Abstract
Aim: The aim of this study was to prove the diagnostic value of interim 18F-Fluorodeoxyglucose-positron-emission tomography combined with computed tomography (PET/CT) scan for predicting pathological complete response (pCR) compared to other factors in neoadjuvant chemotheraphy. Patients and Methods: Twenty-seven patients with breast cancer were included in this retrospective study. They all underwent scheduled neoadjuvant chemotherapy. Patients underwent PET/CT at baseline, mid-point (interim), and preoperatively (after completion of chemotherapy). The metabolic response was calculated as follows: ΔStandardized uptake value (SUV)(%)=(1st SUVmax–2nd SUVmax)/1st SUVmax×100. Results: The change in SUVmax between baseline and interim PET/CT scans was significantly larger than between interim and preoperative PET/CT scan. An optimal cut-off ΔSUV value of 78.3% was proposed for discriminating patients with pCR from those without pCR. Metabolic CR, defined as a change of SUVmax greater than the cut-off value, can predict pCR according to univariate analysis (p=0.012; Relative risk (RR)=25.3). Furthermore, metabolic CR was the most powerful factor for predicting pCR than other possible factors according to multivariate analysis (p=0.003). Conclusion: It is possible to use interim 18F-FDG PET-CT as an effective method to predict early response in patients with breast cancer treated with neoadjuvant chemotherapy.
Neoadjuvant chemotherapy has been regarded as an effective way to treat patients with locally advanced breast cancer to reduce tumor volume and enhance the opportunity for breast-conserving surgery (1, 2). Pathological complete response (pCR) to neoadjuvant chemotherapy has been proven to be a significant prognostic factor for disease-free and overall survival (3-5). That is to say, pCR following neoadjuvant chemotherapy improves the prognosis of patients with breast cancer. Previous studies have shown that 13%-26% of patients show pCR after completion of neoadjuvant chemotherapy (1, 6). The Nottingham histological grading system is the most widely used method to predict prognosis of those patients (7-9). Therefore, it is thought that early prediction of pathological response in neoadjuvant chemotherapy may provide an early opportunity to change the treatment plan in case of ineffectiveness. It is also possible to avoid unnecessary side-effects from ineffective chemotherapy, such as nausea, alopecia, hematological toxicity, cardiotoxicity, or neurotoxicity (10).
Positron-emission tomography combined with computed tomography (PET/CT) using 18F-fluorodeoxyglucose (FDG) is widely used in patients with malignant cancer. It can be used in detection of the malignant lesion, finding the metastatic lesion, staging the tumor, and monitoring the response to therapeutic approaches. In addition, 18F-FDG PET/CT has been playing a major role for the early prediction of response to neoadjuvant chemotherapy in many types of malignant cancers such as esophageal, rectal and lung cancer and some types of aggressive lymphomas (11-14).
18F-FDG PET/CT is regarded as one of the essential imaging modalities for evaluation of breast cancer in patients (15, 16). Several studies have reported a correlation between early changes in 18F-FDG uptake after one or two cycles of neoadjuvant chemotherapy with the pathological response in patients with breast cancer (10, 17-18).
The aim of this study was i) to assess the feasibility of interim 18F-FDG PET/CT scan for early response evaluation, ii) to propose an optimal cut-off value of ΔSUV(%) for predicting pCR, and iii) to justify the effectiveness of an optimal cut-off value for predicting pCR compared to other possible factors.
Patients and Methods
Patients. Twenty-seven patients [mean (±SD) age=50±9 years] with newly diagnosed, non-inflammatory, large or locally advanced breast cancer, were included in this study, retrospectively (four patients with stage IIA, 21 patients with stage IIIA and two patients with IIIC). The study population and the characteristics of the 27 patients are shown in Tables I and II. Initial core needle biopsy was performed in all patients. One patient had invasive lobular carcinoma and the others had invasive ductal carcinoma subtype. They then underwent 4, 6 or 8 cycles of neoadjuvant chemotherapy. PET/CT scan was taken at baseline (before initiating neoadjuvant chemotherapy), and after the 2nd, 3rd or 4th cycle of chemotherapy (interim). Additionally, among the 27 patients, 19 patients also underwent preoperative PET/CT scan after completion of neoadjuvant chemotherapy. Breast surgery was performed for all patients and final pathological reports were also presented. This study was approved by the Hospital Institutional Review Board (AN 13022-002).
