values in bivariate analysis. This limitation also excludes the possibility of a missed diagnosis case (Antonio and Crespi, 2010). Second, the sample is small which may cause decreased rate of interobserver agreement. Third, the observers did not have exchange of correspondence in BI-RADS assessment. The interpretation was based on observer’s previous knowledge of BI-RADS guideline. Even they may have observed the same thing; they can interpret it differently (Elmore et al., 1994). These limitations lead to lower agreement in this study as compared to former studies. CONCLUSION The malignancy rate in subcategories 4B and 4C was significantly higher than in subcategory 4A. PPV was comparable to previous studies. Interobserver agreement varies from slightly lower to comparable with prior studies. Interpreting breast imaging has a wide variability, depending on radiologist experience and previous knowledge of BI-RADS guideline. The accuracy of mammographic and breast ultrasound interpretation can be improved by consensus evaluation in uncertain cases to reduce underestimated malignancy. ACKNOWLEDGMENT The authors gratefully thank Assistant Professor Vithya Varavithya for his suggestions and editorial assistant. This work was supported by a grant from the Srinakharinwirot University.

REFERENCE American College of Radiology (2003). Breast imaging reporting and data system: breast imaging atlas. 4th ed. American College of Radiology. Antonio AL, Crespi CM (2010). Predictor of interobserver agreement in breast imaging using the breast imaging report and data system. Breast Cancer Res. Treat. 120: 539–546. Balleyguier C, Ayadi S, Nguyen KV, Vanel D, Dromain C, Sigal R (2007). BIRADS classification in mammography. Eur. J. Radiol. 61: 192-194 Carmen MG, Halpern EF, Kopans DB, Moy B, Moore RH, Gross PE, Hughes KS (2007). Mammographic breast density and race. AJR Am. J. Roentgenol. 188: 1147-1150. Chaiwerawatana A (2007). Breast: cancer in Thailand: 4th ed. Bangkok (Thailand): Bangkok Medical Publisher. pp. 48-50. Elmore JG, Wells CK, Lee CH, Howard DH, Feinstein AR (1994). Variability in radiologists’ interpretation of mammograms. N. Engl. J. Med. 331: 1493-1499 Eradat J, Bassett L (2007). BI-RADS category 4: option for subcategories 4A, 4B, and 4C. Womensimagingonline [Online]. Aug 1 [cited 2009 Dec 14]. Available from: URL: http://womensimagingonline.arrs.org Harvey JA, Nicholson BT, LoRusso AP, Cohen MA, Bovbjerg VE (2009). Short term follow up of palpable breast lesions with benign imaging features: evaluation of 375 lesions in 320 women. AJR Am J Roentgenol. 193: 1723-1730. Hong AS, Rosen EL, Soo MS, Baker JA (2005). BI-RADS for sonography positive and negative predictive values of sonographic features. AJR Am. J. Roentgenol. 184: 1260-1265. Kim EY, Ko KH, Oh KK, Kwak JY, You JK, Kim MJ, Park BW (2008). Clinical application of the BI-RADS final assessment to breast sonography in conjunction with mammography. AJR Am. J. Roentgenol. 190: 1209-1215. Lazarus E, Maineiro MB, Schepps B, Koelliker SL, Livingston LS (2006). BI-RADS lexicon for US and mammography: interobserver variability and positive predictive value. Radiol. 239: 385-391. Liberman L, Abramson AF, Squires FB, Glassman JR, Morris EA, Dershaw D (1998). The breast imaging reporting and data system: positive predictive value of mammographic features and final assessment categories. AJR Am. J Roentgenol. 171:35-40. Nascimento JH, Silva VD, Maciel AC (2010). Accuracy of mammographic findings in breast cancer: correlation between BIRADS classification and histological findings. Radiol. Bras. 43(2): 91–96. Orel SG, Kay N, Reynold C, Sullivan DC (1999). BI-RADS categorization as a predictor of malignancy. Radiol. 211(3): 845-850. Parkin DM, Bray F, Ferlay J, Pisani P (2005). Global cancer statistic, 2002. CA Cancer J. Clin. 55: 74-108. Wiratkapan C, Boonyapaiboonsri W, Wibulpolprasert B, Lertsithichai P (2010). Biopsy rate and positive predictive value for breast cancer in BI-RADS category 4 breast lesions. J. Med. Assoc. Thai. 93(7): 830837.

Suggest Documents