Tumour characteristics and change in tumour biomarkers after NET
104 patients with early ER+/HER2 breast cancer were included in our study. The study population presented a mean age at diagnosis of 69 (47–93) years and the mean NET duration before surgery was 7 months (3–39). The mean tumour size was 25 mm (10–90) assessed by MRI and 18 mm (40–70) by USS. The main administered NET drug was letrozole (n = 100), but some patients also received anastrozole (n = 2), tamoxifen (n = 1), or exemestane (n = 1). One patient was diagnosed with bilateral disease and her two tumours were independently considered in our analyses. The principal characteristics of the tumours and their surgical management as well as the pathological changes after NET are summarised in Table 1. No significant decrease of histological grade was observed after NET (p = 0.12). Ki67, ER and PgR expression was assessed in all tumours pre- and post- NET treatment. As seen in previous studies using similar cohorts (Toi et al. 2011; Martí et al. 2022), NET significantly decreased all these three parameters, being the changes in Ki67 and PgR the most significant (both p < 0.0001, Table 1 and supplementary material, Figure S1). While only 13 patients (12%) were cN + before treatment, 26 patients (25%) were pN + at pathological assessment. Regarding pathological tumour response to NET, only one case of pCR was recorded and most cases (72%) showed partial path-TR (Table 2). Most patients (81%) fell into low (I, n = 34, 35%) and intermediate (II, n = 45, 46%) mPEPI risk groups. These results are in agreement with the response rates obtained in similar series8,10,12.
Radiological examination of tumour size after NET underestimates pathological tumour size
To determine which is the best radiological technique to predict pathological tumour size (path-TS) after NET, we compared tumour size measured by MRI and USS before and after treatment. As expected, radiological tumour size (rad-TS), measured by MRI or USS, both at diagnosis and after NET (just before surgery), significantly correlated with path-TS (Fig. 1A-D). Surprisingly, our results showed that path-TS correlated better with tumour size assessed by MRI and USS at diagnosis than after NET (Fig. 1A). This may suggest that the radiological evaluation before surgery may not be a very precise technique to assess tumour size after NET. To better visualize this, we compared the mean value of tumour size assessed by each radiological technique, before and after NET, and by path-TS. As shown in Fig. 1E, MRI/USS measurements after NET were significantly lower than path-TS and, interestingly, radiological measures at diagnosis were more similar to path-TS than the measures after treatment. Actually, MRI and USS before surgery underestimated path-TS in 77% (76/99) and 92% (84/91) of the cases, respectively.
Importantly, we also found that this disagreement in tumour size estimation by imaging and histopathological analysis affects the concordance between radiological (rad-TR) and pathological (path-TR) tumour response (Table 2). Complete rad-TR was observed in 27 (by MRI) and 16 (by USS) patients while only one patient presented a pCR by pathological assessment. To better visualize these discrepancies, we plotted the correlation between rad- and path-TR. As shown in Fig. 2A-B, we found that rad-TR assessed by MRI correlated better with path-TR than rad-TR assessed by USS, although both associations were statistically significant. Interestingly, we observed that a considerable number of tumours presented a complete (100%) rad-TR after NET but presented a low path-TR (G2 or G3, highlighted in red in Fig. 2A-B). Taken together, our data indicate that the radiological examination of tumour size after NET and before surgery underestimates path-TS bearing surgical implications for the definition of tumour area.
Next, we evaluated the association between rad- and path-TR with the two most accepted prognostic markers after NET: Ki67 levels and mPEPI score11,18,28. As expected, pathological responders presented significantly lower Ki67 levels at surgery and mPEPI score (Fig. 2C-D). Regarding rad-TR, both prognostic markers were associated with tumour response assessed by MRI (Fig. 2E-F), but, in the case of USS, there was not association between tumour response and Ki67 and PEPI score (Fig. 2G-H).
In summary, our data support that radiological evaluation of tumour size after NET underestimates pathological tumour size and indicate that MRI could be more reliable than USS to assess response to NET.
Tumour cellularity size is a new parameter to standardize the assessment of residual tumour content after NET
Diffuse cell loss has been observed as a common pattern of tumour response after neoadjuvant therapies in ER+ (luminal) tumours23. In an attempt to better assess tumour response after NET, we propose a novel parameter called tumour cellularity size (TCS). TCS is the product of tumour cellularity (%) and tumour diameter (path-TS, in mm) and estimates the volume of remaining cells in the tumour bed after NET. First, we evaluated how TCS relates to radiological tumour size and response (Fig. 3). As seen in Fig. 3A, TCS values were much lower than path-TS and more similar to MRI or USS measures after NET compared to rad-TS at diagnosis or path-TS. We then analysed how TCS associates with radiological and pathological response (Fig. 3B-D). Our results showed that TCS inversely correlated with path-TR and with MRI rad-TR (Fig. 3B-C). However, the association between TCS and rad-TR determined by USS was not significant (Fig. 3D), in line with previous results supporting that MRI may be more adequate than USS to quantify response to NET. Taken together, our data indicate that TCS can quantify the tumour ‘’diffuse cell loss’’ response observed in ER + BC tumours after NET, and may capture better the biological response of those tumours and explain why the radiological pre-operative assessment of tumour size underestimates the path-TS.
In order to further evaluate if TCS can be used as a biomarker of response and prognosis for patients undergoing NET, we evaluated its association with changes in Ki67 (∆Ki67) (Fig. 3E-G) and Ki67 levels at surgery (supplementary material, Figure S2), well-established prognostic markers after NET. We observed that ∆Ki67 and Ki67 expression at surgery correlated better with TCS than with tumour cellularity or path-TS (Fig. 3E-G, and supplementary material, Figure S2A-C). Consequently, tumours with high residual Ki67 expression (∆Ki67 > 0 and high Ki67 expression at surgery) also present a high TCS, suggesting that TCS could be a promising biomarker of response to NET.
Finally, to identify an initial cut-off value for which TCS can divide patients responding to NET from no responder patients, we analysed the relationship of TCS quartiles with Ki67 (Fig. 4 and supplementary material, Figure S3A-B). As mentioned before, TCS is positively correlated with Ki67 at surgery and ∆Ki67 (Figure S3). Tumours with TCS < 2.5 mm (Q1) showed significantly lower Ki67 levels at surgery and ∆Ki67 compared with tumours with TCS ≥ 2.5 mm (Q2, Q3 and Q4, Fig. 4), suggesting that a TCS value < 2.5 mm could be used as a good cut-off value to identify patients responding to NET. However, more studies in independent cohorts with associated survival data are needed.