CT imaging… Potential tool for diagnosis of breast cancer

Clinical Radiology (2008) 63, 1213e1227

ORIGINAL ARTICLE

Dual time point FDG-PET/CT imaging. Potential tool for diagnosis of breast cancer A.A. Zytoona,b,*, K. Murakamib, M.R. El-Kholya, E. El-Shorbagyc a

Radiology Department, Menoufiya University School of Medicine, Egypt, bPET Center, Dokkyo University School of Medicine, Japan, and cRadiology Department, National Liver Institute, Egypt Received 13 September 2007; received in revised form 1 March 2008; accepted 14 March 2008

AIM: This prospective study was designed to assess the utility of the dual time point imaging technique using 2- [18F]-fluoro-2-deoxy-D-glucose (FDG) positron-emission tomography/computed tomography (PET/CT) to detect primary breast cancer and to determine whether it is useful for the detection of small and non-invasive cancers, as well as cancers in dense breast tissue. METHODS: One hundred and eleven patients with newly diagnosed breast cancer underwent two sequential PET/CT examinations (dual time point imaging) for preoperative staging. The maximum standardized uptake value (SUVmax) of FDG was measured from both time points. The percentage change in SUVmax (DSUVmax%) between time points 1 (SUVmax1) and 2 (SUVmax2) was calculated. The patients were divided into groups: invasive (n ¼ 82), non invasive (n ¼ 29); large (>10 mm; n ¼ 80), small (10 mm; n ¼ 31); tumours in dense breasts (n ¼ 61), and tumours in nondense breasts (n ¼ 50). The tumour:background (T:B) ratios at both time points were measured and the DSUVmax%, DT:B% values were calculated. All PET study results were correlated with the histopathology results. RESULTS: Of the 111 cancer lesions, 88 (79.3%) showed an increase and 23 (20.7%) showed either no change [10 (9%)] or a decrease [13 (11.7%)] in the SUVmax over time. Of the 111 contralateral normal breasts, nine (8.1%) showed an increase and 102 (91.9%) showed either no change [17 (15.3%)] or a decrease [85 (76.6%)] in the SUVmax over time. The mean  SD of SUVmax1, SUVmax2, D%SUVmax were 4.9  3.6, 6.0  4.5, and 22.6  13.1% for invasive cancers, 4.1  3.8, 4.4  4.8, and 2.4  18.5% for non-invasive cancers, 2.3  1.9, 2.7  2.3, and 12.9  21.1% for small cancers, 5.6  3.7, 6.8  4.8, and 17.3  17.1% for large cancers, 4.9  3.7, 5.8  4.8, and 15.1  17.6% for cancers in dense breast, and 4.5  3.6, 5.4  4.5, and 17.2  19.2% for cancers in non-dense breast. The receiver-operating characteristic (ROC) analysis suggested DSUVmax% of 8% as the only significant cut-off for discrimination between invasive and non-invasive cancer (sensitivity 84.1%, specificity 75.9%, p < 0.0001). CONCLUSION: Dual time point FDG-PET/CT improves the discrimination between non-invasive and invasive cancers, and provided superior sensitivity for the detection of small cancers and cancers in dense breast. ª 2008 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved.

Introduction Breast cancer is the commonest cancer in women and is the second leading cause of cancer death.1 Although it is curable when detected early, about one third of women with breast cancer die of the disease.2 * Guarantor and correspondent: A.A. Zytoon, Faculty of Medicine, Radiology Department, Menoufiya University Shebin El-Koom, Menoufiya, Egypt. Tel.: þ002 048 2323888; fax: þ002 048 2228302. E-mail address: [email protected] (A.A. Zytoon).

Positron emission tomography (PET) has gained widespread acceptance for the diagnosis, staging, and management of a variety of malignancies, including breast cancer.3e5 2- [18F]-fluoro-2-deoxyD-glucose (FDG)-PET has been quite effective in detecting nodal and distant metastases by acquiring a single whole-body examination in patients with breast cancer.6e9 It is likely that certain early breast tumours, with less biologically aggressive features, are less glycolytic than more advanced breast cancer and will evade detection due to insufficient FDG uptake, not simply because of

0009-9260/$ - see front matter ª 2008 The Royal College of Radiologists. Published by Elsevier Ltd. All rights reserved. doi:10.1016/j.crad.2008.03.014

1214

instrumentation limitations.10 Some types of cancers, for example, well-differentiated and lobular carcinomas of the breast have an abnormally low FDG uptake, which is well below the diagnostic threshold for FDG uptake in malignant lesions.11 This leads to false-negative readings, which result in a lower sensitivity of PET in detecting true malignancies.12 In primary breast cancer diagnosis, FDGPET has sensitivities of 63e96% and specificities of 75e100%.6,7,13,14 Single-time-point standardized uptake value (SUV) analysis is a suboptimal method for assessing suspected breast cancer, and any method to improve the accuracy of FDG-PET in breast cancer characterization would be of value.12 The biological characteristics of the tumour, such as histological subtype (invasive versus noninvasive), are considered important prognostic factors in patients with breast cancer and may influence glucose metabolism as detected by PET.15 The variation in body habitus, duration of uptake, plasma glucose levels, and partial-volume effects are important factors that influence the SUV.16 Changes in these parameters, especially elevated plasma glucose levels, short uptake periods, and small lesion sizes, will lead to low SUVs in malignant lesions.17 PET usually is performed 1 h after FDG administration. The accuracy of FDG-PET/computed tomography (CT) could be further enhanced by delayed or dual-time point imaging.18e23 In animal experiments, FDG accumulation in tumours increases over a period of 2 or 3 h.19 Added advantages of dual time point imaging have been reported in head and neck, lung, pancreatic, and cervical malignancies; substantially higher SUVs are seen on delayed imaging compared with the initial scans.17,19,21,24e27 This approach can also improve the sensitivity of the technique for primary and metastatic sites.12,17,19,21,25,28,29 The purpose of this study was to determine whether there is an advantage of dual time point FDG-PET/CT, measuring the SUVs, in patients who have primary breast cancer.

