The impact of extensive intraductal component (EIC) on the genomic risk of recurrence in early hormone receptor positive breast cancer

Abstract

Background

Early invasive ductal carcinoma (IDC) breast cancer often presents with a coexisting ductal carcinoma in situ (DCIS) component, while about 5 % of cases present with an extensive (>25 %) intraductal component (EIC). The impact of EIC on the genomic risk of recurrence is unclear.

Methods

Patients with early hormone receptor-positive HER2neu-negative (HR + HER2-) IDC breast cancer and a known OncotypeDX Breast Recurrence Score® (RS) who underwent breast surgery at our institute were included. Using a rule-based text-analysis algorithm, we analyzed pathological reports and categorized patients into three groups: EIC, non-extensive DCIS (DCIS-L), and pure-IDC (NO-DCIS). Genomic risk was determined using OncotypeDX RS.

Results

A total of 33 (4.6 %) EIC cases, 377 (57.2 %) DCIS-L cases and 307 (42.8 %) NO-DCIS cases were identified. Patients in the EIC group were younger and had lower tumor grades than other groups. The distribution of genomic risk varied between the groups, with EIC tumors significantly less likely to have a high RS (>25) compared to DCIS-L and No-DCIS tumors (3 % vs 20 % and 20 %, respectively; p = 0.03). When adjusted to age, tumor size, grade and LNs involvement, both DCIS-L and NO-DCIS groups were significantly correlated with a higher probability of high RS compared to the EIC group (OR 12.3 and OR 13.1, respectively; p < 0.02). Moreover, patients with EIC had a lower likelihood for adjuvant chemotherapy recommendation.

Conclusions

In early HR + HER2- IDC, an EIC correlates with a reduced genomic recurrence risk. The impact on genomic risk seems to be influenced by the extent, not merely the presence, of DCIS.

Highlights

• •Invasive ductal carcinoma (IDC) breast cancer often presents with a coexisting ductal carcinoma in situ (DCIS) component.
• •About 5 % of cases present with an extensive (>25 %) intraductal component (EIC).
• •The presence of EIC is correlated with a lower Oncotype DX recurrence score (RS) compared to pure IDC.
• •The presence of coexisting non-extensive DCIS does not predict a lower genomic risk compared to pure IDC.
• •The use of an automatic rule-based text-analysis algorithm in pathology reports was efficient and valid.

1. Introduction

Breast cancer is the most common cancer among women with invasive ductal carcinoma (IDC) being the most prevalent form. IDC develops when epithelial cells in the milk ducts undergo cancerous transformation and invade the basement membrane of the duct walls into the surrounding breast tissue (stroma) [1]. Ductal carcinoma in situ (DCIS) is non-invasive precancerous lesion, characterized by abnormal cells inside milk ducts that do not invade the surrounding breast tissue [2]. If left untreated, about 30 % of DCIS cases may develop into invasive carcinoma [[3], [4], [5]].

IDC presents with a coexisting DCIS component in up to 60 % of cases [6,7]. In 5–20 % of those cases IDC co-present with an extensive intraductal component (EIC), defined as DCIS within the invasive tumor that occupies more than 25 % of the tumor and DCIS area [[7], [8], [9], [10]]. The co-presence of a DCIS component within IDC, was previously shown to be associated with favorable clinicopathological characteristics and survival outcomes compared to pure IDC [6,7,[11], [12], [13]]. EIC was found to be associated with worse local recurrence rates when surgical margins are positive [9,14], but with better overall survival compared to pure IDC and a less extensive (<25 %) DCIS component when surgical margins are negative [7]. However, currently, the presence and extent of a coexisting DCIS component have no effect on tumor staging or treatment decisions.

The treatment of early hormone receptor positive HER2neu-negative (HR + HER2-) breast cancer includes local therapy (surgery and radiation) and systemic therapy. Adjuvant systemic chemotherapy is recommended in addition to standard preventive endocrine therapy for these patients, based on a combination of clinical parameters and genomic expression profiles [15,16]. The OncotypeDX Breast Recurrence Score (RS) is both prognostic for disease recurrence [17] and predictive of adjuvant chemotherapy benefit in node-negative patients [[18], [19], [20], [21]] and postmenopausal women with 1–3 positive nodes [[22], [23], [24], [25]]. Zeng et al. showed that the presence of DCIS component within IDC is associated with a lower OncotypeDX score and therefore with a lower genomic risk of recurrence [26]. It is unclear, however, whether this observation applies for all tumors with coexisting DCIS or primarily to tumors with EIC.

The current study aimed to investigate the association between the presence of EIC and the genomic risk of recurrence in patients with early HR + HER2- IDC. We employed a rule-based text-analysis algorithm to identify the presence and extent of coexisting DCIS in surgical pathology reports of consecutive women with early HR + HER2-breast cancer and a known OncotypeDX RS. We then conducted univariate and multivariate analyses to evaluate the impact of the EIC on the genomic and clinical risk of recurrence, as well as on distant disease recurrence rates.

2. Methods

2.1. Patients

All patients who underwent surgery between 2004 and 2021 at Tel Aviv Sourasky Medical Center (TASMC) for early HR + HER2- IDC, and had a known OncotypeDX RS were included in this study. Patients with a non-IDC breast cancer were excluded from the study. The OncotypeDX test was used to guide chemotherapy recommendation for women with HR + HER2-early breast cancer. The test was performed for patients with a node-negative disease or 1–3 positive nodes (N1 disease), as well as for a minority of patients with >3 nodes for whom the institutional tumor board considered chemotherapy could potentially be omitted based on the test results. The test was not performed for women for whom the institutional tumor board believed that the results of the test would not alter the treatment recommendation. The majority of the patients underwent upfront surgery, and the test was conducted on the surgical pathology. For 45 patients (6.2 %), the test was performed on biopsy specimens as neoadjuvant treatment was considered. