We selected a total of 196 newly diagnosed breast cancer patients who underwent ¹⁸F-FDG PET/CT inspection in our hospital from 2017 to 2020, excluded 116 cases, and at last included 80 cases for analysis. The inclusion criteria and exclusion criteria were as follows:

Inclusion criteria: (1) Newly diagnosed breast cancer. (2) Never received treatment. (3) Female or male. (4) Had a ¹⁸F-FDG PET/CT inspection. (5) Single lesion.

Exclusion criteria: (1) Bilateral breast cancer. (2) Diffuse lesions. (3) The maximum diameter of the lesion was less than 1 cm [NCCN guidelines and many other studies mentioned that the sensitivity and specificity of ¹⁸F-FDG PET/CT inspection are not high enough when faced with lesions with a maximum diameter of less than 1 cm (3, 4, 6, 8)]. (4) No biopsy. (5) Metastases cannot be diagnosed without follow-up. (6) Had other malignant tumors simultaneously.

The patient should rest 1 day in advance, avoid strenuous exercise to prevent muscle intake, fast for 4 h in advance, and forgo sugary fluids. On the examination day, the patient should drink 500 ml of water before the injection to dilute the ¹⁸F-FDG concentration in the urinary system, and keep warm to prevent brown fat from FDG avidity. The patient’s blood glucose level before examination needs to be kept <200 mg/dl to prevent the influence on ¹⁸F-FDG intake. Injection should be performed on the contralateral upper arm of the patient’s affected side to prevent the influence of human factors on the axillary lymph node metastasis of the affected side. Before the examination, the patient should urinate to excrete the higher concentration of FDG accumulated in the bladder. Imaging started 60 min after injection, and the range was generally from the skull to the mid-thighs of both sides.

Instrument model: Germany Siemens Biograph m CT·S. CT scan parameters: CT tube pressure 120 kv, tube current: 300MA, layer thickness 3–4 mm, matrix 512×512, radiopharmaceutical ¹⁸F-FDG, generated by GE800trace accelerator, imaging agent quality control radiochemical purity >95%. ¹⁸F-FDG was injected intravenously at 5.55 MBq/kg body weight. After the drug was injected, the patient should close their eyes and rest quietly for 40–60 min. PET/CT scan was performed after urination. Scan time: 15 min. The obtained images were transferred to the Medix workstation for reading and analysis after attenuation correction processing.

Tumor tissues resected intraoperatively were formalin-fixed and paraffin-embedded in time, cut into 3 μm sections, and baked at 68°C for 2 h. The tissue sections were stained immunohistochemically for ER, PR, and Her-2, respectively, using automated immunostainers (BOND-MAX; Leica Microsystems, Tokyo, Japan). The tissue sections were treated with biotinylated anti-rabbit secondary antibody, followed by further incubation with streptavidin-horseradish peroxidase complex. They were then stained with diaminobenzidine (DAB), and the sections were counterstained with hematoxylin.

All the suspected or unsuspected extraaxillary nodal metastases and distant metastasis needed to be confirmed by pathological results. In addition, the lesion, such as small distant bone metastases, that cannot be confirmed by pathological results needed to be confirmed by more than two imaging methods during follow-up at least 3 months, and the stage finalized by the clinician and the corresponding treatment plan were recorded. The case follow-up was completed by a nuclear medicine physician and a non-nuclear medicine physician. When there was a disagreement, the opinion of the pathologist would be discussed and decided together.

To compare continuous variables between groups, the statistical significance of differences was determined using Student’s t test or non-parametric test, as appropriate. Categorical variables were compared using the chi-square test. Simple correlation analysis and logistic regression analysis of the association between variables and N1, M1, and C1 after adjustment of potential confounders were performed. Finally, ROC (receiver operating characteristic) curve was used to test diagnostic power and determine the best cutoff value. The statistical software was SPSS 25. Statistical significance was defined as a two-tailed probability of less than 0.05.

Among all the 80 patients included in this research, there were three false positives (two related to lymph node metastasis were included in group N0 and C0 [sensitivity 93.55%], two related to distant metastasis was included in M0 and C0 [sensitivity 92.86%], there was one patient who overlapped) and three false negatives (two related to lymph node metastasis was included in group N1and C0 [specificity 96.23%], another one related to bone metastasis was included in group M1 and C0 [specificity 98.18%]). The total accuracy of ¹⁸F-FDG PET/CT inspection in this research was 93.02%.