Patients and methods Patient population This prospective study was approved by the review board of our institute. Written informed consent was obtained from all patients. One hundred and eleven patients (all women; age mean 55.8  10.4, 95%CI 53.9e57.8, median 58 years, range 32e78 years) with newly diagnosed unilateral breast cancer were analysed. Sixty-one (55%) patients had

A.A. Zytoon et al.

grade III or IV mammographic density (dense breasts), whereas 50 (45%) patients had grade I or II breast density (non-dense breasts) according to the American College of Radiology (ACR) Lexicon criteria. On the basis of the surgical histopathological examination, the mean breast cancer diameter was 2.1  1 cm (95%CI 1.9e2.3, median 2.3 cm, range 0.6e5.3 cm; Table 1). The diagnosis was suggested by clinical examination (performed by a breast surgeon) and mammography. The patients were imaged using digital mammography, ultrasonography, CT, magnetic resonance imaging (MRI), and FDG-PET/CT. After FDG-PET/CT, all of the patients had surgery (breast-conserving surgery or mastectomy). Surgical histology results were considered to provide the definitive diagnosis against which the FDG-PET/CT study results were compared. The mean interval between surgery and dual time point FDG-PET/CT was 16.3  10.4 days. None of the patients had received chemotherapy or radiation therapy before they underwent dual time point FDG-PET/CT for preoperative staging. Table 1 Basic clinical and pathological characteristics of the patients No. of patients Sex Age (years); mean (95% CI, median, range)

111 All female 55.8  10.4 (53.9e57.8, 58, 32e78)

Histopathology DCIS/IDC/ILC 29/76/6 Size (cm); 2.1  1.0 (1.9e2.3, 2.3, 0.6e5.3) mean (95% CI, median, range) SUVmax breast tumour; mean (95% CI, median, range) SUVmax1 4.7  3.6 (4e5.4, 3.6, 0.6e18.5) SUVmax2 5.6  4.6 (4.8e6.5, 4.6, 0.6e23.8) D SUVmax% 16.1  18.3 (12.6e19.5, 19.4, 52.9e 46.5) SUVmax normal contralateral breast; mean (95% CI, median, range) SUVmax1 1.2  0.4 (1.1e1.3, 1.1, 0.5e2.5) SUVmax2 1  0.5 (0.9e1.1, 0.9, 0.2e3.2) D SUVmax% 16.6  17.9 (19.9e13.2, 19, 80e31) T:B ratio; mean (95% CI, median, range) T:B1 4.3  4.5 (3.5e5.2, 3.3, 0.5e31) T:B2 6.5  7.1 (5.2e7.9, 4.8, 0.7e47) D T:B% 48.2  36.9 (41.3e55.1, 51.2, 48.0e 160) DCIS, ductal carcinoma in situ, IDC, invasive ductal carcinoma; ILC, invasive lobular carcinoma; SUVmax, maximum standardized uptake value; D SUVmax%, percentage change in SUVmax; T:B tumour:background ratio; DT:B%,percentage change over time in the T:B ratio.

Dual time point FDG-PET/CT imaging

1215

Figure 1 Maximum SUV change over time between the time point 1 and time point 2 in breast cancer (a) and normal breast (b). While tumour maximum SUV increased over time (p < 0.0001), the physiological uptakes of normal breast tissue decreased over time (p < 0.0001) in its mean value.

FDG-PET/CT imaging and assessment Patients fasted for at least 6 h before the PET examination and had blood glucose levels less than 140 mg/dl at the time of injection. PET/CT was performed with a dedicated whole-body PET machine (Biograph Sensation 16 PET-CT; Siemens Medical Systems; CTI/Siemens, Knoxville, TN, USA). All the patients underwent dual time point imaging for preoperative staging with an average interval of approximately 50e60 min between the two phases. The first examination was performed as whole-body images from head to thigh, with acquisition of six to seven bed positions; resulting in a complete axial length of 80e100 cm. Imaging began 55e60 min after injection of FDG (4.5 MBq/kg of body weight). In all cases, FDG radiochemical purity was >95% and specific activity was >47 GBq/ mmol. Eight to 15 ml FDG was infused over 2e3 min in the antecubital vein contralateral to the affected breast. The second examination imaged the chest only, with acquisition of one or two bed positions; resulting in a complete axial length of 15e25 cm. Imaging began at approximately 110e120 min after FDG injection. Transmission scans were performed for all patients to provide attenuation correction with CT. The PET section thickness was 3.4 mm. Three-dimensional (3D) data were acquired without septa, with image reconstruction and scatter correction using a computer system (e-soft nuclear medicine acquisition, processing, and viewing software). Visual assessment included assessing both sets of images at the same time. Analysis was based on measuring the maximum FDG uptake in the tumour site and in the contralateral normal breast at both time points.

Image analysis PET images were reconstructed using measured attenuation correction, dead-time correction, and

decay correction to the beginning of each image. The maximum SUV (SUVmax) of FDG was measured from both time points: time point 1 (SUVmax1) and time point 2 (SUVmax2). Visual assessment, image interpretation, and data analysis were performed independently by two nuclear medicine physicians. They were aware of the patients’ clinical history, which was provided by the referring physician, but were blinded to the results of other imaging studies. When there was a difference between the two observers, a mean was calculated to determine the final SUV. After image reconstruction, a free-hand region of interest (ROI) was carefully overlaid onto three to six PET image sections of the lesion and contralateral normal breast tissue. The SUVs for the second time point were obtained with the same technique as those used for the first time point. The maximum FDG uptake on the consecutive early images was obtained for quantitative measurement of the metabolic activity of the tracer. From the ROIs, the SUV was calculated according to the following formula: SUV ¼

Mean ROI activityðMBq=gÞ Injected dose ðMBqÞ=body weightðgÞ

The DSUVmax% change over time [retention index (RI)] is defined as follows: ðSUVmax2  SUVmax1Þ ðSUVmax1Þ  100 where SUVmax1 and SUVmax2 are the values of SUVmax at the initial and delayed phases, respectively. DSUVmax% is usually expressed as a percentage. The percentage change in the SUVmax of breast cancer and contralateral normal breast between the two time points was calculated. For the purpose of statistical analysis, patients were divided into groups according to surgical

1216

A.A. Zytoon et al.

Breast Cancer (n = 111)

Normal Contralateral Breast (n = 111)

25.0

3.25 3.00

22.5

2.75 2.50

17.5

2.25

15.0

2.00

SUVmax

SUVmax

20.0

12.5 10.0

1.75 1.50 1.25 1.00

7.5

0.75

5.0

0.50 2.5 0.0

0.25 SUVmax1

0.00

SUVmax2

SUVmax1

SUVmax2

(b)

(a) 50

Tumor-Background Ratio (n = 111)

45 40 35

T-B ratio

30 25 20 15 10 5 0

T-B1

T-B2

(c) Figure 2 Before and after, symbols and lines vertical graph. Comparison of FDG uptake by breast tumour and physiological background (normal contralateral breast) at time point 1 and time point 2, evaluated by the SUVmax (a and b) and by the tumour:background (T:B) ratio (c). Whilst there was a significant increase in the SUVmax and T:B ratio of the tumour (p < 0.0001), there was a significant decrease in the SUVmax of the normal breast tissue (p < 0.0001) from time point 1 to time point 2.