Neoadjuvant chemotherapy. Three different regimens were used for chemotherapy in this patient series. Twenty patients (74%) received six cycles of docetaxel/epirubicin (75/75 mg/m2 of body surface area). Six patients (22%) received an initial four cycles of doxorubicin/cyclophosphamide (60/600 mg/ m2 of body surface area) and followed by four cycles of docetaxel (75 mg/m2 of body surface area). One patient (4%) received four cycles of doxorubicin/cyclophsphamide (60/600 mg/m2 of body surface area). Chemotherapy was repeated every three weeks.
Response to chemotherapy. All specimens were confirmed by histopathological analysis after breast surgery. Pathological response was classified into two groups: pCR and non-pCR. pCR was defined as no invasive and no in situ residuals in breast and regional lymph nodes. Pathological grades were assessed as grade 1 to 3 according to the Nottingham histological grade (11-13). In addition, biological subgroups were defined as using hormonal receptor and Ki-67 status (luminal A type: Estrogen receptor positive (ER+)/Human epidermal growth factor receptor 2-negative (HER2−) and Ki-67 expression <20%; luminal B type: ER+HER2− and Ki-67 expression ≥20%, HER2 type: ER−PR− and HER2+; triple-negative type: ER−PR− and HER2−).
PET/CT imaging. Images were obtained with PET-CT scanner (Gemini TF, Philips Medical Systems, Cleveland, OH, USA). All patients fasted for at least six hours and serum glucose level was less than 180 mg/dl before scanning. Forty-five to sixty minutes after intravenous injection of 370 to 480 MBq (10 to13 mCi) 18F-FDG, CT scans were obtained followed by PET emission scans for one minute per bed. The PET unit had an axial field of view of 18 cm and a spatial resolution of 4.4 mm. A low-dose CT scan was obtained for attenuation correction and for localization, with a 16-slice multidetector helical CT unit, using the following parameters: 120 kVp; 50 mA; 0.75-s rotation time; 0.75-mm slice collimation; 4-mm scan reconstruction, with a reconstruction index of 4 mm; 60-cm field of view; and 512×512 matrix. PET data were reconstructed iteratively using a 3-dimensional row action maximum likelihood algorithmwith low-dose CT datasets for attenuation correction. Maximum intensity projection and cross sectional views and fusion images were generated and reviewed.
PET/CT image analysis. Two experienced nuclear physicians evaluated the PET/CT images. Malignant breast lesions were classified as positive if there was focally increased 18F-FDG uptake, compared with the uptake in surrounding normal soft-tissue. A region of interest (ROI) was targeted on each malignant breast lesion by manual adjustment. The maximum standardized uptake value (SUV) was calculated as follows:
SUV=Mean activity (ROI) (MBq/ml)/injected dose (MBq)/total body weight (g)
From these SUVs from targeted ROIs, the maximum standardized uptake values (SUVmax) were acquired for analysis. The metabolic response after the interim PET/CT was calculated as follows: ΔSUV(%)=(baseline SUVmax – interim SUVmax)/baseline SUVmax ×100 (%)
Statistical analysis. The Mann-Whitney U-test, receiver-operating characteristic (ROC), logistic regression analysis, and multivariate regression anlysis were used as statistical methods. A value of p<0.05 was defined as being statisticaIly significant. SPSS 17.0 (SPSS inc, Chicago, IL, USA) and Medcalc software (Medcalc Software, Mariakerke, Belgium) were used for data analysis.
Results
A total of 27 lesions were identified on baseline PET/CT scan in 27 patients. SUVmax of the lesion in baseline and interim PET/CT scan was 8.6±3.8 (mean±SD) and 2.7±1.9, respectively (p<0.001) (Figure 1). ΔSUV(%) was 66.9±14.9%.
