Tumor‐infiltrating lymphocytes in breast FNA biopsy cytology: A predictor of tumor‐infiltrating lymphocytes in histologic evaluation

Abstract

Background Tumor‐infiltrating lymphocytes (TILs) are associated with various clinicopathological features. Using cytologic specimens for assessing TILs remains to be established. This retrospective study aimed to establish a practical method to assess TILs in cytologic samples. Methods The authors found 1101 breast fine‐needle aspiration biopsy (FNAB) cytology samples in their hospital, and 214 of them met the inclusion criteria. The TILs score was evaluated using histologic slides, and breast cancers were divided into 2 groups: low‐ (<60%) and high‐TILs (≥60%). Training and validation tests composed of 50 breast cancer samples each were constructed. A cytologic TILs (cTILs) score was introduced to evaluate lymphocytes in FNAB cytology and it was compared with histologically evaluated TILs. The cTILs score was calculated by subtracting the number of neutrophils from the number of lymphocytes surrounding the tumor cells. Results In the training test, a 2‐tier system with low‐ and high‐TILs groups showed a large area under the curve (AUC) (0.943; 95% confidence interval [CI], 0.84‐0.99). A cTILs score cutoff value of >8 had 87.5% sensitivity and 90.5% specificity. In the validation test, the AUC was 0.79 (95% CI, 0.6‐0.93) whereas sensitivity and specificity were 57% and 89.5%, respectively. When small tumors <0.5 cm were excluded, the AUC improved to 0.93 (95% CI, 0.83‐1.0), and sensitivity and specificity were 80% and 88.5%, respectively. Conclusions The cTILs scoring system had acceptable reproducibility and concordance with TILs on histologic samples for tumors ≥0.5 cm. Cytologic evaluation can potentially substitute for histologic evaluation of TILs.

Investigating the role of core needle biopsy in evaluating tumor-stroma ratio (TSR) of invasive breast cancer: a retrospective study

“…The 95% con dence interval (95%CI) of sensitivity and speci city were calculated, using 1000 bootstrap samples. The methodology of the present study was similar to our previous study [10].…”

Limitations and opportunities for tumor infiltrating lymphocytes in breast FNA

“…Thus, evaluating cytologic TILs should be reserved for larger tumors and cases, where the FNA is obtained with no peripheral blood contamination, and the smears are well prepared with preserved architecture. Odate et al 13 are careful not to overstate the value of their observation by raising these issues directly in their article. We commend the authors on this interesting and novel application for breast FNA biopsy to establish and validate a methodology for evaluating TILs in cytologic specimens.…”

“…Perhaps, digital imaging and artificial intelligence might increase the precision of measuring the exact number of TILs per mm 2 stromal tissue. Through an interesting and forthright re‐evaluation of breast FNA, Odate et al 13 may have found a role for many new and exciting applications of breast cytology in the future.…”

“…In a novel and interesting approach to this problem, Odate et al 13 examined TILs in breast fine‐needle aspiration (FNA) biopsy and correlated their findings with histologic evaluation. Although sampling is always a concern with FNA biopsy, it has been shown that FNA frequently provides better cellularity and higher tumor fraction than concurrently acquired core needle biopsy samples 14 .…”

“…By dividing their samples into low (<60%) and high TILs (≥60%), the authors reduce the confounding variable of reporting TILs in 10th percentiles with histologic evaluation. Limiting factors in their study included peripheral blood contamination, and the authors note that “even a small amount of peripheral blood contamination increased the number of lymphocytes.” 13 The authors subtracted the number of neutrophils from the number of lymphocytes to reduce the effect of peripheral blood contamination. Additionally, they were only able to show “a good association between the histologic and cytologic TILs for tumors larger than 0.5 cm” with an area under the curve of 0.93 (95% confidence interval, 0.83‐1.0).…”