Dual time point FDG-PET/CT imaging

1217

25

SUVmax2

20

15

10

5

0 0

5

10

15

20

2.0

2.5

SUVmax1

(a) 3.5 3.0

SUVmax2

2.5 2.0 1.5 1.0 0.5 0.0 0.5

1.0

1.5

SUVmax1

(b) 50

histopathology results into: invasive and noninvasive cancer, large (>10 mm) and small (10 mm), and according to ACR Lexicon criteria applied to the underlying breast into: cancer in dense breast tissue and cancer in non-dense breast tissue. The tumour:background (T:B) ratio of SUVmax was calculated at both time points; 1 (ratio 1) and 2 (ratio 2) for each lesion for all groups, and the percentage change in these ratios over time was calculated. It has been reported that there is no significant difference in the maximum and mean SUV between right and left breasts, as well as the right and left nipple.30 Thus, for the purpose of creation of T:B ratio, the same patient’s contralateral normal breast tissues SUVmax results were used as a reference. The SUVmax of the tumour was divided by the patient’s contralateral normal breast SUVmax for both time points to generate these ratios. For interpretation of visual assessment of early and delayed images, tumour FDG uptake less intense than, or equivalent to, the intensity of the normal breast tissue background physiological FDG uptake was regarded as undetectable (T:B ratio 1). Uptake higher than the normal breast tissue background physiological uptake was regarded as detectable (T:B ratio >1). The PET results from different semiquantitative parameters (SUV, T:B ratio) were then compared together between different groups to estimate the optimal method with the highest accuracy for detection of breast cancer.

Statistical analysis

40

T/B2

30

20

10

0 0

5

10

15

20

25

30

35

T/B1

(c) Figure 3 Scattergrams with linear regression show the comparison between breast cancer SUVmax (a), physiological background SUVmax (b) and the tumour:background (T:B) ratio (c) change over time between time point 1 (early phase) and point 2 (delayed phase). Linear regression analysis shows significant correlation for the parameters change over time. Slopes are 0.8, 0.7 and 0.6, respectively. r ¼ 0.981 (95%CI 0.972e0.987,

All semiquantitative data were expressed in terms of mean  SD. All statistical analyses were two-sided with significance defined as p < 0.05. Data analyses were performed using MedCalc statistical software for windows. In order to identify the power of the SUVmax for discrimination between invasive and non-invasive cancer, the cut-off scores were created by receiver-operating characteristic (ROC) curve analysis. Clinically, relevant cut-off scores were generated such that they maximized both sensitivity and specificity for each outcome under study. Semiquantitative parameters (SUVmax and T:B ratio) of each group were analysed using the non-parametric Wilcoxon signed-rank and ManneWhitney U test. Sensitivity, specificity, positive predictive value, and negative predictive value were obtained

p < 0.0001), 0.804 (95%CI 0.726e0.861, p < 0.0001), 0.950 (95%CI 0.928e0.965, p < 0.0001), respectively.

8e46.0 53e22 26 (19e29) 0 (12e4) 22.6  13.1 (19.7e25.5) 2.4  18.5 (9.5e4.6) <0.0001 0.6e23.8 0.6e18.6

Results One hundred and eleven breast cancers were investigated in the study; 82 lesions (73.9%) were invasive cancer (invasive ductal carcinoma IDC, n ¼ 76; invasive lobular carcinoma ILC, n ¼ 6) and the remaining 29 (26.1%) were non-invasive cancers (ductal carcinoma in situ, DCIS). The mean breast cancer diameter at histology was; 2.1  1 cm (median 2.3 cm, range 0.6e5.3 cm). Eighty lesions were >1 cm; mean 2.6  0.9 (median 2.4 cm, range 1.3e5.3 cm), whereas 31 were 1 cm; mean 0.9  0.1 cm (median 1 cm, range 0.6e1 cm).

SUV measurements

Values in parentheses are 95% confidence intervals (95%CI). D%, percent change.

0.6e18.5 0.7e15.3 3.6 (3.5e4.3) 2.6 (1.8e5.6) 4.9  3.6 (4.1e5.7) 4.1  3.8 (2.7e5.6) 0.3305 Invasive cancer (n ¼ 82) Non-invasive cancer (n ¼ 29) p-Value

Median

with 95% confidence intervals (CI). The Spearman’s rank test was performed to determine the correlation coefficient of the change of semiquantitative parameters between the early and delayed phases.

6  4.5 (5.1e7) 4.4  4.8 (2.6e6.3) 0.1077

4.7 (4.3e5.2) 2.7 (1.6e5.2)

Range Median Range Median SUVmax2

Mean Mean

Range SUVmax1 Breast Cancer

Table 2

Maximum standardized uptake value (SUVmax) changes over time in invasive and non-invasive breast cancer

DSUVmax%

A.A. Zytoon et al.

Mean

1218

In all patients, the SUVmax could be calculated for breast tumours and for the contralateral normal breasts. Of the 111 cancers, 88 (79.3%) showed an increase and 23 (20.7%) showed either no change [10 (9%)] or a decrease [13 (11.7%)] in the SUVmax over time (Fig. 1a). Breast cancer SUVmax1, SUVmax2, and DSUVmax% mean values were; 4.7  3.6, 5.6  4.6, and 16.1  18.3%. The percentage change (increase) between SUVmax1 and SUVmax2 over time was significant (p < 0.0001; Fig. 2a). Fig. 3a is a plot of the change in tumour SUVmax against time interval between the dual scans. It demonstrates a significant correlation between SUVmax change and time interval, with a slope of 0.8 and a correlation coefficient (r) of 0.981, 95%CI 0.972e0.987, p < 0.0001. Of the 111 contralateral normal breasts, nine (8.1%) showed an increase and 102 (91.9%) showed either no change [17 (15.3%)] or a decrease [85 (76.6%)] in the SUVmax over time (Fig. 1b). Of the nine contralateral normal breasts that showed an increase in the SUVmax over time, there were no obvious focal lesions in the follow-up imaging. Normal contralateral breasts SUVmax1, SUVmax2, and DSUVmax% mean values were: 1.2  0.4, 1  0.5, and 16.6  17.9% (Table 1). The percentage change (decrease) between SUVmax1 and SUVmax2 over time was significant (p < 0.0001; Fig. 2b). Fig. 3b is a plot of the change in contralateral normal breast SUVmax against the time interval between the dual scans. It demonstrates a significant correlation between SUVmax change

