Utility of 18F-Fluorodeoxyglucose Positron Emission Tomography/Computed Tomography Fusion Imaging for Prediction of Metastasis to Sentinel and Nonsentinel Nodes in Patients with Clinically Node-Negative Breast Cancer

Purpose 18F-Fluorodeoxyglucose positron emission tomography/computed tomography fusion imaging (18F-FDG PET/CT) is an important diagnostic tool in breast cancer. The utility of maximum standardized uptake values (SUVmax) of primary tumors has been evaluated to predict sentinel node (SN) and non-SN metastasis in clinically node-negative (cN0) patients. Patients and Methods 18F-FDG PET/CT was performed on 414 cN0 patients. The following parameters were evaluated: SUVmax at 60 min (SUVmax1), SUVmax at 120 min (SUVmax2), percent change between SUVmax1 and SUVmax2 (ΔSUVmax%), SN metastasis foci maximum size (SN meta size), and ratio of metastatic SNs to total SNs or SN ratio (SNR). It was assessed whether these were risk factors for SN metastasis. The relationship between these parameters and the status of SN and/or non-SN metastasis was retrospectively explored to predict non-SN metastasis. Results All SUV parameters significantly correlated with pathological T factor (pT), nuclear grade, lymphatic invasion (Ly), and Ki-67 labeling index. On multivariate analysis, pT and Ly were independent predictive factors for SN metastasis. In SN meta-positive cases, SN meta size, SNR, and ΔSUVmax% were predictors for non-SN metastasis on univariate analyses, and the former two were independent predictors on multivariate analysis. The combination of SUVmax2 and ΔSUVmax% was an independent predictor of non-SN metastasis (P = 0.0312) and was associated with prediction of non-SN metastasis negative status with high probability (92.3%). Conclusions In patients with cN0 breast cancer, SUV parameters of the primary tumor were correlated with pathological features. The combination of SUVmax2 and ΔSUVmax% may be useful for predicting non-SN metastasis. Electronic supplementary material The online version of this article (10.1245/s10434-020-08269-0) contains supplementary material, which is available to authorized users.

studies have reported the correlation between the 18 F-FDG uptake value of primary tumors and their histological and biological features such as tumor size, nuclear grade (NG), Ki-67 labeling index (LI), and prognosis. [4][5][6][7] Usually, 18 F-FDG uptake is measured with the maximum standardized uptake value (SUVmax) 60 min after its injection, but some articles report the utility of SUVmax levels both at 60 min and 120 min after injection (SUVmax1 and SUV-max2, respectively). [8][9][10] The percentage change between SUVmax1 and SUVmax2 (DSUVmax%) in the primary tumor was also easily measured. However, the utility of dual time point (DTP) measurement has not yet been established for primary tumors.
Sentinel node biopsy (SNB) is a standard technique for patients with clinically node-negative (cN0) breast cancer, 11 and axillary lymph node dissection (ALND) may be considered when macrometastasis is observed in a SN. Staging of axillary lymph node (ALN) was evaluated by physical examination and ultrasound. Nonetheless, reports of up to 30% of SN metastasis have been found in cN0 patients, 12 and in this population, the frequency of metastasis to non-SN resected by ALND was reported to be around 40%. 13 As the result of the American College of Surgeons Oncology Group Z0011 trial, 14 axillary dissection in clinically node-negative individuals has come to be less common, and ALND has come to be optional for the patients who had SN-metastasis positive in two or less nodes, underwent breast-conserving surgery, and received whole-breast irradiation with adjuvant systemic therapy.
On the other hand, several nomograms were developed to predict metastasis to SN and non-SN from clinicopathological parameters, including properties of the primary tumor. 15,16 The validity of these nomograms was also reported in Japanese patients. 17 Therefore, the biological properties of the primary tumor, detected with 18 F-FDG PET/CT, are expected to help predicting SN and/or non-SN metastasis in cN0 patients.
The aim of this study is to investigate whether the prediction of SN and non-SN metastasis is possible by the examination of SUV parameters in the primary tumor.

