Percentage change of primary tumor on 18F-FDG PET/CT as a prognostic factor for invasive ductal breast cancer with axillary lymph node metastasis: Comparison with MRI.

We evaluated the prognostic value of quantitative parameters using dual time point (DTP) F-FDG PET/CT (PET/CT) in invasive ductal breast cancer (IDC) with metastatic axillary lymph nodes (ALN) as compared with dynamic contrast-enhanced (DCE) and diffusion-weighted (DW) MRI.Seventy patients with IDC and metastatic ALN were retrospectively registered. Static PET parameters including maximum standardized uptake value (SUVmax), metabolic tumor volume (MTV), total lesion glycolysis (TLG) of primary tumor, SUVmax of ALN (SUVALN), and percentage changes (Δ%) in those parameters were measured with DTP PET/CT. From DCE MRI, peak enhancement value, total tumor angio volume, and proportions of kinetic curve types on delayed-phases were investigated. The average apparent diffusion coefficient (ADCavg) was estimated on DWI. To demonstrate the prognostic value of quantitative imaging parameters for recurrence-free survival (RFS), univariate and multivariate analyses were performed using those parameters and clinicohistologic variables.All static PET parameters, %ΔSUVmax, %ΔMTV, and %ΔSUVALN on DTP PET/CT and ADCavg on DWI were significantly predictive for disease recurrence. Of clinicohistologic variables, pathologic tumor (pT) diameter, pathologic ALN stage, tumor grade, and hormonal status also were significantly prognostic. After multivariate analysis, %ΔSUVmax > 25.05 (P = .043), ADCavg ≤ 1016.55 (P = .020), pT diameter > 3 cm (P = .001), and ER negative status (P = .002) were independent prognostic factors for poor outcome.Only %ΔSUVmax of the primary tumor on PET/CT together with ADCavg, pT diameter, and ER status was an independent prognostic factor for predicting relapse in IDC with metastatic ALN. Percentage change of primary tumor on preoperative PET/CT may be a valuable imaging marker for selecting IDC patients that require adjunct treatment to prevent relapse.

Introduction

18F-fluorodeoxyglucose (FDG) positron emission tomography/computed tomography (PET/CT) has been used to detect enhanced cancer cell glycolysis and identify malignant lesions, stage work-up, determine disease recurrence, and monitor treatment response in various oncologic settings, including breast cancer. FDG is transported by glucose transporters, metabolized by the enzyme hexokinase and builds up in cancer cells. FDG uptake within cells mainly depends on cellular metabolic activity and the number of glucose transporters. However, this mechanism is not specific to cancer cells. FDG can accumulate in inflammatory cells and benign processes, causing false-positive findings.[ Recently, several studies have been conducted to overcome these limitations using dual time-point (DTP) PET/CT.[ It is widely assumed that continuous increasing FDG uptake over time after intravenous administration indicates malignancy, whereas a decreasing or stable tendency for FDG affinity indicates inflammation or benign processes. This assumption can also be applied to DTP PET/CT, not only to improve diagnostic performance, but also determine the degree of aggressiveness in breast cancer.

Another imaging modality, magnetic resonance imaging (MRI) is being increasingly performed for breast cancer diagnosis. Dynamic contrast enhanced MRI (DCE-MRI) can increase accuracy for differentiating breast cancer from normal tissue by assessing morphologic characteristics and quantitative kinetic profiles of breast lesions.[ Diffusion-weighted imaging (DWI) can measure the mobility of water molecules in biologic tissues and represent the microscopic cellular environment, and provide additional information such as biophysical features complementary to DCE-MRI.[ These advanced MRI techniques play clinically important roles in the identification of malignant lesions and determination of disease prognosis and treatment response.[

There are many clinicohistologic prognostic factors in breast cancer patients. Among these factors, axillary lymph node (ALN) status is highly associated with disease prognosis and has been demonstrated to be the most significant prognostic factor in previous studies.[ The next logical step is identification of prognostic factors in ALN-positive breast cancer to identify high-risk subpopulations that could benefit from adapted treatment and follow-up plans. To the best of our knowledge, no previous studies have investigated the prognostic value of imaging parameters using DTP PET/CT, DCE-MRI, and DWI, simultaneously.

The purpose of this study was to demonstrate prognostic factors by comparing each imaging parameter derived from DTP PET/CT, DCE-MRI, and DWI and clinicohistologic variables for recurrence-free survival (RFS) in invasive ductal breast carcinoma (IDC) with metastatic ALN.