Interim and preoperative PET/CT scans. In the 19 patients that underwent baseline, interim, and preoperative PET/CT scans, SUVmax of the lesion in baseline, interim, and preoperative PET/CT scan was 8.7±4.1, 2.6±2.0, and 1.9±1.2 respectively. The SUVmax of the baseline was significantly higher than interim and preoperative PET/CT scan (p<0.001). There was no significant statistical difference between the SUVmax of the interim and preoperative PET/CT scan (p=0.07) (Figure 2a). ΔSUV(%) between baseline and interim and between interim and preoperative PET/CT scan was 67.8±15.0% and 22.6±14.4%, respectively (p<0.001) (Figure 2b).
pCR group vs. non-pCR group. Among the 27 patients, five were confirmed as having pCR, but 22 patients had residual invasive cancer (non-pCR). SUVmax of pCR and non-pCR groups in the baseline PET/CT scan were 8.9±5.1 and 8.6±3.5, respectively (p=0.74) (Figure 3a). SUVmax of pCR and non-pCR groups in the interim PET/CT scan were 1.6±0.3 and 3.0±2.0, respectively (p=0.03) (Figure 3b). ΔSUV(%) between baseline and interim PET/CT scan of pCR and non-pCR groups were 75.8±15.9% and 64.9±14.3%, respectively (p=0.04) (Figure 3c).
Determination of ΔSUV cut-off value to discriminate the pCR group from non-pCR group. ROC analyses were performed to determine the optimal cut-off value of ΔSUV(%) to differentiate pCR from non-pCR patients. The ROC curve is presented in Figure 4. A cut-off ΔSUV(%) of 78.3% was found to identify those patients with pCR. The area under the ROC curve (AUC) was 0.8 [standard error=0.1; 95% confidence interval (CI)=0.6-0.9]. The sensitivity and specificity were 80.0% and 90.9%, respectively
Metabolic CR and pCR. We defined metabolic CR (complete response) as a change of SUVmax greater than the cut-off value. Univariate analysis was performed on the pCR-related factors. As shown in Table III, metabolic CR significantly predicted the pCR through univariate analysis (p=0.012; relative risk (RR)=25.3; 95% CI=2.1-310.8). Furthermore, according to multivariate analysis, metabolic CR showed superior predictability of the pCR to other known parameters such as HER2 type and Ki-67 status (p=0.003 vs. p=0.171 and 0.131, respectively).
Possible variables and metabolic CR. Possible variables that may have an effect on the metabolic CR were assessed by univariate analysis through separate logistic regression analysis. The variables included age, clinical stage, tumor grade, receptor status of ER, PR, HER2, Ki-67 expression status, and biological subgroups mentioned above. According to the logistic regression analysis, luminal B type group had a significant possibility of presenting metabolic CR (p=0.049; RR=5.427; 95% CI=1.007-29.255) (Table IV). As shown in Table IV, those in the triple-negative type group might also have a possibility of presenting metabolic CR. The p-value was of marginal significance (p=0.061; RR=0.111; 95% CI=0.011-1.106).
Discussion
Systemic neoadjuvant chemotherapy is increasingly being used nowadays and has been proven useful in patients with locally advanced breast cancer (19, 20). The main purpose of the study was to evaluate early changes caused by neoadjuvant chemotherapy in malignant tumor FDG uptake that have highly predictive value for the pathological response in patients with breast cancer.
Therefore, firstly, we assessed the feasibility of interim PET/CT scan for early response evaluation. Secondly, we attempted to propose an optimal cut-off value for predicting pCR. Thirdly, we tried to justify the effectiveness of the optimal cut-off value for predicting pCR compared to other possible factors.
As shown in Figure 2, the change in the SUVmax was greater between baseline and interim PET/CT than between interim and preoperative PET/CT scan (p<0.001). There was no significant statistical difference between the SUVmax of the interim and preoperative PET/CT scan (p=0.07). From these observations, we could expect that the therapeutic effect of neoadjuvant chemotherapy was early with interim PET/CT and the metabolic change was maintained until preoperative PET/CT scans. Therefore, if the chemotherapy regimen was not effective, it is possible to give an early insight using interim PET/CT to enable the treatment plan to be modified and to avoid adverse side-effects. It is feasible to use interim PET/CT scan for early assessment of response to neoadjuvant chemotherapy.