Patient Population
This study was approved by the institutional review board of the National Defense Medical College. Informed consents were obtained from all patients with regard to 18 F-FDG-PET/CT examination and entry into this study. From September 2005 to December 2017, 18 F-FDG-PET/CT was performed for 820 consecutive preoperative patients who received histological diagnosis of primary breast carcinoma. Of these, 406 patients were excluded from the study because of (1) preoperative medication therapy (n = 123), (2) ductal carcinoma in situ (DCIS) (n = 23), (3) distant metastasis (n = 5), (4) SN not being identified by SNB (n = 20), (5) ALND without SNB (n = 180), (6) difficulty measuring SUVmax (n = 111), (7) acquisition of one time point with 18 F-FDG PET/CT (n = 7), and/or (8) diabetes mellitus (n = 47). These eight factors frequently overlapped. There were four cases of SN metastasis negative and non-SN metastasis positive, but these four had received preoperative medication therapy and were excluded from the study. For the 123 patients who received preoperative medication therapy, the medication was only aromatase inhibitors (AI) in 13 patients (10.6%), AI followed by tamoxifen in 1 patient (0.8%), only chemotherapy in 84 patients (68.3%), chemotherapy followed by AI in 9 patients (7.3%), chemotherapy combined antihuman epidermal growth factor receptor 2 (HER2) therapy in 14 patients (11.4%), and chemotherapy combined antiHER2 therapy followed by AI in 2 patients (1.6%). Ultimately, 414 cN0 patients were eligible for the study.
Additionally, 56 cN0 patients with SN macrometastasis and ALND were eligible (Fig. 1). ALNDs were performed for 5 of the 21 patients with micrometastasis and 56 of the 63 patients with macrometastasis. In the five patients with micrometastasis, the decisions of ALNDs were made by the surgeons during the surgery. Among the seven patients who had macrometastasis but did not receive ALNDs, three refused ALND, but the details of other four patients were unknown.
Altogether, the 414 patients had no clinical evidence of ALN metastasis by physical examination and image findings, e.g., mammography, ultrasound examination, and 18 F-FDG PET/CT. When the axillary node status was equivocal in a patient, fine needle aspiration cytology was performed, and the case was judged cN0 if cytological examination was negative. In all these cases, the histological diagnosis of breast cancer was made by core needle biopsy before surgery. After these examinations, 18 F-FDG PET/CT was performed prior to surgery, and the interval between core needle biopsy and surgery was 42 days on average. administration of 3.7 Mbq/kg 18 F-FDG. The first scan was of whole-body images from head to thigh, and the second scan was chest only within 50-60 min after first examination.
After image reconstruction, regions of interest (ROI) were placed in one area of the primary breast lesion showing the highest 18 F-FDG uptake. The SUV was calculated using decay-corrected tissue activity divided by the injected dose per patient body as represented by this formula SUV ¼ activity in ROI MBq/ml ð Þ = injected dose MBq/kg body weight ð Þ : The SUVmax1 and SUVmax2 were obtained at dual time points: the SUVmax at the early (60 min) and delayed (120 min) phase, respectively. The DSUVmax% was calculated using the formula

Pathological Evaluation of SN
From September 2005 to March 2008, SNs were in principle identified using radioactive tin colloid alone (82 cases, 19.8%). After April 2008, SNs were in principle identified using both radioactive tin colloid and blue dye in all cases (332 cases, 80.2%). In the former era, positive rates of metastasis were 17.1% (14/82) on the patient basis and 9.8% (16/164 nodes) on the SN basis. In the latter era, positive rates of metastasis were 21.1% (70/332) on the patient basis and 15.2% (89/586 nodes) on the SN basis.
For intraoperative frozen section diagnosis, each SN was sliced into 2-mm-thick pieces, cut into 5-10-lm-thick sections, fixed with formalin for a short time, and stained with hematoxylin and eosin. Tumor macrometastasis, micrometastasis, and isolated tumor cells were defined in accordance with the Union for International Cancer Control (UICC) eighth edition. For the tumor deposit size, the diameter of the largest metastatic deposit in the frozen or paraffin-embedded permanent section (maximum SN metastasis size, SN meta size) was measured. SNR was defined as numerical ratio of metastasis-positive SNs (macro-and micrometastasis) to all resected SNs.