The breast cancer patients with extraaxillary lymph node metastasis (N1) had lower proportion of the expression of ER (%) and PR (%) in tumor cells than N0 group (ER: 2 [0–80] vs. 80 [15–95]; PR: 1 [0–10] vs. 20 [0–45], p<0.001). Patients in N1group had higher diameter (mm, 40 [28–58] vs. 29 [21–42], p<0.001) than N0 group, as did initial stage (IIA [0, 0%], IIB [7, 24.14%], IIIA [13, 44.83%], IIIB [5, 17.24%], IIIC [2, 6.9%], IV [2, 6.9%] vs. IIA [11, 21.57%], IIB [24, 47.06%], IIIA [11, 21.57%], IIIB [3, 5.88%], IIIC [1, 1.96%], IV [1, 1.96%]). The time from the discovery of the abnormality, such as lump, pain, and so on, to the first visit to the doctor in N1 group was larger than that in N0 group (day, 150 [30–240] vs. 120 [15–365], p<0.001). Also, breast cancer patients with N1 had lower proportion of ER and PR by the box plots than that with N0 as shown in Figure 2A . Her-2 status, age, gender, histology, and grade were similar between N0 and N1 group in Table 1 and Supplementary Table 1 (p>0.05).

The distribution of the proportion of ER and PR between M1 and M0 group had no significant difference as shown in Figure 2B , did nor did Her-2 status, gender, and grade in Supplementary Table 1 . But age (year, 50 [41.75–56.75] vs. 49 [41.5–56], p<0.001), time (day, 183 [60–365] vs. 90 [15–183], p<0.001), diameter (mm, 38 [25.25–61] vs. 32 [22–42], p<0.001), and initial stage (IIA [2, 7.69%], IIB [9, 34.62%], IIIA [7, 26.92%], IIIB [5, 19.23%], IIIC [0, 0%], IV [3, 11.54%] vs. IIA [9, 16.67%], IIB [22, 40.74%], IIIA [17, 31.48%], IIIB [3, 5.56%], IIIC [3, 5.56%], IV [0, 0%]) in M1 group was higher than in M0. Furthermore, patients with the histological type of infiltrating ductal carcinoma had less distant metastasis (47 [83.93%] in M0 vs. 16 [66.67%] in M1, p<0.001), while patients with infiltrating lobular carcinoma had more distant metastasis (0 [0%] in M0 vs. 4 [16.67%] in M1, p<0.001) in Supplementary Table 1 .

The breast cancer patients with unsuspected metastasis found and treatment plan changed (C1) had lower proportion of the expression of ER (%) and PR (%) in tumor cells than the C0 group (ER: 10 [0–85] vs. 80 [15–90], p=0.026; PR: 1 [0–10] vs. 20 [0–70], p=0.041), as presented in Figure 2C . While patients in the C1 group had higher time (day, 183 [30–365] vs. 120 [15–365], p<0.001) and diameter (mm, 42 [28.25–57.25] vs. 27 [20.75–36.25], p<0.001) than in the C0 group. C1 group had more patients with IIIA, IIIB, IIIC initial stage than the C0 group (IIIA [15, 40.54%], IIIB [6, 16.22%], IIIC [2, 5.41%] vs. IIIA [9, 20.93%], IIIB [2, 4.65%], IIIC [1, 2.33%], p<0.001), while had fewer patients with IIA, IIB, IV initial stage than C0 group (IIA [2, 5.41%], IIB [12, 32.43%], IV [0, 0%] vs. IIA [9, 20.93%], IIB [19, 44.19%], IV [3, 6.98%]). Her-2 status, age, gender, histology, and grade were similar between C0 and C1 group in Table 1 and Supplementary Table 1 (p>0.05).

Furthermore, the association between the receptors and extraaxillary lymph node metastasis or distant metastasis was analyzed by simple correlation. Data showed that the proportions of ER (r=−0.339, p=0.002) and PR (r=−0.247, p=0.011) were negatively correlated with the occurrence of extraaxillary lymph node metastasis (suspected and unsuspected), while diameter (r=0.3, p=0.008) and initial stage (r=0.455, p<0.001) were positively correlated with it, as presented in Table 2A . Similarly, the proportions of ER (r=−0.258, p=0.021) and PR (r=−0.217, p=0.054) were negatively correlated with the finding of unsuspected metastasis and a change in treatment plan. Diameter (r=0.399, p<0.001) and initial stage (r=0.262, p=0.019) were positively correlated with it, as presented in Table 2C . However, the proportion of ER (r=0.001, p=0.992) and PR (r=0.088, p=0.437) had no obvious correlation with distant metastasis, scilicet, no obvious correlation was found between the factors and the occurrence of distant metastasis, as shown in Table 2B .