Methods

Patients

A retrospective review of our hospital database between November 2010 and April 2014 identified 230 eligible patients who were diagnosed with IDC and underwent preoperative DTP PET/CT, DCE-MRI, and DWI for initial staging workup. The mean time interval between DTP PET/CT and the 2 MRI modalities was 0.2 ± 1.7 days. From the eligible patients, we excluded those who had a previous history of any other malignancy, neoadjuvant chemotherapy or radiotherapy before those imaging modalities, follow-up less than 6 months, or bilateral breast cancer. We also excluded cases that had either pathologic tumor diameters smaller than 1 cm based on the full-width-at-half-maximum of PET[ or were negative for pathologic ALN metastasis. Ultimately, we enrolled 70 patients (mean age 50.5 ± 10.0 years; range, 31–77 years) pathologically confirmed to have metastatic ALN.

This clinical information review was approved by the institutional review board of our hospital and the requirement for informed patient consent was waived because of the retrospective design.

Treatment and clinical follow-up

In all patients, breast conserving surgery (BCS) or modified radical mastectomy (MRM) with axillary lymph node dissection (ALND) was done according to tumor size and location, results of the sentinel lymph node biopsy, and the corresponding physician's discretion. Systemic chemotherapy was prescribed with a combined regimen containing doxorubicin and cyclophosphamide, adding either paclitaxel or docetaxel, followed by radiotherapy when the necessity arose. After that, hormonal therapy was given to patients depending on individual age and hormonal status. Patients with human epidermal growth factor receptor 2 (HER2) overexpressing breast cancer were treated with trastuzumab with one or more chemotherapy regimens.

During the follow-up period, patients were asked to visit our hospital for regular examinations every 3 to 6 months in the first 2 years and every 6 to 12 months after that. When recurrence was suspected based on clinical symptoms, physical examination and/or imaging studies, additional procedures for pathologic confirmation or relevant imaging modalities (e.g., chest CT, abdomen CT, brain MRI, etc) were performed. RFS was measured from the date of initial diagnosis to the date of the objective result suggesting disease recurrence.

Clinicohistologic analysis

Clinicohistologic variables, including age, pathologic tumor (pT) diameter, pathologic ALN stage, Black's nuclear grade, and Bloom-Richardson's histologic grade were analyzed. Immunohistochemistry was also evaluated by a direct immunoperoxidase method with antibodies against estrogen receptor (ER), progesterone receptor (PR), and HER2 according to the Allred system and American Society of Clinical Oncology/College of American Pathologists guidelines.[ ER and PR positivity were defined using both scoring parameters of the percentage of positive cells (from 0 to 5) and staining intensity (from 0 to 3). A total score of 0–2 was considered negative for each hormone; 3–8 was considered positive. For HER2 positivity, tumor cells with more than 10% staining were regarded as positive, and those having less than 10% were regarded as negative. For the marker Ki67, overexpression was defined as more than 14% of tumor cells in this study.[

Dual time point 18F-FDG PET/CT acquisition protocol

All patients fasted at least 6 hours before intravenous injection of 5.18 MBq/kg (0.14 mci/kg), and fasting serum glucose levels <140 mg/dL were required to proceed. Patients were asked to rest for 1 hour before image acquisition using a Siemens Biograph mCT with 128-slice CT (Siemens Medical Solutions, Knoxville, TN). Before PET scanning, a low-dose CT scan was obtained without contrast enhancement from the skull base to the proximal thigh in the supine position for attenuation correction. To obtain PET image parameters, an emission scan of 5 to 7 bed positions with 2 minutes per bed position and a maximum spatial resolution of 2.0 mm were also acquired. PET images were reconstructed with 200 × 200 matrices with an ordered-subset expectation maximum iterative reconstruction algorithm (21 subsets, 2 iterations), a Gaussian filter of 5.0 mm, and a slice thickness of 3.0 mm. Approximately 2 hours after administration of 18F-FDG (mean, 56.8 ± 5.8 minutes; range, 48–71 minutes), delayed-phase imaging was obtained only in the chest region in the same supine position as the earlier scan. Patients were asked to remain in a resting position on the bed to minimize FDG uptake by muscles during both early- and delayed-phase PET/CT acquisition.

PET/CT image analysis

PET/CT images were interpreted by two experienced nuclear physicians (JY and BSK with 5 years and 13 years of experience, respectively) who were unaware of the clinical information. To assess semiquantitative PET/CT parameters, regions of interest (ROI) were placed over the primary breast tumor and ALN using the automated delineation feature in commercial software (Syngo.via, Siemens Medical Solutions). The maximal standardized uptake value (SUVmax), which represents the highest FDG avidity within the ROI of a tumor lesion, was corrected with the individual patient body weight. Metabolic tumor volume (MTV) was measured based on spherical-shaped delineation of a three-dimensional volume of interest (VOI) covering the primary tumor with an isocontour threshold of 40% of the SUVmax, as reported previously.[ Total lesion glycolysis (TLG) was estimated by multiplying MTV by average SUV within the VOIs of the tumor lesion. ROIs were drawn over the area of maximal FDG avidity of ALN to measure the SUVmax of ALN (SUVALN).