Histological Study
Two observers (H.T. and Y.Y.) performed pathological diagnosis. Pathological tumor size was defined as the largest diameter of a tumor including both invasive and noninvasive components, and pathological invasive tumor size was defined as the largest diameter of the invasive component of a tumor. NG was given according to the General Rules for Clinical and Pathological Recording of Breast Cancer, seventeenth edition. 18 Estrogen receptor (ER) and progesterone receptor (PgR) were assessed by immunohistochemistry and defined as positive if 1% or higher of constituent carcinoma cells were immunoreactive. 19  or higher of constituent carcinoma cells were immunoreactive. 22 Pathological T (pT) and N (pN) factors and stage were determined by the clinical and pathological recording of breast cancer by UICC eighth edition.

Memorial Sloan Kettering Cancer Center (MSKCC) Nomogram
The MSKCC nomograms were available on the MSKCC web site (http://www.mskcc.org/nomogram). 15,23 The nomogram for SN metastasis required nine factors, including primary tumor features such as tumor size, grade, and lymphovascular involvement. 15 The nomogram for non-SN metastasis required nine factors, including primary tumor features and SN status. 23 According to the sum of points for each factor, the probabilities of SN metastasis and non-SN metastasis were calculated for each patient. The values of the probabilities were compared using the nonparametric Wilcoxon test.

Statistical Analysis
Statistical analyses were performed using JMP Ò 13 (SAS Institute Inc.; Cary, NC). The correlations between SUVmax parameters (SUVmax1, SUVmax2, and DSUV-max%) and clinicopathological factors were evaluated using the nonparametric Wilcoxon test and the Kruskal-Wallis test. Receiver operating characteristic (ROC) curves were drawn to find the optimal cutoff value of SUVmax parameters for the prediction of SN. ROC curves were also drawn to find the optimal cutoff values of SUVmax parameters, the number of SN metastasis, SNR, and SN meta size for the prediction of non-SN metastasis. The Youden index [= sensitivity -(1 -specificity) of each cutoff value] was calculated, and the highest value was taken as the optimal cutoff point. All statistical analyses were two-sided with significance defined as a P value of \ 0.05.

Comparison Between SN Metastasis-Positive and Metastasis-Negative Groups
The number of patients with SN metastasis, including macrometastasis and micrometastasis, was 84 (20.3%) (Fig. 1). All patients were classified into either SNmetastasis positive (group A) or SN-metastasis negative (group B). Clinicopathological factors were compared between the two, and results are presented in Table 2.

Optimal Cutoff Values of SUVmax Parameters for Prediction of SN Metastasis
The optimal cutoff values of SUVmax1, SUVmax2, and DSUVmax% for the prediction of SN metastasis were 3.

Univariate and Multivariate Analyses for Predictor of SN Metastasis
By univariate and multivariate logistic analyses in comparing pre-and postoperative factors, the odds ratios for SN metastasis were found to be significantly higher in the cT3 group than in the cT1/2 group, higher in the pT3 group than in the pT1/2 group, higher in the ER-positive group than in the ER-negative group, and higher in Lypositive group than in Ly-negative group (Table 3). PgR was univariately significant but excluded from the multivariate analyses because of its collinearity with ER. Although SUVmax1 (C 3.4 versus \ 3.4), SUVmax2 (C 3.0 versus \ 3.0), and DSUVmax% (C 2.5 versus \ 2.5) were also correlated with the risk of SN metastasis in the univariate analyses, these factors were not significant in the multivariate analyses (Table 3A, B). SUVmax1, SUVmax2, and DSUVmax% were correlated with each other. Additionally, multivariate analyses were conducted incorporating parameters that are available preoperatively, i.e., cT, ER, and SUVmax parameters, which revealed that cT, ER, and SUVmax2 were significant in one of the multivariate analyses (Table 3C).