Dual time point FDG-PET/CT imaging

1219

Table 3 Discrimination between histopathological types of breast cancer (invasive and non-invasive): comparative performance by maximum standardized uptake value (SUVmax)1, SUVmax2, and DSUVmax% cut-off values as generated by receiveroperating characteristic (ROC) curve analysis Parameter

Cut-off value

Sensitivity

Specificity

PPV

NPV

Accuracy

SUVmax1 SUVmax2 DSUVmax%

3 3.1 8%

65.9% (54.5e75.7%) 72% (60.8e81%) 84.1% (74e91%)

65.5% (45.7e81.4%) 62.1% (42.4e78.7%) 75.9% (56.1e89%)

84.4% (72.7e91.9%) 84.3% (73.2e91.5%) 90.8% (81.4e95.9%)

40.4% (26.7e55.7%) 43.9% (28.8e60.1%) 62.9% (44.9e78%)

65.8% 69.4% 82%

Values in parentheses are 95% confidence intervals (95%CI). PPV, positive predictive value; NPV, negative predictive value.

and time interval, with a slope of 0.7 and an r of 0.804, 95%CI 0.726e0.861, p < 0.0001. The T:B ratio of the SUVmax was generated from both time point 1 (T:B1 ratio) and time point 2 (T:B2 ratio) for each lesion, and the percentage change in these ratios over time was calculated. For all tumours (n ¼ 111); the mean T:B1, T:B2, and DT:B% ratios were; 4.3  4.5, 6.5  7.1, and 48.2  36.9 (Table 1). The percentage change (increase) between T:B1 ratio and T:B2 ratio over time was significant (p < 0.0001; Fig. 2c). Fig. 3c is a plot of the change in T:B ratio against the time interval between the dual scans. It

demonstrates a significant correlation between the T:B ratio change and time interval, with a slope of 0.6 and an r of 0.950, 95%CI 0.928e0.965, p < 0.0001. While tumour SUVmax increased over time by 19.1%, the mean physiological uptake of normal breast tissue decreased over time by 16.7% in mean value. Comparing changes in tumour uptake over time against those of matched contralateral normal breast tissue (evolution over time of the ratio tumour maximum FDG uptake/physiological background maximum FDG uptake; T:B ratio) revealed a mean increase by 51.2%.

Figure 4 A 45-year-old woman with 1.5 cm right breast mass. Dual time point FDG-PET/CT imaging was performed. (a) Axial sections were obtained at the first time point and revealed a focus of faint FDG activity hardly discriminated from the background physiological FDG uptake, corresponding to the location of the nodule (thin arrows). (b) Corresponding images were acquired at a second time point (thick arrows). The SUVmax of the right breast cancer in first image set was 1.4, whereas that of second set was 1.8. The percent increase of DSUVmax% was 28.6%. The background FDG uptake SUVmax was 1.4 in first image set, whereas that of second set was 1.3. The percent decrease of DSUVmax% was 7.1%. Therefore, the increase in the T:B ratios between first images set and second image set was 40%. Surgical histopathology confirmed ductal carcinoma in situ (papillotubular type).

1220

A.A. Zytoon et al.

SUVmax1

SUVmax2

20

25

18 16

20

14 12

15

10 8

10

6 4

5

>3.0 Sens: 65.9 Spec: 65.5

2 0 CIS

>3.1 Sens: 72.0 Spec: 62.1

0 CIS

IC

IC

(a)

(b)

SUVmax%

100

60 40 20 >8.0 Sens: 84.1 Spec: 75.9

0 -20

Sensitivity

80 60 40 SUVmax1 SUVmax2 SUVmax%

20

-40 -60

0 IC

CIS

0

20

40

60

80

100

100-Specificity

(c)

(d)

Figure 5 Interactive dot diagram of the findings of SUVmax1 (a), SUVmax2 (b), DSUVmax% (c), against tumour histopathology. In the graph the data of the invasive and non-invasive cancer groups are displayed as symbols on two vertical axes. The horizontal line indicates the cut-off point with the best separation (minimal false-negative and false-positive results) between the two groups. The corresponding test characteristics sensitivity and specificity are shown at the right side of the screen display. (d) Comparative ROC analysis of the three parameters in the differentiation between invasive and non-invasive breast cancer. The area under curve (Az)  SE for SUVmax1 ¼ 0.6460.056, SUVmax2 ¼ 0.663  0.055 and DSUVmax% ¼ 0.813  0.041. Comparative ROC analysis revealed significant differences among the performances of DSUVmax% versus SUVmax1 (p ¼ 0.007) and DSUVmax% versus SUVmax2 (p ¼ 0.008).

Dual time point FDG-PET/CT and non-invasive breast cancer The means  SD (median) of the SUVmax1, the SUVmax2, and D SUVmax% for invasive cancer (n ¼ 82) were 4.9  3.6 (3.6), 6  4.5 (4.7), and 22.6  13.1% (26%). The same values for noninvasive cancer (n ¼ 29) were 4.1  3.8 (2.6), 4.4  4.8 (2.7), and 2.4  18.5% (0%), respectively. Because of some outlier values that resulted in high SD, the confidence intervals (CIs) for these data were calculated as reported. When the SUVmax1, SUVmax2 and DSUVmax% among non-invasive, and invasive cancer groups were compared separately, the difference between

the parameters was statistically significant for DSUVmax% (p < 0.0001), but not for SUVmax2 (p ¼ 0.1077), and SUVmax1 (p ¼ 0.3305; Table 2). The accuracy of various SUVmax parameters for discrimination between invasive and non-invasive breast cancer was investigated by ROC analysis. The ROC analysis suggested the best cut-off value of DSUVmax% for the accurate discrimination between invasive and non-invasive cancer was 8% (sensitivity 84.1%, specificity 75.9%, accuracy 82%). Maximum FDG uptake change over time (DSUVmax%) has a sensitivity of 96.6% (95CI% 82.2e99.9%; p ¼ 0.0007), for the detection of non-invasive breast cancer versus 73.3% (95CI% 44.9e92%; p ¼ 0.2303), for SUVmax2, and 44.4%

Dual time point FDG-PET/CT imaging

1221

Figure 6 A 55-year-old woman with a 4.4 cm left breast mass examined using dual time point FDG-PET/CT. (a) PET images, coronal sections in top row were obtained at the first time point. (b) Corresponding axial images in bottom row were acquired at the second time point. Images in both sets clearly show the primary lesion. However, the intensity of the uptake was substantially higher on the delayed images. In addition, axillary lymph node metastasis was faintly visualized on first set but was clearly demonstrated on second set. The SUVmax1 of the primary lesion in the first image set was 18.5, whereas that of the second set was 23.8. The percent increase in SUV of the lesion was 29%. SUVmax1 of metastatic right axillary lymph node was 2.3 in first set (thin arrow) and increased to 4.2 in second set (thick arrow). Surgical pathology confirmed invasive ductal carcinoma (mixed papillotubular and scirrhous types) with axillary lymph node metastasis.

