- Email: [email protected]
Correlation Between Ultrasound Characteristics, Mammographic Findings and Histological Grade in Patients with Invasive Ductal Carcinoma of the Breast
Clinical Radiology (2000) 55, 40–44
Correlation Between Ultrasound Characteristics, Mammographic Findings and Histological Grade in Patients with Invasive Ductal Carcinoma of the Breast PHIL IPPA M. L AMB, NI CH OL AS M . PE RRY, SARAH J. VIN NICOMBE , C L IVE A. W E L L S Department of Radiology, The Royal Hospitals NHS Trust, St. Bartholomew’s Hospital, West Smithfield, London EC1A 7BE, U.K. Received: 12 February 1999 Revised: 16 June 1999 Accepted: 7 July 1999 AIM: To investigate the relationship between ultrasound characteristics, mammographic findings and histological grade in cases of invasive ductal carcinoma which produce a mass on ultrasound. MATERIAL AND METHODS: A retrospective review was performed of the imaging findings in 120 patients diagnosed with invasive ductal carcinoma of the breast between January 1996 and December 1997. Imaging findings were correlated with the histological grade of tumour. RESULTS: High-grade tumours were significantly larger both on ultrasound and mammography (P < 0.016). A spiculated margin on mammogram was documented in 72% of low-grade tumours compared with only 24% of high-grade tumours (P ¼ 0.001). Twenty-two per cent of low-grade tumours had a poorly defined margin on mammography compared with 66% of high-grade tumours (P ¼ 0.001). At ultrasound, 16% of high-grade tumours (95% confidence limits 7–29%) had a welldefined margin. Acoustic enhancement was seen in 36% of high-grade tumours compared with only 9% of low and intermediate-grade tumours (P ¼ 0.003): 22% of all tumours showed acoustic enhancement. Acoustic shadowing was seen in 71% of low-grade tumours compared with only 28% of high-grade tumours (P ¼ 0.003). Malignant-type microcalcification was seen on mammogram in 6% of low-grade tumours compared with 31% of high-grade tumours (P ¼ 0.045). CONCLUSION: The classical appearance of a malignant breast mass as a spiculated mass on mammogram associated with acoustic shadowing on ultrasound is more typical of a low-grade tumour. In comparison, high-grade tumours are more likely to demonstrate posterior acoustic enhancement, and a proportion has a well-defined margin on ultrasound. Therefore, high-grade invasive ductal carcinoma may paradoxically display similar imaging features to a benign breast mass. Lamb, P. M. et al. (2000). Clinical Radiology 55, 40–44. q 2000 The Royal College of Radiologists Key words: breast carcinoma, invasive ductal carcinoma, ultrasound, attenuation characteristics, histological grade.
INTRODUCTION
The use of ultrasound as an adjunct to mammography in the diagnosis of breast cancer is well established [1–4]. An abnormal mass on mammography is reported in 50% of carcinomas of less than 10 mm diameter and 88% of those greater than 10 mm diameter [5]. A spiculated mass on mammography has been reported in up to 84% of breast malignancies [6] (Fig. 1). Historically, at ultrasound examination posterior acoustic shadowing was said to be associated with the majority of malignant breast masses: 70–80% overall and almost 100% of scirrhous (spiculated) lesions [3] (Fig. 2). However, it is now widely recognized that not all invasive Author for correspondence and guarantor of study: P. M. Lamb, 21 Old Sun Wharf, 40 Narrow Street, London E14 8DG, U.K. 0009-9260/00/010040+05 $35.00/0
breast masses demonstrate posterior shadowing: they may have no distal effect or even be associated with acoustic enhancement. In our clinical practice we have observed that tumours associated with distal acoustic enhancement tend to be those which are designated high-grade at subsequent histological examination (Fig. 3). The purpose of this study was to evaluate further the relationships between attenuation characteristics, mammographic findings and histological grade of breast cancer. The aims of the study were as follows. (i) To categorize the sonographic characteristics and mammographic features of invasive ductal carcinoma. (ii) To correlate the imaging findings with the histological grade of invasive ductal carcinoma. (iii) To ascertain whether the histological grade of invasive ductal carcinoma can be predicted from the imaging appearances. q 2000 The Royal College of Radiologists
IMAGING CHARACTERISTICS AND HISTOLOGICAL GRADE IN CARCINOMA OF THE BREAST
41
Fig. 2 – Acoustic shadowing associated with a low-grade (grade 1) invasive ductal carcinoma of the breast.
Fig. 1 – A low-grade (grade 1) invasive ductal carcinoma producing a spiculated mass on mediolateral oblique mammogram.