Comparison Between Non-SN Metastasis-Positive and Metastasis-Negative Groups
The clinicopathological factors of these groups (group C and D) are presented in Table 4. There were significant differences in the mean pathological invasive tumor size (40.6 mm ± 23.3 SD versus 26.6 mm ± 19.1 SD, P = 0.0106), mean SN meta size (8.7 mm ± 4.2 SD versus 5.7 mm ± 3.0 SD, P = 0.0080), SN meta size (C 6.0 mm versus \ 6.0 mm, P = 0.0111), SNR (C 0.67 versus \ 0.67, P = 0.0131), and DSUVmax% (C 20.0 versus \ 20.0, P = 0.0458). Although there was no significant difference in SUVmax1 and SUVmax2 between these two groups, they tended to be higher in group C than in group D.

Univariate and Multivariate Analyses for Prediction of Non-SN Metastasis
On univariate analyses, SN meta size, SNR, and DSUVmax% were statistically significant factors for the prediction of non-SN metastasis (Table 5). On multivariate analysis, SN meta size and SNR were independent predictive factors of metastasis to non-SN in patients with SN metastasis (Table 5). DSUVmax% was nearly significant as a predictive factor (odds ratio 3.60, 95% CI 0.95-13.6, P = 0.0586).

Combination of SUVmax2 and DSUVmax% for Prediction of Non-SN Metastasis
There were 13 patients with low SUVmax2 (\ 3.0) and low DSUVmax% (\ 20.0). Of these, 12 were patients without non-SN metastasis (92.3%) ( Table 6). The sensitivity, specificity, positive predictive value, negative predictive value (NPV), and accuracy of their combination for non-SN metastasis were 94.7%, 32.4%, 41.9%, 92.3%, and 53.6%, respectively. In predicting non-SN metastasis, the combination of SUVmax2 and DSUVmax% showed higher sensitivity and NPV than the SUVmax1, SUVmax2, DSUVmax% and the combination of SUVmax1 and DSUVmax%. By univariate and multivariate logistic analyses, the combination of SUVmax2 and DSUVmax% was an independent predictive factor of metastasis to non-SN in patients with SN macrometastasis (P = 0.0470, 0.0312, respectively) ( Table 7). However, the combination of SUVmax1 and DSUVmax% did not show any significant difference on univariate analysis. The combination of SUVmax2 and DSUVmax% was a useful predictor of metastasis to non-SN.