(95CI% 21.6e69.2%; p ¼ 0.9590), for SUVmax1. This means that the accuracy of SUVmax1 and SUVmax2 for discrimination between invasive and non-invasive breast cancer was lower than DSUVmax%, and the difference was not statistically significant (Tables 2 and 3). Furthermore, comparative ROC analysis for the SUVmax1, SUVmax2, and DSUVmax%, revealed significant differences among the performances of the DSUVmax% versus SUVmax1 (p ¼ 0.007) and DSUVmax% versus SUVmax2 (p ¼ 0.008) for differentiation between invasive and non-invasive cancer (Figs. 4, 5 and 6). The means  SD (median) of the T:B1 ratio, T:B2 ratio, and percentage change over time between the T:B ratios at dual time imaging (DT:B ratio%) were 4.2  2.9 (3.3), 6.4  4.6 (5.8), and 52.6  32.0% (56.3), for invasive cancer, 4.6  7.5 (2.2), 6.8  11.6 (3.4), and 35.7  46.5% (25.0), for non-invasive cancer. When T:B1 ratio, T:B2 ratio, and DT:B ratio% among invasive, and noninvasive cancer groups compared separately, the difference between parameters was statistically significant for DT:B ratio% (p ¼ 0.0093) but not

for T:B2 ratio (p ¼ 0.8499) and T:B1 ratio (p ¼ 0.7696;Table 4). Seventeen of 29 (58.6%) non-invasive breast cancers with an SUVmax higher than the normal background uptake could be discriminated well by visual assessment in the early phase (T:B1 ratio >1). More cancers (n ¼ 21; 72.4%) were found to have T:B2 ratio >1 at the delayed phase. Comparing both phases, more cancers (n ¼ 25; 86.2%) were found to have T:B ratio >0%. Concerning non-invasive cancer, regression analysis demonstrates a significant correlation between T:B ratio change and time interval, with a slope of 0.6 and an r of 0.904, 95%CI 0.804e0.954, p < 0.0001.

Dual time point FDG-PET/CT and small breast cancer The means  SD (median) of the SUVmax1, the SUVmax2, and D SUVmax% were 2.3  1.9 (1.8), 2.7  2.3 (1.8) and 12.9  21.1% (19%) for small (10 mm) cancers (n ¼ 31). The same values for large cancers were 5.6  3.7 (4.2), 6.8  4.8 (5.0)

A.A. Zytoon et al.

56.3 (48.3e63.1) 25 (13.5e51.6) 0.7e22 0.7e47

52.6  32 (45.6e59.7) 35.7  46.5 (18e53.4) 0.0093

16e160 48e143

and 17.3  17.1% (20.5%), respectively (Table 5). Statistical analysis of SUVmax1, SUVmax2, and D SUVmax% reveal that all are significant for detection of small breast cancer (p ¼ 0.0002, <0.0001, <0.0001, respectively; Figs 7e8); however, by means of visual assessment, most lesions became more intense on delayed imaging and were detected and diagnosed with greater confidence. Twenty of 31 (64.5%) small cancers with an SUVmax higher than the physiological background uptake could be discriminated well by visual assessment in the early phase (T:B1 ratio > 1.0). Twenty-four cancers (77.4%) were found to have a T:B2 ratio >1 at the delayed phase. Comparing both phases, more cancers (n ¼ 27; 87.1%) were found to have a T:B ratio >0%. Regarding small cancers, regression analysis demonstrates a significant correlation between T:B ratio change and time interval, with a slope of 0.6 and an r of 0.875, 95%CI 0.755e0.938, p < 0.0001.

Dual time point FDG-PET/CT and breast cancer development in dense breast

Values in parentheses are 95% confidence intervals (95%CI). D%, percent change.

0.5e13.2 0.6e31

6.4  4.6 (5.1e7) 6.8  11.6 (2.4e11.2) 0.8499

5.8 (3.8e6.2) 3.4 (1.5e4.5)

Median

3.3 (3e4.3) 2.2 (1.4e3.5) 4.2  2.9 (3.6e4.8) 4.6  7.5 (1.8e7.5) 0.7696 Invasive cancer (n ¼ 82) Non-invasive cancer (n ¼ 29) pevalue

DT:B ratio%

Median Range Mean

Median T:B2 ratio

Mean

Range T:B1 ratio Breast cancer

Tumour:background (T:B) ratio changes over time in invasive, and non-invasive breast cancer Table 4

Mean

Range

1222

The means  SD (median) of the SUVmax1, the SUVmax2, and DSUVmax% for cancers in dense breasts (n ¼ 61) were 4.9  3.7 (4.2), 5.8  4.8 (4.7) and 15.1  17.6% (16%). The same values for physiological background FDG uptake of normal contralateral breast were 1.5  0.4 (1.4), 1.2  0.5 (1.1), and 15.3  18.6% (18.0%), respectively (Table 6). Statistical analysis of the SUVmax1, SUVmax2 and DSUVmax% reveal that all parameters are significant for discrimination of cancers in dense breast tissue (p < 0.0001); however, by means of visual assessment, most cancers became more intense on delayed imaging and were detected and diagnosed with greater confidence. Forty-nine of 61 (80.3%) cancers in dense breasts with an SUVmax higher than the physiological background uptake could be discriminated well by visual assessment in the early phase (T:B1 ratio >1). Fifty-five cancers (90.2%) were found to have a T:B2 ratio >1 at the delayed phase. Comparing both phases, 58 cancers (95.1%) were found to have a T:B ratio >0% (Fig. 9). Regarding cancers in dense breast tissue, regression analysis demonstrates a significant correlation between T:B ratio change and time interval, with a slope of 0.6 and an r of 0.963, 95%CI 0.939e0.978, p < 0.0001.

Discussion The uptake of FDG in breast malignancies increases over several hours; therefore, it would

53e46 80e31.3 20.5 (12.9e28) 20 (25e18.8)

25e44.4 64e0 19 (0e26) 14 (18e0)

5 (4.6e5.8) 1 (0.8e1.1)

0.8e23.8 0.2e3.2

12.9  21.1 (5.2e20.7) 13.4  13.6 (18.4e8.4) <0.0001 17.3  17.1 (13.5e21.1) 17.8  19.2 (22.1e13.5) <0.0001 0.6e8.8 0.5e1.4 1.8 (1.3e2.9) 0.9 (0.9e1.1)

Values in parentheses are 95% confidence intervals (95%CI). D%, percent change.

4.2 (3.6e5.6) 1.2 (1.1e1.3) (4.8e6.5) (1.2e1.4)

1.3e18.5 0.5e2.5

2.7  2.3 (1.8e3.5) 0.9  0.2 (0.8e1) <0.0001 6.8  4.8 (5.7e7.8) 1  0.5 (0.9e1.2) <0.0001 0.6e7 0.7e1.8 1.8 (1.3e2.4) 1.1 (0.9e1.1)

Median

(1.7e3) (1e1.2)

2.3  1.9 1.1  0.3 0.0002 5.6  3.7 1.3  0.4 <0.0001 Tumour  10 mm (n ¼ 31) Normal breast (n ¼ 31) peValue Tumour >10 mm (n ¼ 80) Normal breast (n ¼ 80) peValue

Mean

DSUVmax%

Range Mean

Median SUVmax2

Mean

Range SUVmax1 Breast cancer

Table 5

Maximum standardized uptake value (SUVmax) changes over time in normal breast, large, and small breast cancer

Median

Range

Dual time point FDG-PET/CT imaging

1223

be more valuable to delay imaging beyond the usual 45e60 min after contrast medium injection. In theory, this should lead to improved contrast between the lesion and the background, and improved diagnostic accuracy. This would make dual time point imaging unnecessary. However, although single, delayed imaging may improve sensitivity, the specificity may remain low because of the overall low uptake of breast cancers. We believe that changes in the dual time point SUVs would be a more valuable diagnostic tool than a delayed single time point alone.15 Hustinx et al.21 performed dual time point imaging on 21 patients with 18 head and neck malignant tumours and nine inflammatory or infectious lesions. The authors noted that tumours had a mean SUV increase of 12% between the first and second scan, whereas inflammatory lesions and structures with physiological uptake of FDG (tongue, larynx) showed essentially stable uptake over time or a slight decline. Zhuang et al.19 demonstrated that the FDG activity in cancers increases over time by 19% between the early and delayed scans. The present results agree with these reports. The tumour SUVmax increased over time by 19.1% (p < 0.0001), whereas the physiological uptake of normal breast tissue decreased over time by 16.7% (p < 0.0001) in its mean value. One possible reason for this difference in the pattern of FDG uptake over time is that cancers have an average lower glucose-6-phosphatase activity than normal breast tissue. Although dual time point imaging will prolong the examination, this technique appears to improve the accuracy of the characterization of breast cancers with a low SUV value or cancers in dense breast tissue with a high-uptake background that make the discrimination of cancers difficult, especially if they are small. The non-specific increase in FDG uptake in benign inflammatory processes has been a source of false-positive results for PET. The single time point SUV of 2.5e3.8, cited as the optimal threshold for diagnosing lung cancer,31 may not be applicable to breast cancer because the reported mean SUV is lower in breast cancer cells as a result of less complete phosphorylation of FDG.28 In the current study, 35 of the 111 (31.5%) lesions had an SUVmax1 < 2.5, but only 19 of the 111 (17.1%) lesions had an SUVmax2 < 2.5. If an SUV of 2.5 were considered the threshold for diagnosing malignancy, then many cancers would not be diagnosed on the early phase imaging. Conversely, three of the 111 (2.7%) normal breasts expressed high FDG uptake with an SUVmax1 of 2.5 and none had an SUVmax1 > 2.5. In the present study

1224

A.A. Zytoon et al.

Correlation between Dual Time FDG-PET/CT and Small Breast Cancer 20

P = 0.0002

P<0.0001

P<0.0001

15

SUVmax

10 5 0 -5 -10 -15 -20

SUVmax1n SUVmax1t

SUVmax2n

SUVmax2t

SUVmax%n

SUVmax%t

Figure 7 Comparison of the maximum FDG uptake by small breast cancer against the physiological background FDG uptake at time point 1 (SUVmax1) and time point 2 (SUVmax2), and maximum FDG uptake change over time (DSUVmax%). The difference between the SUVmax values of the breast cancer and the physiological background was significant for the three parameters (p ¼ 0.0002, 0.0001 and <0.0001, respectively). The bar shows the SUVmax mean value while the line shows the SEM. n, Normal breast; t, tumour.

it was shown that the majority of cancer lesions (79.3%) demonstrated a significant increase (p < 0.0001), the majority of normal breast tissues showed a significant decrease (76.6%) in maximum SUV over time (p < 0.0001). This suggests that dual time point imaging will improve the sensitivity of the test because it is expected that normal tissue would not accumulate FDG over an extended period of time. The percentage change over time in the T:B ratios (DT:B% ratio) between ratio 1 and ratio 2 was higher for invasive (mean 52.6, median 56.3) compared with non-invasive cancers (mean; 35.7, median; 25.0), and was higher for large (mean 54.7, median 58.4) compared with small cancers (mean 31.4, median; 25.0). Accordingly, by using this method, despite the low initial SUV of non-invasive and small tumours, the intensity of the contrast ratio was higher on the delayed images, which resulted in improved tumour detectability. The two areas that may benefit from dual time point FDG-PET/CT are tumour grade and size. Noninvasive breast cancer has been previously shown to be poorly detected by FDG-PET.32 The majority of FDG-PET research studies have been performed

Figure 8 A 50-year-old woman with a right breast mass of 7 mm. Dual time point FDG-PET/CT imaging was performed. (a) Axial sections were obtained at the first time point and revealed a focus of very faint FDG activity that could have easily been overlooked on the early PET images (thin arrows). (b) The corresponding images acquired at the second time point (thick arrows). The SUVmax of the right breast cancer in first image set was 0.7, whereas that of second set was 0.9. The percent increase of DSUVmax% was 28.6%. The background FDG uptake SUVmax was 1.1 in first image set, whereas that of second set was 1.0. The percent decrease of DSUVmax% was 9.1%. Therefore, the increase in the T:B ratios between first images set and second image set was 40.6%. Surgical histopathology confirmed invasive ductal carcinoma (mixed papillotubular and scirrhous).