This retrospective study demonstrated that patients with pCR can be distinguished by interim 18F-FDG PET/CT during the interim neoadjuvant chemotherapy. The pCR and non-pCR goups had similar mean SUVmax in baseline PET/CT scan. However, the pCR group presented significantly lower mean SUVmax than the non-pCR group on the interim PET/CT scan (p=0.03). Furthermore, the pCR group had a significantly larger change in SUVmax than did the non-pCR group (p=0.04). An optimal cut-off ΔSUV value of 78.3% was proposed for discriminating pCR patients (change of SUVmax greater than cut-off value) from non-pCR patients in ROC analysis.
Similar to our results, several studies have suggested a cut-off value of ΔSUV in neoadjuvant chemotherapy for discriminating pCR from non-pCR patients, acquiring values ranging from 40 to 88% (17, 18, 21-23). Thus, differentiation of the pCR from non-pCR group using interim PET/CT scan may be possible. However, the wide range of cut-off values limits application in clinical practice. Several factors can contribute to the wide range of cut-off values (24). Firstly, the timing of PET/CT evaluation is very variable. Many groups performed PET/CT after one or two cycles of neoadjuvant chemotherapy (21, 22). McDermott et al. took PET/CT at midpoint and end of neoadjuvant chemotherapy (23). Secondly, breast carcinomas consist of different subtypes depending on hormonal receptors, such as ER+ tumors, HER2 overexpression, and triple-negative tumors. Thus, heterogenous characteristics of tumor biology can cause differences in response to neoadjuvant chemotherapy.
Using a cut-off ΔSUV(%) of 78.3%, patients were classified according to metabolic CR (change of SUVmax greater than the cut-off value). According to univariate analysis, metabolic CR significantly predicted the pCR. As shown in Table III, in predicting pCR, relative risks of Ki-67 status and HER2 type were quite high but the p-values were not significant. These factors are well-known for predicting pCR on receiving neoadjuvant treatment (25-28). Considering these factors, the present study demonstrated that metabolic CR was a strong or predictor for pCR than other variables.
Regarding metabolic CR, as shown in Table IV, luminal B type was significantly associated with metabolic CR than in patients non-luminal B type. Luminal B type had been known to be more responsive to chemotherapy than luminal A type (28). Luminal B type was also regarded as more proliferative than luminal A type (29). Therefore, these factors might explain the association of metabolic CR and luminal B type.
Another impressive finding was the triple-negative type group might also have a greater possibility of achieving metabolic CR (p=0.061) than the non triple-negative type. Patients with triple-negative breast cancer are known to have better responsiveness to neoadjuvant chemotherapy than those with ER+ tumor (30). In other words, metabolic CR could be interpreted as a good response to neoadjuvant chemotherapy.
Although a limitation of this study is the small number of patients, it clearly identifies the feasibility of interim PET/CT scan for early response evaluation and presents an optimal cut-off ΔSUV value to predict pCR. Metabolic CR is proven to be a powerful predictor of pCR.
Conclusion
In patients with breast cancer treated with neoadjuvant chemotherapy, the change in 18F-FDG uptake at midpoint (interim) of chemotherapy provides valuable information of therapeutic response in early time. An optimal cut-off ΔSUV value of 78.3% was proposed for discriminating patients with pCR from those non-pCR. Using this cut-off value, metabolic CR in interim PET-CT showed better predictability for pCR than other possible factors. It is possible to use interim 18F-FDG PET/CT as a valuable method for predicting early response of neoadjuvant chemotherapy. This may be helpful for establishing individual treatment strategies for patients with breast cancer.
Acknowledgements
This study was partly supported by Korea University Clinical Research Grants (2011-K1132941).
Footnotes
-
Conflicts of Interest
The Authors have no conflicts of interests.
- Received May 8, 2014.
- Revision received June 16, 2014.
- Accepted June 17, 2014.
- Copyright© 2014 International Institute of Anticancer Research (Dr. John G. Delinassios), All rights reserved