These semiquantitative parameters were also obtained from delayed-phase images using the same techniques as with early-phase images. The percentage change of each parameter over interval time was calculated by subtracting the early measurement from the delayed and dividing by the early, as follows: Percentage change (%Δ) = (delayed parameter – early parameter)/early parameter × 100.

DCE-MRI and DWI acquisition protocol

MRI was performed with a 3.0-Tesla (T) Achieva system (Philips Medical System, Best, the Netherlands) using a bilateral dedicated seven-channel SENSE breast coil in the prone position. An axial, fat-suppressed, fast-echo T2-weighted image was acquired in advance (TR/TE, 5521/70; flip angle, 90 degrees; field of view, 320 mm; matrix, 332/261; slice thickness, 3 mm with no gap; bandwidth, 289 Hz; time acquisition, 4 minutes 23 seconds). Thereafter, axial DWI with single-shot echo planar imaging (b = 0 and 1000 s/mm2; TR/TE, 6984/78; field of view, 360 mm; matrix, 128/128; number of excitations, 2; slice thickness 4 mm with a 1-mm slice gap; acquisition time, 1 minutes 30 seconds) and pre- and postcontrast axial and sagittal three-dimensional T1-weighted gradient echo with fat saturation sequences were obtained. DWI was obtained along each of the x-, y-, and z-axes. Gadolinium (0.1 mmol/kg body weight of Gd-DTPA; Gadovist, Bayer Schering Pharma AG, Berlin, Germany) as a bolus was injected intravenously at a velocity of 2 mL/s, followed by 25-cc saline flush using an automatic injector. After administration of contrast media, 6 phases of dynamic enhanced images were acquired at 55.4 seconds (axial), 110.8 seconds (axial), 146 seconds (sagittal), 221.6 seconds (axial), 292 s (sagittal), and 438 seconds (axial), respectively. Dynamic axial and sagittal MR parameters were as follows: TR/TE, 4.42/2.17; flip angle, 12 degrees; field of view, 320 mm; matrix, 320/320; receiver bandwidth, 621.4 Hz/pixel; and slice thickness, 1 mm and 4.37/2.15; 12 degrees; 25 cm; 250/250; 704.5 Hz/pixel; 1 mm, respectively. Subtraction images were obtained by subtracting the pre-contrast images from the series of 6 postcontrast images on a pixel-by-pixel basis.

DCE-MRI and DWI analysis

Precontrast and DCE-MRIs were transferred to a commercially available MR computer-aided diagnosis workstation (CAD STREAM version 5.2.8.591, Confirma, Kirkland, WA) and were retrospectively reviewed. The kinetic analyses for breast lesions were reviewed by a radiologist (JC with 10 years of experience in breast imaging) blinded to clinical information and final histopathologic results. A color overlay map was placed on all enhancing lesions at a 50% enhancement threshold level in a pixel-by-pixel comparison across precontrast, first postcontrast, and last postcontrast series. The delayed enhancement type after the first postcontrast series was categorized as persistent, plateau, or washout pattern according to the changes of signal intensity after reaching the highest value.

Angio volume (cc) of the primary breast lesion, which represents total enhancing lesion volume, was measured automatically using CAD software. The quantitative parameters from kinetic curves were investigated including the proportion (%) of washout and plateau patterns. The percentage of peak enhancement in breast tumor lesions, which was defined as the peak value of the signal intensity, was also recorded using the following formula: (peak of signal intensity − signal intensity of precontrast phase)/signal intensity of precontrast phase × 100.

All DWIs were analyzed to measure the apparent diffusion coefficient (ADC) value (×10−6 mm2/s) by the same radiologist, who manually drew the ROI with care to avoid cystic and necrotic portions of the tumor and normal breast parenchyma. The ADC value was estimated using the following equation: −(1/b)ln(S0/S1), where b is the diffusion factor, S1 is the attenuated signal (b value of 1000 s/mm2) and S0 is the full spin-echo signal without diffusion gradient (b value of 0 s/mm2). Three ROIs were initially drawn on the most representative image corresponding to tumor location and size, after that the average ADC (ADCavg) value was used for further analysis. Consensus was achieved in all patients.