Dual time point FDG-PET/CT imaging

1225

Figure 9 A 32-year-old woman with a left breast mass of 1 cm in diameter originating in dense breast tissue, was examined with dual time point FDG-PET/CT. (a) PET images, coronal sections in the top row were obtained at the first time point. (b) The corresponding axial images in the bottom row were acquired at the second time point. The SUVmax of the lesion in the first image set was 1.6 (thin arrow), whereas that of the second set was 2.6 (thick arrow). The percent increase in SUVmax of the lesion was 62.5%. The increase in the T:B ratio between the first image set and the second image set was 100%, which is consistent with the increasing discriminatory power between the cancer lesion and the background physiological uptake as time passed. Surgical pathology confirmed invasive ductal carcinoma (scirrhous type).

on patients with invasive breast cancer. In a metaanalysis, Wu and Gambhir33 reported that the overall sensitivity, specificity, and accuracy of FDG-PET in the detection of primary invasive breast cancer are 90, 92, and 93%, respectively.

In a large study of single-phase PET for beast tumour imaging, Avril et al.,32 found that the major limitation was a poor detection rate for small (pT1a and pT1b) breast carcinomas. None of four stage pT1a (0.5 cm) cancers and only one of eight

Table 6 Maximum standardized uptake value (SUVmax) changes over time in breast cancer (SUVmaxet) originated in dense breast versus normal dense breast physiological uptake (SUVmaxen) Parameter

Mean

Median

Range

SUVmax1t SUVmax1n T:B1 ratio peValue

4.9  3.7 (3.9e5.8) 1.5  0.4 (1.4e1.6) 3.4  2.7 (2.7e4.1)

4.2 (2.7e5.3) 1.4 (1.3e1.4) 2.5 (1.7e3.5) <0.0001

0.7e18.5 1.1e2.5 0.5e13.2

SUVmax2t SUVmax2n T:B2 ratio peValue

5.8  4.8 (4.6e7) 1.2  0.5 (1.1e1.4) 5.3  4.6 (4.1e6.5)

4.7 (2.7e5.4) 1.1 (1e1.2) 3.8 (2.4e5.4) <0.0001

0.7e23.8 0.5e3.2 0.7e22

DSUVmaxet% DSUVmaxen% DT:B ratio peValue

15.1  17.6 (10.6e19.6) 15.3  18.6 (20.1e10.5) 49.4  36.2 (40.2e58.7)

16 (6.1e25) 18 (20e15.4) 47.3 (38.2e58.4) <0.0001

23e46 64e31.3 25e160

Values in parentheses are 95% confidence intervals (95%CI). t, tumour; n, normal breast.

1226

breast carcinomas at stage pT1b (0.5e1 cm) cancers were identified. The sensitivity for noninvasive breast cancer was low (41.7%). These findings suggest that detection of in-situ carcinomas may not be improved by PET imaging.32 Compared with Avril et al.,32 the present results using dual time imaging show higher sensitivity than single time imaging for the detection of non-invasive cancer. It was possible to discriminate between invasive and non-invasive breast cancers. In a recent study by Mavi et al.,15 the sensitivity of the dual time point imaging method in detecting small invasive breast cancers (4e10 mm), was 82.7%. The slightly higher sensitivity for small cancers achieved by dual time point FDG-PET/CT may be related to the use of PET/CT image fusion, whereas Mavi et al., used dual time point FDGPET without CT fusion. In comparison, the sensitivity of single time point FDG-PET for detecting cancers <10 mm was 57% in another report.32 The present results using dual-time imaging show higher sensitivity and specificity than single-time imaging for the detection of small breast cancers. The DSUVmax% has a sensitivity of 100% for the detection of small breast cancers versus 91.7% for SUVmax2, and 85.7% for SUVmax1. Several potential benefits of dual time point FDG-PET/CT were demonstrated during this study. The technique was sensitive enough to demonstrate the details of the primary breast cancers and capable of depicting small invasive and noninvasive cancers, as well as cancers in dense breast tissue. This technique demonstrated focal areas of abnormal FDG uptake in several lesions that were smaller than 1 cm. All malignancies in the present study manifested as focal areas of increased FDG activity and were identified with the aid of semiquantitative assessment. It has previously been reported that FDG-PET may play an important adjunctive role in selected patients with dense breasts where mammography has a much poorer sensitivity.34 Furthermore a recent study reported that both the maximum and mean SUVs values of dense breasts were found to be significantly higher than those of non-dense breasts.30 Based on the current results, dual time point FDG-PET/CT appears to be advantageous for identifying the most likely region of malignancy within the breast either dense or non-dense. The improved sensitivity may help identify patients with early disease who are candidates for breast-conserving surgery. One limitation of the present study is that, because of the nature of the study, the specificity of the technique in excluding cancer could not be determined and is the subject of further investigation.

A.A. Zytoon et al.

In conclusion, the present study suggests that dual time point FDG-PET/CT imaging is useful for the assessment of primary breast cancer. Single-phase FDG-PET/CT imaging also has a role in breast cancer diagnosis. Dual time point FDG-PET/CT imaging with SUV change over time improved the discrimination between non-invasive and invasive cancers, and had superior sensitivity for the detection of small cancers and cancers in dense breast tissue. An analysis of both time points offers the potential advantage over single time point imaging for the diagnosis of breast cancer in its early stages when it is small or non-invasive. Dual time point FDG-PET/CT imaging is a non-invasive method that can improve the sensitivity of FDG-PET/CT in assessing patients with primary breast cancer.

