This study focused on the reliability of evaluating FIs with a chemical probe to detect differences between breast tissue features. For this purpose, we applied gGlu-HMRG, a chemical probe that can detect GGT enzyme activity via green fluorescence. Previously, we reported the detection of breast lesions and lymph node metastasis using this probe[22, 23]. We established a procedure to obtain quantitatively reliable FI using a newly developed dedicated apparatus for fluorescence measurements. Unexpectedly, the fluorescence measurements using the previous method were not reproducible. In this study, we found that soaking the tissue entirely in the solution yielded better results than spraying or dripping the tissues. Using this method, a sufficient amount of probe molecules was provided to the cancer cells. In addition, rinsing with PBS improved the FIs of cancer tissues, probably because this step can remove blood, fat, and other materials that sometimes cover the cell surface (Fig. 1). In addition, rinsing with such a calcium-free solution may loosen cell–cell adhesion via cadherin molecules, and thus, facilitate probe binding with GGT, which is expressed on the cell surface[21].
First, we developed a standardized protocol and applied it in a multicenter study. The temperature was predicted to have the greatest influence on the FI measurements. However, we did not need to precisely control the temperature when measurements were taken in the range of room temperature (15–30°C). Correlation between the temperature and 5 min FI was weak and not significant (P > 0.05, Table 1), indicating that the variations in GGT activities between cells and tissues were more significant than the effect of temperature. Therefore, we chose not to control the temperature to simplify the protocol and develop a cost-effective measurement method. However, significant differences in the 15 min FI and temperature were observed (Table 1). Therefore, controlling the temperature can still be considered for more accurate measurements when considering the 15 min FI.
The reliability of the established protocol was confirmed in both the multicenter and validation studies. In the multicenter study, no significant difference between institutes was observed (Table 2). This finding indicates that we could establish a protocol to detect FI with high reproducibility. The results of the validation study were also consistent. No cancer sample showed a 5 min FI below the negative threshold in this study (Fig. 5, Table S6).
Based on the data obtained from the 251 samples analyzed in the multicenter study, the 5 min FI of gGlu-HMRG appeared to correlate with the malignancy of the lesions. However, significant differences between proliferative lesions, low-grade DCIS, middle/high-grade DCIS, and invasive cancer were not detected. The distributions of the 5 min FIs of these lesions overlapped (Fig. 4). This result indicates that we could not distinguish these lesions using this probe, including differences between proliferative lesions and invasive cancers. Therefore, false positives could not be avoided. Surgeons must be careful not to perform unnecessary surgery. To establish a more reliable method for ex vivo diagnosis, the application of other chemical probes or their use in combination with other probes is required. Indeed, several new probes that can detect breast lesions have been recently developed[27, 28]. The simple protocol we developed here could be applied to or used with these probes for a better and more cost-effective diagnosis.
Our three-step study confirmed that the negative threshold of the 5 min FI (0.061) could be used to verify the negative margins within 10 min. At least, we confirmed that no IC was left on the margin surface if the 5 min FI was below the threshold. In the multicenter study, two samples below the threshold contained low-grade DCIS. Thus, if a surgeon judged that the lesions could be controlled with radiotherapy or chemotherapy, then no additional surgery was required. Alternatively, by setting the negative threshold of the 15 min FI to 0.170, false negatives could be avoided for low-grade DCIS (Fig. S5).
Furthermore, our fluorescence-based diagnostic procedure did not prevent further pathological examination of the same sample. Analyzing pathological specimens with the procedure developed here may also help in determining postoperative therapy. Therefore, in combination with the improvement of probes, the method described here can provide reliable intraoperative navigation for surgeons.