Statistical analysis

All data are presented as mean ± standard deviation (SD) and the 95% confidence interval (CI). Statistical analysis was performed using the MedCalc software package (Ver. 9.5, MedCalc Software, Mariakerke, Belgium). Receiver operating characteristic (ROC) analysis was performed to estimate optimal cutoff values for all continuous variables for the prediction of disease recurrence. The Kaplan–Meier method with a log-rank test was assessed in univariate analyses of RFS using optimal cutoff values. Prognostic factors that were statistically significant in the Kaplan–Meier analysis were included in the multivariate analysis using Cox proportional hazard modeling to demonstrate independent prognostic factors for RFS. Statistical significance was defined as a P < .05.

Results

Patient characteristics and clinicohistologic features

Table 1 shows the patient characteristics. Disease recurrence occurred in 18 patients during the follow-up period (mean period 36.1 ± 13.6 months; range, 6.1–70.4 months). Of these patients, 5 showed locoregional recurrence, 4 experienced pulmonary metastasis, 3 had axillary lymph nodal metastasis, and 2 each had bone, brain, and liver metastases, respectively.

Table 1

Patient characteristics.

Table 2 includes the univariate analyses of the correlation between clinicohistologic features and RFS. Patients with pT diameter greater than 3 cm (P < .001) and higher pN stage (P < .001) showed poorer prognosis compared to those without. Higher nuclear and histologic grade was also significantly associated with poor prognosis (P = .006 and .006). Negative ER (P = .005), negative PR (P = .044), and positive Her-2 (P = .049) status were predictive of poor prognosis, but age (P = .955) and Ki67 overexpression (P = .189) were not. Figure 1 contains Kaplan-Meier curves according to clinicohistologic features.

Table 2

Results of univariate analyses of clinicopathologic variables.

Figure 1

Kaplan–Meier curves of RFS for clinicohistologic variables: (A) Age, (B) pT diameter, (C) pN stage, (D) nuclear grade, (E) histologic grade, (F) ER status, (G) PR status, (H) Her-2 status, and (I) Ki67 overexpression. ER = estrogen receptor, pN = pathologic lymph node, RFS = recurrence-free survival.

Imaging parameters and recurrence-free survival

Table 3 lists the results of univariate analyses of DTP PET/CT parameters and RFS according to the optimal cutoff values from ROC curves. Patients with higher values of SUVmax (P = .012), MTV (P < .001), TLG (P < .001), SUVALN (P < .001), %ΔSUVmax (P < .001), and %ΔSUVALN (P = .008) had significantly lower RFS than those with lower values for those parameters. In addition, patients with a lower %ΔMTV value (P = .002) had a poor prognosis compared with those with a high value. There was no significant difference in RFS according to a cutoff value of %ΔTLG (P = .431). Kaplan-Meier curves for each group are illustrated in Fig. 2.

Table 3

Results of univariate analysis of DTP PET/CT parameters.

Figure 2

Kaplan–Meier curves of RFS for subgroups according to DTP PET/CT parameters: (A) SUVmax, (B) MTV, (C) TLG, (D) SUVALN, (E) %ΔSUVmax, (F) %ΔMTV, (G) %ΔTLG, and (H) %ΔSUVALN. DTP = dual time point, ER = estrogen receptor, MTV = metabolic tumor volume, PET/CT = positron emission tomography/computed tomography, RFS = recurrence-free survival, SUV = standardized uptake value, TLG = total lesion glycolysis.

Univariate analyses of DCE-MRI and DWI parameters used to predict RFS are listed in Table 4. Kaplan–Meier curves revealed that only ADCavg was significantly correlated with RFS (P = .041). Angio volume, proportion of washout pattern, proportion of plateau pattern, and peak enhancement were not associated with RFS (P = .105, .1460, .098, and .161, respectively). Kaplan–Meier curves of each group are illustrated in Fig. 3.

Table 4

Results of univariate analysis of DCE-MRI and DWI parameters.

Figure 3

Kaplan-Meier curves of RFS for subgroups according to DCE-MRI and DWI parameters: (A) Angio volume (cc), (B) proportion of washout (%), (C) proportion of plateau (%), (D) peak enhancement (%), and (E) ADCavg (×10−6 mm2/s). ADC = apparent diffusion coefficient, DCE-MRI = dynamic contrast enhanced magnetic resonance imaging, DWI = diffusion-weighted imaging, ER = estrogen receptor, RFS = recurrence-free survival.

Multivariate analysis

The multivariate regression analysis (Table 5), %ΔSUVmax (P = .043) showed independent prognostic value along with ADCavg (P = .020), ER status (P = .002), and pT diameter (P = .001). Figure 4 depicts a representative case.