Next, logistic regression was performed to analyze the correlation between the proportion of ER, PR, and the occurrence of extraaxillary lymph node metastasis, as in Table 3A . After adjustment of age, time, diameter, and initial stage, we found that the proportions of ER (OR=0.980, 95% CI of OR [0.964–0.996], p=0.013) and PR (OR=0.944, 95% CI of OR [0.909–0.980], p=0.003) were still negatively correlated with the occurrence of extraaxillary lymph node metastasis detected by ¹⁸F-FDG PET/CT, as did the time (OR=0.994, 95% CI of OR [0.990–0.998], p=0.002). And the positive correlation between age (OR=1.073, 95% CI of OR [1.003–1.149], p=0.042) and initial stage (OR=4.983, 95% CI of OR [2.152–11.537], p<0.001) between N1 also was found, as presented in Table 3A . The proportion of ER (OR=0.998, 95% CI of OR [0.973–1.024], p=0.890) and PR (OR=1.000, 95% CI of OR [0.972–1.029], p=0.988) had no obvious correlation with distant metastasis, nor did age and time. The maximum diameter of primary lesion (OR=1.037, 95% CI of OR [1.008–1.068], p=0.013) and initial stage (OR=1.617, 95% CI of OR [1.012–2.585], p=0.044) still had a positive correlation with M1, as shown in Table 3B . At last, after adjustment of the traditional factors, we found ER (OR=0.986, 95% CI of OR [0.972–0.999], p=0.039) had a negative correlation with the finding of unsuspected metastasis and treatment plan change. And the diameter (OR=1.058, 95% CI of OR [1.018–1.100], p=0.004) had a positive correlation with C1, as presented in Table 3C .

Finally, the receiver operating characteristic (ROC) curve was used to evaluate predictive values of the proportion of ER for N1 and C1 by ¹⁸F-FDG PET/CT inspection. Figure 3 showed the area under the curve (AUC) for the relationships between the proportion of ER and N1 (AUC =0.704, 95% CI [0.584–0.824], p=0.003)and C1 (AUC=0.647, 95% CI [0.525–0.770], p=0.024). Youden index was calculated to obtain the best cutoff value, and all of N1 and C1 groups were 25% (the proportion of ER).

There were 37 patients in C1 group. They had extraaxillary lymph node metastasis or distant metastasis lesions, which were ambiguous or neglected by conventional imaging (ultrasound, CT, and/or MRI) and clinical examination (lymph node biopsy) but identified by ¹⁸F-FDG-PET. The detailed sites of lesions that led to changing the treatment plan are shown in Table 4 , p= 0.037<0.005 (chi-square test). There were 15 (40.54%) patients with clavicle area and/or inner mammary area lymph node metastases, 12 (32.43%) of which with the proportion of ER ≤ 25% (ER−), 3 (8.11%) with the proportion of ER>25% (ER+). Three (8.11%) patients had liver metastasis, two (5.41%) patients with ER−, and one (2.7%) patient with ER+. One (2.7%) patient had lung metastasis and with ER−. Seven (18.92%) patients had bone metastasis, one (2.7%) patient with ER−, and six (16.22%) patients with ER+. Eleven (29.72%) patients had multiple metastases, in which one (2.7%) patient had lung and bone metastases with ER−, two (5.41%) patients had lung, clavicle area, and/or inner mammary area lymph node metastases with ER−, eight (21.62%) patients had bone, clavicle area, and/or inner mammary area lymph node metastases (five [13.51%] patients with ER− and three [8.11%] patients with ER+).

Figure 4 shows an example of a 48-year-old newly diagnosed breast cancer woman detected by ¹⁸F-FDG PET/CT inspection. Unsuspected extraaxillary lymph node metastasis in the left internal mammary near the breastbone was found by ¹⁸F-FDG PET/CT, which was neglected by MR. The ER proportion of this patient was 1% (less than 25%). This patient originally planned to undergo surgical treatment, but after the lesion was discovered by ¹⁸F-FDG PET/CT, the treatment plan was updated; this patient would take neoadjuvant chemotherapy first and then surgery. This example indicated that when the proportion of ER was less than 25%, ¹⁸F-FDG PET/CT might help newly diagnosed breast cancer patients find unsuspected extraaxillary lymph node metastasis and change the initial treatment plan.

Figure 5 shows another example of a 63-year-old newly diagnosed breast cancer patient detected by ¹⁸F-FDG PET/CT inspection. There was no metastasis found except the primary lesion in initial diagnosis, but ¹⁸F-FDG PET/CT inspection was performed at the request of the patient, then unsuspected distant metastases in lung, hilar lymph node, and bone were detected, confirmed by needle biopsy, which were neglected by ultrasound and MRI. The treatment plan was updated to chemotherapy and targeted drug therapy. The ER proportion of this patient was 10% (less than 25%). This example indicated that although no axillary lymph node metastasis was found in the initial diagnosis, it is still possible to find unsuspected metastases by FDG PET/CT inspection when the ER was less than 25%.