References 1. Jemal A, Murray T, Samuels A, et al. Cancer statistics. Cancer J Clin 2003;2003(53):5e26. 2. Scheidhauer K, Walter C, Seemann MD. FDG PET and other imaging modalities in the primary diagnosis of suspicious breast lesions. Eur J Nucl Med Mol Imaging 2004;31 (Suppl. 1):S70e9. 3. Rohren EM, Turkington TG, Coleman RE. Clinical applications of PET in oncology. Radiology 2004;231:305e32. 4. Kumar R, Bhargava P, Bozkurt MF, et al. Positron emission tomography imaging in evaluation of cancer patients. Indian J Cancer 2003;40:87e100. 5. Quon A, Gambhir S. FDG-PET and beyond: molecular breast cancer imaging. J Clin Oncol 2005;23:1664e73. 6. Schirrmeister H, Kuhn T, Guhlmann A, et al. Fluorine-18 2-deoxy-2-fluoro-D-glucose PET in the preoperative staging of breast cancer: comparison with the standard staging procedures. Eur J Nucl Med 2001;28:351e8. 7. Rose C, Dose J, Avril N. Positron emission tomography for the diagnosis of breast cancer. Nucl Med Commun 2002; 23:613e8. 8. Lind P, Igerc I, Beyer T, et al. Advantages and limitations of FDG PET in the follow-up of breast cancer. Eur J Nucl Med Mol Imaging 2004;31(Suppl. 1):S125e34. 9. Yap CS, Seltzer MA, Schiepers C, et al. Impact of wholebody 18F-FDG PET on staging and managing patients with breast cancer: the referring physician’s perspective. J Nucl Med 2001;42:1334e7. 10. Mankoff DA. Are we ready for dedicated breast imaging approaches? J Nucl Med 2003;44:594e5. 11. Buck A, Schirrmeister H, Kuhn T, et al. FDG uptake in breast cancer: correlation with biological and clinical prognostic parameters. Eur J Nucl Med Mol Imaging 2002;29:1317e23. 12. Kumar R, Loving VA, Chauhan A, et al. Potential of dual-time-point imaging to improve breast cancer diagnosis with 18F-FDG PET. J Nucl Med 2005;47:1440e6. 13. Adler LP, Crowe JP, Al-Kaisi NK, et al. Evaluation of breast masses and axillary lymph nodes with [F-18] 2-deoxy-2fluoro-D-glucose PET. Radiology 1993;187:743e50. 14. Walter C, Scheidhauer K, Scharl A, et al. Clinical and diagnostic value of preoperative MR mammography and FDG-PET in suspicious breast lesions. Eur Radiol 2003;13:1651e6. 15. Mavi A, Urhan M, Yu JQ, et al. Dual time point 18F-FDG PET imaging detects breast cancer with high sensitivity and

Dual time point FDG-PET/CT imaging

16. 17.

18.

19.

20.

21.

22.

23.

24.

25.

correlates well with histologic subtypes. J Nucl Med 2006; 47:1440e6. Keyes JW. SUV: Standard uptake or silly useless value? J Nucl Med 1995;36:1836e9. Matthies A, Hickeson M, Cuchiara A, et al. Dual time point 18F-FDG PET for the evaluation of pulmonary nodules. J Nucl Med 2002;43:871e5. Boerner AR, Weckesser M, Herzog H, et al. Optimal scan time for fluorine-18 fluorodeoxyglucose positron emission tomography in breast cancer. Eur J Nucl Med 1999;26:226e30. Zhuang H, Pourdehnad M, Lambright ES, et al. Dual time point F-18-FDG PET imaging for differentiating malignant from inflammatory processes. J Nucl Med 2001;42: 1412e7. Kubota K, Itoh M, Ozaki K, et al. Advantage of delayed whole-body FDG-PET imaging for tumour detection. Eur J Nucl Med 2001;28:696e703. Hustinx R, Smith RJ, Benard F, et al. Dual time point fluorine-18 fluorodeoxyglucose positron emission tomography: a potential method to differentiate malignancy from inflammation and normal tissue in the head and neck. Eur J Nucl Med 1999;26:1345e8. Conrad GR, Sinha P. Narrow time-window dual-point F-18-FDG PET for the diagnosis of thoracic malignancy. Nucl Med Commun 2003;24:1129e37. Yen TC, Ng KK, Ma SY, et al. Value of dual-phase 2-fluoro-2deoxy-D-glucose positron emission tomography in cervical cancer. J Clin Oncol 2003;21:3651e8. Demura Y, Tsuchida T, Ishizaki T, et al. 18F-FDG accumulation with PET for differentiation between benign and malignant lesions in the thorax. J Nucl Med 2003;44: 540e8. Nakamoto Y, Higashi T, Sakahara H, et al. Delayed 18F-fluoro-2-deoxy-D-glucose positron emission tomography scan for differentiation between malignant and benign lesions in the pancreas. Cancer 2000;89:2547e54.

1227

26. Ma SY, See LC, Lai CH, et al. Delayed 18F-FDG PET for detection of paraaortic lymph node metastases in cervical cancer patients. J Nucl Med 2003;44:1775e83. 27. Mavi A, Lakhani P, Zhuang H, et al. Fluorodeoxyglucose-PET in characterizing solitary pulmonary nodules, assessing pleural diseases, and the initial staging, restaging, therapy planning, and monitoring response of lung cancer. Radiol Clin North Am 2005;43:1e21. 28. Beaulieu S, Kinahan P, Tseng J, et al. SUV varies with time after injection in 18F-FDG PET of breast cancer: characterization and method to adjust for time differences. J Nucl Med 2003;44:1044e50. 29. Xiu Y, Bhutani C, Dhurairaj T, et al. Dual-time point FDG PET imaging in the evaluation of pulmonary nodules with minimally increased metabolic activity. Clin Nucl Med 2007;32:101e5. 30. Kumar R, Chauhan A, Zhuang H, et al. Standardized uptake values of normal breast tissue with 2-deoxy-2-[F-18]fluoro-D-glucose positron emission tomography: variations with age, breast density, and menopausal status. Mol Imaging Biol 2006;8:355e62. 31. Torizuka T, Zasadny KR, Recker B, et al. Untreated primary lung and breast cancers: correlation between F-18 FDG kinetic rate constants and findings of in vitro studies. Radiology 1998;207:767e74. 32. Avril N, Rose CA, Schelling M, et al. Breast imaging with positron emission tomography and fluorine-18 fluorodeoxyglucose: use and limitations. J Clin Oncol 2000;18: 3495e502. 33. Wu D, Gambhir SS. Positron emission tomography in diagnosis and management of invasive breast cancer: current status and future perspectives. Clin Breast Cancer 2003; 4(Suppl. 1):S55e63. 34. Vranjesevic D, Schiepers C, Silverman DH, et al. Relationship between 18F-FDG uptake and breast density in women with normal breast tissue. J Nucl Med 2003;44: 1238e42.