Table 5

Results of multivariate analysis for RFS.

Figure 4

A 56-year-old patient with IDC in the upper outer quadrant of the right breast and metastatic ALN: %ΔSUVmax, 25.25, ADCavg, 925.36 × 10−6 mm2/s, pT diameter, 3.2 cm, negative ER: (A) early-phase fusion PET/CT, (B) delayed-phase fusion PET/CT, (C) DCE MRI with the kinetic curve, and (D) ADC map showing ROI for measuring the ADCavg. The patient experienced recurrence in the right chest wall after 17.8 months. ADC = apparent diffusion coefficient, ALN = axillary lymph node, ER = estrogen receptor, IDC = invasive ductal breast cancer, ROI = region of interest, SUV = standardized uptake value.

Discussion

ALN status is the most significant prognostic factor and is considered in the application of adjuvant therapy after surgical treatment.[ Postoperative adjuvant therapy can help eradicate cancer cells that might already have spread at the time of diagnosis. Even though adjuvant therapy improves survival, many patients still remain at risk of disease recurrence.[ Therefore, we proposed that it would be clinically effective to identify breast cancer patients with metastatic ALN at higher risk of disease recurrence based on quantitative imaging parameters during preoperative evaluation. Although DCE-MRI and DWI are valuable imaging techniques for improving the sensitivity and specificity of malignant breast lesion detection, the prognostic significance of parameters derived from them compared with DTP PET/CT parameters has not been evaluated.

Our major finding was that only the percentage change in SUVmax during interval image acquisition time was an independent prognostic factor, along with ADCavg, pT diameter, and ER status for RFS in IDC with metastatic ALN. According to the results of our multivariate analysis, both pT diameter and ER status were stronger prognostic markers than the 2 imaging parameters. Given that these clinicohistologic variables are obtained from surgical specimens, they might be expected to have greater prognostic value. However, in light of the ability to evaluate prognosis using imaging parameters in a preoperative assessment, our results will have clinical implications. Our current findings are consistent with previously reported results from Garcia et al,[ who suggested that the 18F-FDG retention index showed prognostic value, and Choi et al,[ who demonstrated that both SUVmax at a single time point and ADC values could be useful for predicting IDC prognosis. However, these prior studies did not include DTP PET parameters for survival analysis.

SUVmax of breast cancer is a representative value of glucose metabolism that reflects cancer cell glycolysis and metabolic activity. Based on this hypothesis, increasing SUVmax during the interval time would indicate the cancer cell aggressiveness. Univariate survival analyses revealed that both SUVmax and %ΔSUVmax of the primary tumor had statistical value, whereas the multivariate analysis demonstrated that only %ΔSUVmax was an independent prognostic factor. SUVmax only represents the highest radiotracer concentration at a single time point, and so the percentage change in FDG accumulation would be expected to reflect metabolic characteristics of breast cancer better than a static parameter.

The quantitative DCE-MRI parameters from the CAD program that predominantly depend on microvascular permeability and blood flow rate did not provide prognostic information for risk stratification in this study. However, our results differ from those of Pickles et al,[ who asked whether DCE-MRI vascular kinetic variables and MRI-based volume had a prognostic value in locally advanced breast cancer patients. This dissimilarity may be related to the different methods of imaging analysis to obtain tumor size, vascular kinetic parameters, texture, and shape-based metrics.

This study has several limitations. First, the exclusive focus on ALN-positive patients might limit the generalizability of our results. Second, given the limited resolution of PET (e.g., partial volume effect), the SUVmax of either small primary tumor lesions (e.g., smaller than 2 cm) or ALN could be underestimated.[ However, the partial volume effect should be negligible in this study because there were no patients with disease recurrence in the T1c group and only 1 patient with an SUVALN lower than 1 had disease recurrence during follow-up. Third, we did not analyze the morphologic characteristics of breast lesions such as tumor shape, tumor margin or contrast-enhancement patterns using DCE-MRI, which may be associated with histopathologic grade.[ Therefore, further prospective studies that complement our approach and evaluate the specific criteria of morphologic characteristics to determine prognosis are necessary in the future.

Despite these limitations, the current study demonstrated that quantitative imaging parameters using DTP PET/CT and DWI had independent prognostic significance for patients with ALN-positive IDC. As breast cancer patients with metastatic ALN have poorer prognoses than those without, further classification of this patient group according to objective prognostic factors during preoperative clinical assessment would be valuable for optimizing treatment approach. The prognostic value of those imaging parameters can be evaluated to provide evidence for risk stratification, guiding the treatment plan for individual patients and ultimately improving prognosis.