MATERIALS AND METHODS
A retrospective review was performed of the ultrasound and mammographic findings of 120 consecutive patients diagnosed with invasive ductal carcinoma at our institution between January 1996 and December 1997. The sample was obtained from a histopathology database: invasive ductal carcinoma accounts for 80% of all cases of invasive breast cancer and by excluding other histological subtypes (e.g. lobular and mucinous carcinomas) from our study we aimed to minimize confounding variables. Symptomatic and screen-detected cases were included and the women were aged 24 to 97 years (mean 58 years). Ultrasound scans were performed using an Acuson Sequoia 5L8 linear probe for the majority of cases, although some of the earlier studies were performed using an Aloka 121 SSD (KeyMed) dedicated 7.5 MHz sector probe. The images were analysed by two experienced breast radiologists. Features assessed on ultrasound were: (i) size of mass (mm); (ii) depth/width ratio; (iii) margin of mass (well or poorly defined); (iv) attenuation characteristics (shadowing, enhancement, mixed or no distal effect); (v) echotexture (homogeneous or heterogeneous); (vi) halo (present, absent). The term ‘halo’ refers to an irregular zone of increased echogenicity that may surround a tumour mass [7]. Similarly, corresponding mammograms were assessed for:
Fig. 3 – Acoustic enhancement associated with a high-grade (grade 3) invasive ductal carcinoma of the breast.
42
CLINICAL RADIOLOGY
120 patients diagnosed with invasive ductal carcinoma for the period January 1996–December 1997 was as follows: 19 grade 1 (low-grade) tumours; 37 grade 2 (intermediate-grade) tumours; 64 grade 3 (high-grade) tumours.
(i) size of mass (mm); (ii) margin (well-defined, poorly defined, partially defined or spiculated) [8]; (iii) microcalcification: the presence of malignant-type microcalcification was documented in case the calcification was altering the attenuation characteristics of the mass itself. The observers were not restricted to specific mammographic projections and their assessment related to those images that in their opinion provided the optimum image quality for the features being analysed. The observers were blinded to the histological grade assigned to each case and an attempt was made by them to predict the grade based on the combined mammographic and sonographic appearances. Histological grading was performed according to the Elston and Ellis protocol [9]. This involves the assessment of three components of tumour morphology: tubule formation, nuclear pleomorphism and mitotic frequency. Each is scored from 1 to 3, and adding the scores gives the overall grade. Score 3–5 6–7 8–9
Grade 1 2 3
Ultrasound Findings Ultrasound images were available for 104 cases: of these 17 were grade 1 tumours, 36 were grade 2 tumours and 51 were grade 3 tumours. The following features were found to be statistically significant when correlated with the histological grade (Table 1: Kruskal–Wallis ANOVA test by rank, chisquared test for trend and the Holm–Bonferroni correction): (i) an increase in size of tumour mass with increasing histological grade; (ii) 71% of grade 1 tumours demonstrated acoustic shadowing compared with only 28% of grade 3 tumours; (iii) 36% of grade 3 tumours demonstrated acoustic enhancement compared with only 9% of grade 1 and grade 2 tumours. Overall (for all histological grades), 41% of tumours showed acoustic shadowing while 22% showed acoustic enhancement. The size of tumour was compared between grades and the presence of acoustic shadowing or enhancement (Table 2). For grade 1 tumours, those that showed acoustic enhancement tended to be larger, but for grade 3 tumours this was not the case. There was no significant relationship between the size of the tumour and the nature of the distal effect. With regard to the depth/width ratio, 24% of all tumour masses had an increased ratio, but this was not grade specific (Table 1). Of those tumours which had a well-defined margin and showed acoustic enhancement, only one showed an increased depth/width ratio. The majority of masses had a heterogeneous appearance and the echo-texture was not grade specific. A halo was noted in 39% of all cases, but its presence was not significantly related to the histological grade.
Low Intermediate High
Statistical Methods Statistical analysis was performed using methods appropriate for non-parametric data: a Kruskal–Wallis analysis of variance (ANOVA) by rank for continuous data, chi-squared test for trend for categorical data and the Holm–Bonferroni correction [10,11]. The significance level was 5%. Regarding the radiologists’ attempt to predict histological grade, the positive predictive value was calculated for both high and low-grade tumours [12].
RESULTS
The distribution of histological grades within the data set of Table 1 – Ultrasound features Size (mm)
Grade 1 Grade 2 Grade 3 P-value Holm corrected
Margin
Echotexture
Median, inter-quarterile range
† Depth/width
Poorly defined
Heterogeneous
12, 8–14 14, 9–22 20, 11–27
6 (35) 12 (33) 7 (14)
17 (100) 34 (94) 43 (84)
13 (76) 28 (78) 38 (76)
0.006* 0.030*
0.203 0.609
0.098 0.392
0.981 0.981
Distal effect
Grade 1 Grade 2 Grade 3 P-value Holm corrected
Halo
Enhancement
Shadowing
None
Mixed
5 (29) 14 (39) 21 (43)
3 (18) 2 (6) 18 (36)
12 (71) 16 (44) 14 (28)
2 (12) 9 (25) 9 (18)
0 9 (25) 9 (18)
0.619 1.000
0.003 0.018
* Kruskal–Wallis ANOVA test by rank; chi-squared test for trend for all other parameters; significance level ¼ 0.05; (%) in parentheses.
43
IMAGING CHARACTERISTICS AND HISTOLOGICAL GRADE IN CARCINOMA OF THE BREAST
Table 2 – Comparison of size of tumours showing acoustic enhancement or shadowing on ultrasound Size (median, inter-quartile range, mm)
All Grade 1 Grade 2 Grade 3
Acoustic enhancement
Acoustic shadowing
16, 15, 22, 18,
12, 9–20 9, 8–12 12, 9–19 16, 10–24
13–26 6–17 14–30 13–26
P-value*
0.635 0.151 0.219
* Kruskal–Wallis ANOVA test by rank.
histological grade by choosing either a high-grade or low-grade category (Table 5). The results indicate that for high-grade tumours there was a positive predictive value of 84%, while for low-grade tumours this was 73%. The specificity of overall radiological prediction of histological grade was 89%, while the sensitivity was 65%. In summary, the significant results in this study are: (i) highgrade (grade 3) tumours tend to be larger both on ultrasound and mammography; (ii) low-grade (grade 1) tumours tend to produce a spiculated mass on mammogram and to demonstrate acoustic shadowing on ultrasound; (iii) 36% of high-grade tumours demonstrate acoustic enhancement at ultrasound examination.
Mammographic Findings
DISCUSSION
Mammograms were available for 117 patients: of those in whom a mass was present on the mammogram (112 cases), 18 were grade 1 tumours, 36 were grade 2 tumours and 58 were grade 3 tumours. The following were found to be statistically significant when correlated with the histological grade (Table 3: Kruskal–Wallis ANOVA test by rank, chi-squared test for trend and Holm–Bonferroni correction): (i) an increase in size of tumour mass with increasing histological grade; (ii) 72% of grade 1 tumours demonstrated a spiculated margin compared with only 24% of grade 3 tumours; (iii) 22% of grade 1 tumours had a poorly defined margin compared with 66% of grade 3 tumours; (iv) malignant-type microcalcification was seen in 6% of grade 1 tumours compared with 31% of grade 3 tumours. The size of tumour was compared between grades and the presence of a poorly defined or spiculated margin (Table 4). There was no significant relationship between the size of the tumour and the assessment of the margin of the tumour mass on mammogram. Analysis of mammographic and sonographic features for all histological grades revealed that 28/100 (28%) of all masses had both a spiculated margin on mammography and demonstrated acoustic shadowing on ultrasound: of these, 10/28 (36%) were grade 1 tumours, 11/28 (39%) were grade 2 tumours and 7/28 (25%) were grade 3 tumours.
Radiological Prediction of Histological Grade Based on the combined appearances on mammography and ultrasound an attempt was made by the radiologist to predict the
The increase in size of tumour mass with increasing histological grade is to be expected, due to the faster mitotic rate in high-grade tumours. However, the differences in attenuation characteristics between the histological grades of invasive ductal carcinoma have not previously been reported and need to be explained. Previous studies which have examined the relationship between ultrasound characteristics and pathological features of breast cancer have attributed particular importance to the presence of fibrous connective tissue in those cases that demonstrate acoustic shadowing, the so-called desmoplastic reaction. Conversely it has been suggested that those tumours which demonstrate acoustic enhancement are more cellular [7,13]. However, a study by Gozzi et al. [14] reported that it is not the absolute amount of fibrous tissue within a tumour, but the organization of the tissue that determines the attenuation characteristics. In fact, in their study the tumours which demonstrated acoustic enhancement were less cellular (and contained more fibrous tissue). Some studies have reported acoustic enhancement in mucincontaining tumours (but these studies have involved only relatively small numbers of patients – eight in each) [15,16]. We intend to carry out further histological assessment with regard to the amount of fibrous tissue, cellularity and organization of tissues within malignant breast masses. The findings will be correlated with the attenuation characteristics that we have reported above. Regarding the radiologist’s attempt to predict histological
Table 3 – Mammographic features Size (mm)
Margin
Median, inter-quartile range
Poorly defined
Partially defined
Spiculated
Well-defined
Microcalcification
Grade 1 Grade 2 Grade 3
12, 10–15 15, 10–20 17, 15–30
4 (22) 14 (39) 38 (66)
1 (6) 0 3 (5)
13 (72) 22 (61) 14 (24)
0 0 3 (5)
1 (6) 6 (16) 19 (31)
P-value Holm corrected
0.016* 0.032*
0.001 0.003
0.045 0.045
* Kruskal–Wallis ANOVA test by rank; chi-squared test for trend for all other parameters; significance level ¼ 0.05; (%) in parentheses.
44
CLINICAL RADIOLOGY
Table 4 – Comparison of size of tumours showing poorly defined or spiculated margin Size (median, inter-quartile range, mm) Poorly defined margin
Spiculated margin
15, 11, 13, 17,
15, 12, 18, 15,
However, in our study these were not reliable indicators and the absence of these signs did not preclude malignancy. All solid lesions at ultrasound should therefore be subject to cytological or histological sampling and correlation made with the imaging appearances.
P-value*
REFERENCES All Grade 1 Grade 2 Grade 3
12–25 9–13 11–18 12–28
11–20 10–15 10–20 12–30
0.459 0.304 0.535
* Kruskal–Wallis ANOVA test by rank.
Table 5 – Radiologists’ prediction of histological grade Actual grade Predicted grade
High grade
Low grade
High grade Low grade
32 17
6 47
Positive predictive value for high-grade tumours ¼ 84%; positive predictive value for low-grade tumours ¼ 73%; sensitivity ¼ 65%; specificity ¼ 89%.
grade of breast cancer, the imaging characteristics appear to be a reliable predictor, particularly for high-grade tumours (positive predictive value ¼ 84%). Theoretically this may have implications for patient management in the future. However, at present core biopsy remains the most reliable method of ascertaining histological grade pre-operatively. In conclusion, the classical appearances of a spiculated mass on mammography associated with acoustic shadowing on ultrasound are more likely to represent a low or intermediategrade tumour (75% of cases in this study). In this study, 22% of all cases of invasive ductal carcinoma demonstrated acoustic enhancement and radiologists performing breast imaging should be aware that many of these could be misinterpreted as benign. Such tumours are more likely to be those which are found to be high-grade at subsequent histological examination. Radiologists should make a careful search for other features which may indicate malignancy, such as an increase in the depth/width ratio or disruption of adjacent normal tissues.
1 Jacob D, Brombart JC, Muller C, Lefebvre C, Massa F, Depoerck A. Analysis of the results of 137 subclinical breast lesions excisions. Value of ultrasonography in the early diagnosis of breast cancer. J Gynecol Obstet Biol Reprod (Paris) 1997;26:27–31. 2 Jackson VP, Reynolds HE, Hawes DR. Sonography of the breast. Semin Ultrasound CT MR 1996;17:460–475. 3 Tohno E, Cosgrove DO, Sloane JP. Ultrasound Diagnosis of Breast Diseases. Edinburgh: Churchill Livingstone; 1994, chapter 9, pp. 158– 162. 4 Tucker AK. Textbook of Mammography, 1st ed. Churchill Livingstone, 1993, chapter 9, pp. 219–226. 5 Andersson I. Radiographic screening for breast carcinoma III: appearance of carcinoma and number of projections to be used at screening. Acta Radiol Diagn 1981;22:407–420. 6 Barth V. Atlas of Diseases of the Breast. Stuttgart: Georg Thieme, 1979. 7 Kobayashi T. Diagnostic ultrasound in breast cancer: analysis of retrotumerous echo patterns correlated with sonic attenuation by cancerous connective tissue. JCU J Clin Ultrasound 1979;7:471–479. 8 Burnett SJD, Ng YY, Perry NM et al. Benign biopsies in the prevalent round of breast screening: a review of 137 cases. Clin Radiol 1995; 50:254–258. 9 Elston CW, Ellis IO. Pathological prognostic factors in breast cancer No. 1. The value of histological grade in breast cancer: experience from a large study with long-term follow-up. Histopathology 1991;19:403– 410. 10 Aicken M, Gensler H. Adjusting for multiple testing when reporting research results: The Bonferroni vs Holm methods. Am J Public Health 1996;86:726–728. 11 Altman DG. Practical Statistics for Medical Research. London: Chapman & Hall, 1991;211–213. 12 Goldin J, Sayre JW. A guide to clinical epidemiology for radiologists: part I. Study design and research methods. Clin Radiol 1996;51:313– 316. 13 Kossoff G. Causes of shadowing in breast sonography. Ultrasound Med Biol 1988;14:211–215. 14 Gozzi G, Cressa C, Bazzocchi M, Stanta G, Vidali C. Causes of attenuation of the sound waves in neoplasm of the breast. Histologic and echographic correlation study. Radiol Med (Torino) 1986;72:195– 198. 15 Conant EF, Dillon RL, Palazzo J, Ehrlich SM, Feig SA. Imaging findings in mucin-containing carcinomas of the breast: correlation with pathologic features. Am J Roentgenol 1994;163:821–824. 16 Chopra S, Evans AJ, Pinder SE et al. Pure mucinous breast cancer-mammographic and ultrasound findings. Clin Radiol 1996; 51:421–424.
Correlation Between Ultrasound Characteristics, Mammographic Findings and Histological Grade in Patients with Invasive Ductal Carcinoma of the Breast PHIL IPPA M. L AMB, NI CH OL AS M . PE RRY, SARAH J. VIN NICOMBE , C L IVE A. W E L L S Department of Radiology, The Royal Hospitals NHS Trust, St. Bartholomew’s Hospital, West Smithfield, London EC1A 7BE, U.K. Received: 12 February 1999 Revised: 16 June 1999 Accepted: 7 July 1999 AIM: To investigate the relationship between ultrasound characteristics, mammographic findings and histological grade in cases of invasive ductal carcinoma which produce a mass on ultrasound. MATERIAL AND METHODS: A retrospective review was performed of the imaging findings in 120 patients diagnosed with invasive ductal carcinoma of the breast between January 1996 and December 1997. Imaging findings were correlated with the histological grade of tumour. RESULTS: High-grade tumours were significantly larger both on ultrasound and mammography (P < 0.016). A spiculated margin on mammogram was documented in 72% of low-grade tumours compared with only 24% of high-grade tumours (P ¼ 0.001). Twenty-two per cent of low-grade tumours had a poorly defined margin on mammography compared with 66% of high-grade tumours (P ¼ 0.001). At ultrasound, 16% of high-grade tumours (95% confidence limits 7–29%) had a welldefined margin. Acoustic enhancement was seen in 36% of high-grade tumours compared with only 9% of low and intermediate-grade tumours (P ¼ 0.003): 22% of all tumours showed acoustic enhancement. Acoustic shadowing was seen in 71% of low-grade tumours compared with only 28% of high-grade tumours (P ¼ 0.003). Malignant-type microcalcification was seen on mammogram in 6% of low-grade tumours compared with 31% of high-grade tumours (P ¼ 0.045). CONCLUSION: The classical appearance of a malignant breast mass as a spiculated mass on mammogram associated with acoustic shadowing on ultrasound is more typical of a low-grade tumour. In comparison, high-grade tumours are more likely to demonstrate posterior acoustic enhancement, and a proportion has a well-defined margin on ultrasound. Therefore, high-grade invasive ductal carcinoma may paradoxically display similar imaging features to a benign breast mass. Lamb, P. M. et al. (2000). Clinical Radiology 55, 40–44. q 2000 The Royal College of Radiologists Key words: breast carcinoma, invasive ductal carcinoma, ultrasound, attenuation characteristics, histological grade.
INTRODUCTION
The use of ultrasound as an adjunct to mammography in the diagnosis of breast cancer is well established [1–4]. An abnormal mass on mammography is reported in 50% of carcinomas of less than 10 mm diameter and 88% of those greater than 10 mm diameter [5]. A spiculated mass on mammography has been reported in up to 84% of breast malignancies [6] (Fig. 1). Historically, at ultrasound examination posterior acoustic shadowing was said to be associated with the majority of malignant breast masses: 70–80% overall and almost 100% of scirrhous (spiculated) lesions [3] (Fig. 2). However, it is now widely recognized that not all invasive Author for correspondence and guarantor of study: P. M. Lamb, 21 Old Sun Wharf, 40 Narrow Street, London E14 8DG, U.K. 0009-9260/00/010040+05 $35.00/0
breast masses demonstrate posterior shadowing: they may have no distal effect or even be associated with acoustic enhancement. In our clinical practice we have observed that tumours associated with distal acoustic enhancement tend to be those which are designated high-grade at subsequent histological examination (Fig. 3). The purpose of this study was to evaluate further the relationships between attenuation characteristics, mammographic findings and histological grade of breast cancer. The aims of the study were as follows. (i) To categorize the sonographic characteristics and mammographic features of invasive ductal carcinoma. (ii) To correlate the imaging findings with the histological grade of invasive ductal carcinoma. (iii) To ascertain whether the histological grade of invasive ductal carcinoma can be predicted from the imaging appearances. q 2000 The Royal College of Radiologists
IMAGING CHARACTERISTICS AND HISTOLOGICAL GRADE IN CARCINOMA OF THE BREAST
41
Fig. 2 – Acoustic shadowing associated with a low-grade (grade 1) invasive ductal carcinoma of the breast.
Fig. 1 – A low-grade (grade 1) invasive ductal carcinoma producing a spiculated mass on mediolateral oblique mammogram.
MATERIALS AND METHODS
A retrospective review was performed of the ultrasound and mammographic findings of 120 consecutive patients diagnosed with invasive ductal carcinoma at our institution between January 1996 and December 1997. The sample was obtained from a histopathology database: invasive ductal carcinoma accounts for 80% of all cases of invasive breast cancer and by excluding other histological subtypes (e.g. lobular and mucinous carcinomas) from our study we aimed to minimize confounding variables. Symptomatic and screen-detected cases were included and the women were aged 24 to 97 years (mean 58 years). Ultrasound scans were performed using an Acuson Sequoia 5L8 linear probe for the majority of cases, although some of the earlier studies were performed using an Aloka 121 SSD (KeyMed) dedicated 7.5 MHz sector probe. The images were analysed by two experienced breast radiologists. Features assessed on ultrasound were: (i) size of mass (mm); (ii) depth/width ratio; (iii) margin of mass (well or poorly defined); (iv) attenuation characteristics (shadowing, enhancement, mixed or no distal effect); (v) echotexture (homogeneous or heterogeneous); (vi) halo (present, absent). The term ‘halo’ refers to an irregular zone of increased echogenicity that may surround a tumour mass [7]. Similarly, corresponding mammograms were assessed for:
Fig. 3 – Acoustic enhancement associated with a high-grade (grade 3) invasive ductal carcinoma of the breast.
42
CLINICAL RADIOLOGY
120 patients diagnosed with invasive ductal carcinoma for the period January 1996–December 1997 was as follows: 19 grade 1 (low-grade) tumours; 37 grade 2 (intermediate-grade) tumours; 64 grade 3 (high-grade) tumours.
(i) size of mass (mm); (ii) margin (well-defined, poorly defined, partially defined or spiculated) [8]; (iii) microcalcification: the presence of malignant-type microcalcification was documented in case the calcification was altering the attenuation characteristics of the mass itself. The observers were not restricted to specific mammographic projections and their assessment related to those images that in their opinion provided the optimum image quality for the features being analysed. The observers were blinded to the histological grade assigned to each case and an attempt was made by them to predict the grade based on the combined mammographic and sonographic appearances. Histological grading was performed according to the Elston and Ellis protocol [9]. This involves the assessment of three components of tumour morphology: tubule formation, nuclear pleomorphism and mitotic frequency. Each is scored from 1 to 3, and adding the scores gives the overall grade. Score 3–5 6–7 8–9
Grade 1 2 3
Ultrasound Findings Ultrasound images were available for 104 cases: of these 17 were grade 1 tumours, 36 were grade 2 tumours and 51 were grade 3 tumours. The following features were found to be statistically significant when correlated with the histological grade (Table 1: Kruskal–Wallis ANOVA test by rank, chisquared test for trend and the Holm–Bonferroni correction): (i) an increase in size of tumour mass with increasing histological grade; (ii) 71% of grade 1 tumours demonstrated acoustic shadowing compared with only 28% of grade 3 tumours; (iii) 36% of grade 3 tumours demonstrated acoustic enhancement compared with only 9% of grade 1 and grade 2 tumours. Overall (for all histological grades), 41% of tumours showed acoustic shadowing while 22% showed acoustic enhancement. The size of tumour was compared between grades and the presence of acoustic shadowing or enhancement (Table 2). For grade 1 tumours, those that showed acoustic enhancement tended to be larger, but for grade 3 tumours this was not the case. There was no significant relationship between the size of the tumour and the nature of the distal effect. With regard to the depth/width ratio, 24% of all tumour masses had an increased ratio, but this was not grade specific (Table 1). Of those tumours which had a well-defined margin and showed acoustic enhancement, only one showed an increased depth/width ratio. The majority of masses had a heterogeneous appearance and the echo-texture was not grade specific. A halo was noted in 39% of all cases, but its presence was not significantly related to the histological grade.
Low Intermediate High
Statistical Methods Statistical analysis was performed using methods appropriate for non-parametric data: a Kruskal–Wallis analysis of variance (ANOVA) by rank for continuous data, chi-squared test for trend for categorical data and the Holm–Bonferroni correction [10,11]. The significance level was 5%. Regarding the radiologists’ attempt to predict histological grade, the positive predictive value was calculated for both high and low-grade tumours [12].
RESULTS
The distribution of histological grades within the data set of Table 1 – Ultrasound features Size (mm)
Grade 1 Grade 2 Grade 3 P-value Holm corrected
Margin
Echotexture
Median, inter-quarterile range
† Depth/width
Poorly defined
Heterogeneous
12, 8–14 14, 9–22 20, 11–27
6 (35) 12 (33) 7 (14)
17 (100) 34 (94) 43 (84)
13 (76) 28 (78) 38 (76)
0.006* 0.030*
0.203 0.609
0.098 0.392
0.981 0.981
Distal effect
Grade 1 Grade 2 Grade 3 P-value Holm corrected
Halo
Enhancement
Shadowing
None
Mixed
5 (29) 14 (39) 21 (43)
3 (18) 2 (6) 18 (36)
12 (71) 16 (44) 14 (28)
2 (12) 9 (25) 9 (18)
0 9 (25) 9 (18)
0.619 1.000
0.003 0.018
* Kruskal–Wallis ANOVA test by rank; chi-squared test for trend for all other parameters; significance level ¼ 0.05; (%) in parentheses.
43
IMAGING CHARACTERISTICS AND HISTOLOGICAL GRADE IN CARCINOMA OF THE BREAST
Table 2 – Comparison of size of tumours showing acoustic enhancement or shadowing on ultrasound Size (median, inter-quartile range, mm)
All Grade 1 Grade 2 Grade 3
Acoustic enhancement
Acoustic shadowing
16, 15, 22, 18,
12, 9–20 9, 8–12 12, 9–19 16, 10–24
13–26 6–17 14–30 13–26
P-value*
0.635 0.151 0.219
* Kruskal–Wallis ANOVA test by rank.
histological grade by choosing either a high-grade or low-grade category (Table 5). The results indicate that for high-grade tumours there was a positive predictive value of 84%, while for low-grade tumours this was 73%. The specificity of overall radiological prediction of histological grade was 89%, while the sensitivity was 65%. In summary, the significant results in this study are: (i) highgrade (grade 3) tumours tend to be larger both on ultrasound and mammography; (ii) low-grade (grade 1) tumours tend to produce a spiculated mass on mammogram and to demonstrate acoustic shadowing on ultrasound; (iii) 36% of high-grade tumours demonstrate acoustic enhancement at ultrasound examination.
Mammographic Findings
DISCUSSION
Mammograms were available for 117 patients: of those in whom a mass was present on the mammogram (112 cases), 18 were grade 1 tumours, 36 were grade 2 tumours and 58 were grade 3 tumours. The following were found to be statistically significant when correlated with the histological grade (Table 3: Kruskal–Wallis ANOVA test by rank, chi-squared test for trend and Holm–Bonferroni correction): (i) an increase in size of tumour mass with increasing histological grade; (ii) 72% of grade 1 tumours demonstrated a spiculated margin compared with only 24% of grade 3 tumours; (iii) 22% of grade 1 tumours had a poorly defined margin compared with 66% of grade 3 tumours; (iv) malignant-type microcalcification was seen in 6% of grade 1 tumours compared with 31% of grade 3 tumours. The size of tumour was compared between grades and the presence of a poorly defined or spiculated margin (Table 4). There was no significant relationship between the size of the tumour and the assessment of the margin of the tumour mass on mammogram. Analysis of mammographic and sonographic features for all histological grades revealed that 28/100 (28%) of all masses had both a spiculated margin on mammography and demonstrated acoustic shadowing on ultrasound: of these, 10/28 (36%) were grade 1 tumours, 11/28 (39%) were grade 2 tumours and 7/28 (25%) were grade 3 tumours.
Radiological Prediction of Histological Grade Based on the combined appearances on mammography and ultrasound an attempt was made by the radiologist to predict the
The increase in size of tumour mass with increasing histological grade is to be expected, due to the faster mitotic rate in high-grade tumours. However, the differences in attenuation characteristics between the histological grades of invasive ductal carcinoma have not previously been reported and need to be explained. Previous studies which have examined the relationship between ultrasound characteristics and pathological features of breast cancer have attributed particular importance to the presence of fibrous connective tissue in those cases that demonstrate acoustic shadowing, the so-called desmoplastic reaction. Conversely it has been suggested that those tumours which demonstrate acoustic enhancement are more cellular [7,13]. However, a study by Gozzi et al. [14] reported that it is not the absolute amount of fibrous tissue within a tumour, but the organization of the tissue that determines the attenuation characteristics. In fact, in their study the tumours which demonstrated acoustic enhancement were less cellular (and contained more fibrous tissue). Some studies have reported acoustic enhancement in mucincontaining tumours (but these studies have involved only relatively small numbers of patients – eight in each) [15,16]. We intend to carry out further histological assessment with regard to the amount of fibrous tissue, cellularity and organization of tissues within malignant breast masses. The findings will be correlated with the attenuation characteristics that we have reported above. Regarding the radiologist’s attempt to predict histological
Table 3 – Mammographic features Size (mm)
Margin
Median, inter-quartile range
Poorly defined
Partially defined
Spiculated
Well-defined
Microcalcification
Grade 1 Grade 2 Grade 3
12, 10–15 15, 10–20 17, 15–30
4 (22) 14 (39) 38 (66)
1 (6) 0 3 (5)
13 (72) 22 (61) 14 (24)
0 0 3 (5)
1 (6) 6 (16) 19 (31)
P-value Holm corrected
0.016* 0.032*
0.001 0.003
0.045 0.045
* Kruskal–Wallis ANOVA test by rank; chi-squared test for trend for all other parameters; significance level ¼ 0.05; (%) in parentheses.
44
CLINICAL RADIOLOGY
Table 4 – Comparison of size of tumours showing poorly defined or spiculated margin Size (median, inter-quartile range, mm) Poorly defined margin
Spiculated margin
15, 11, 13, 17,
15, 12, 18, 15,
However, in our study these were not reliable indicators and the absence of these signs did not preclude malignancy. All solid lesions at ultrasound should therefore be subject to cytological or histological sampling and correlation made with the imaging appearances.
P-value*
REFERENCES All Grade 1 Grade 2 Grade 3
12–25 9–13 11–18 12–28
11–20 10–15 10–20 12–30
0.459 0.304 0.535
* Kruskal–Wallis ANOVA test by rank.
Table 5 – Radiologists’ prediction of histological grade Actual grade Predicted grade
High grade
Low grade
High grade Low grade
32 17
6 47
Positive predictive value for high-grade tumours ¼ 84%; positive predictive value for low-grade tumours ¼ 73%; sensitivity ¼ 65%; specificity ¼ 89%.
grade of breast cancer, the imaging characteristics appear to be a reliable predictor, particularly for high-grade tumours (positive predictive value ¼ 84%). Theoretically this may have implications for patient management in the future. However, at present core biopsy remains the most reliable method of ascertaining histological grade pre-operatively. In conclusion, the classical appearances of a spiculated mass on mammography associated with acoustic shadowing on ultrasound are more likely to represent a low or intermediategrade tumour (75% of cases in this study). In this study, 22% of all cases of invasive ductal carcinoma demonstrated acoustic enhancement and radiologists performing breast imaging should be aware that many of these could be misinterpreted as benign. Such tumours are more likely to be those which are found to be high-grade at subsequent histological examination. Radiologists should make a careful search for other features which may indicate malignancy, such as an increase in the depth/width ratio or disruption of adjacent normal tissues.
1 Jacob D, Brombart JC, Muller C, Lefebvre C, Massa F, Depoerck A. Analysis of the results of 137 subclinical breast lesions excisions. Value of ultrasonography in the early diagnosis of breast cancer. J Gynecol Obstet Biol Reprod (Paris) 1997;26:27–31. 2 Jackson VP, Reynolds HE, Hawes DR. Sonography of the breast. Semin Ultrasound CT MR 1996;17:460–475. 3 Tohno E, Cosgrove DO, Sloane JP. Ultrasound Diagnosis of Breast Diseases. Edinburgh: Churchill Livingstone; 1994, chapter 9, pp. 158– 162. 4 Tucker AK. Textbook of Mammography, 1st ed. Churchill Livingstone, 1993, chapter 9, pp. 219–226. 5 Andersson I. Radiographic screening for breast carcinoma III: appearance of carcinoma and number of projections to be used at screening. Acta Radiol Diagn 1981;22:407–420. 6 Barth V. Atlas of Diseases of the Breast. Stuttgart: Georg Thieme, 1979. 7 Kobayashi T. Diagnostic ultrasound in breast cancer: analysis of retrotumerous echo patterns correlated with sonic attenuation by cancerous connective tissue. JCU J Clin Ultrasound 1979;7:471–479. 8 Burnett SJD, Ng YY, Perry NM et al. Benign biopsies in the prevalent round of breast screening: a review of 137 cases. Clin Radiol 1995; 50:254–258. 9 Elston CW, Ellis IO. Pathological prognostic factors in breast cancer No. 1. The value of histological grade in breast cancer: experience from a large study with long-term follow-up. Histopathology 1991;19:403– 410. 10 Aicken M, Gensler H. Adjusting for multiple testing when reporting research results: The Bonferroni vs Holm methods. Am J Public Health 1996;86:726–728. 11 Altman DG. Practical Statistics for Medical Research. London: Chapman & Hall, 1991;211–213. 12 Goldin J, Sayre JW. A guide to clinical epidemiology for radiologists: part I. Study design and research methods. Clin Radiol 1996;51:313– 316. 13 Kossoff G. Causes of shadowing in breast sonography. Ultrasound Med Biol 1988;14:211–215. 14 Gozzi G, Cressa C, Bazzocchi M, Stanta G, Vidali C. Causes of attenuation of the sound waves in neoplasm of the breast. Histologic and echographic correlation study. Radiol Med (Torino) 1986;72:195– 198. 15 Conant EF, Dillon RL, Palazzo J, Ehrlich SM, Feig SA. Imaging findings in mucin-containing carcinomas of the breast: correlation with pathologic features. Am J Roentgenol 1994;163:821–824. 16 Chopra S, Evans AJ, Pinder SE et al. Pure mucinous breast cancer-mammographic and ultrasound findings. Clin Radiol 1996; 51:421–424.