DISCUSSION
In the present work, the significance of SUV parameters in primary tumor for SN and/or non-SN metastasis was evaluated in patients with cN0 breast cancer. SUV parameters were found to be effective predictors of SN metastasis in cN0 patients, and these parameters could help anticipate metastasis of non-SN in SN-positive patients. Furthermore, in cN0 and SN metastasis-positive patients with low SUVmax2 and low DSUVmax%, the negative status of non-SN could be predicted with high probability (92.3%) by using a combination of SUVmax2 and DSUVmax% values.
Several clinicopathological factors have been described as predictors of SN metastasis in breast cancer. 13 These factors include tumor size, lymphovascular invasion, HER2, ER, multifocality, age, and tumor grade. Furthermore, several studies identified clinicopathological predictors of non-SN metastasis such as primary tumor size, lymphovascular invasion, and SN status. [24][25][26] Several nomograms have been developed to predict SN and non-SN metastasis: the MSKCC nomogram of prediction of metastasis to SN and non-SN utilizes primary tumor features such as tumor size, tumor grade, and lymphovascular invasion. 15,16 In the present cohort, the MSKCC nomograms were confirmed to be useful for prediction of SN and non-SN metastasis. In these nomograms, some pathological parameters of a primary tumor can only be obtained from detailed postoperative pathological examination. Because SUVmax of the primary tumor was correlated with these pathological parameters and was able to be acquired before surgical examination, the measurement of the SUVmax may potentially be of clinical benefit.
In the present study, pT, ER, and Ly were independent predictors of SN metastasis, and tumor invasion size was significantly different between non-SN-metastasis-positive and non-SN-metastasis-negative groups. ER-positive cN0 cases showed significantly higher odds ratio of SN metastasis than ER-negative cN0 cases. Although this result appeared paradoxical, it is in agreement with findings of other large-scale studies. 15 18 F-FDG PET/CT was performed using the DTP method, and it was confirmed that the combination of SUVmax2 and DSUVmax% was useful to predict non-SN metastasis using preoperative features. This combination was also superior in sensitivity (94.7%) and NPV (92.3%) to SUVmax1, SUVmax2, and DSUVmax% alone, and to the combination of SUVmax1 and DSUVmax%. These results might support the idea that the combination of SUVmax and DSUVmax% represents a more biological characteristic of the tumor.
Given the findings of ACOSOG Z0011 and AMAROS trials, in cN0 and SN-positive patients, axillary radiotherapy could be chosen instead of ALND if further axillary SD standard deviation, SN sentinel node, SN meta size maximum sentinel node metastasis size, SN ratio number of metastasis-positive SNs/ number of all resected SNs, SUVmax maximum standardized uptake value, SUVmax1 SUVmax at 60 min, SUVmax2 SUVmax at 120 min, DSUVmax% (SUVmax2 -SUVmax1)/SUVmax1 9 100  NPV negative predictive value, PPV positive predictive value, SN sentinel node, SUVmax maximum standardized uptake value, SUVmax1 SUVmax at 60 min, SUVmax2 SUVmax at 120 min, DSUVmax% (SUVmax2 -SUVmax1)/SUVmax1 9 100 treatment is needed. 14,27 At present, the findings of additional non-SN may not be clinically useful. However, which subset of cN0 and SN-positive patients require axillary treatment is not fully clarified. If SUVmax parameters of PET/CT scans were shown to accurately predict SN and non-SN statuses, these results would open the way to further research to find the optimal axillary management in cN0 and SN-positive patients. PET/CT scans are not performed routinely, and the SUVmax parameters are not introduced clinically for many patients. If SUVmax parameters are shown to be excellent for prediction of SN and/or non-SN metastasis and their utility is widely accepted, this method may be included as one preoperative diagnostic tool in the future.
The cost of 18 F-FDG PET/CT is higher than the total cost of whole-body examinations, including MRI, bone scintigraphy, and abdominal ultrasonography, but 18 F-FDG PET/CT appears to have superiority in that the diagnosis of both local and systemic status of a disease is possible in only 3 h. Therefore, the PET/CT was considered to be more convenient to the patients than the combination of other whole-body examinations.
Limitations of this study include its retrospective nature and that it was conducted in a single facility with a relatively small number of patients. Another prospective multicenter trial is needed to confirm the effectiveness of SUVmax and DSUVmax% in the prediction of SN and non-SN metastasis.
In conclusion, SUVmax of the primary tumor was a predictive factor of SN and/or non-SN metastasis in patients with cN0 breast cancer. Furthermore, it was possible to estimate non-SN metastasis negativity with a high probability by combining SUVmax2 and DSUVmax%. From these results arises the possibility of minimizing unnecessary ALND.
FUNDING Data extraction and data analysis were supported in part by JSPS KAKENHI Grant Number JP 18K07340. JSPS did not influence the study design, the data collection, the data analysis, the data interpretation, or the writing of the manuscript.
DISCLOSURES The authors declare no conflicts of interest.
OPEN ACCESS This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article's Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article's Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons. org/licenses/by/4.0/. CI confidence interval, SN sentinel node, SN meta size maximum sentinel node metastasis size, SN ratio number of metastasis-positive SNs/ number of all resected SNs, SUVmax maximum standardized uptake value, SUVmax1 SUVmax at 60 min, SUVmax2 SUVmax at 120 min, DSUVmax% (SUVmax2 -SUVmax1)/SUVmax1 9 100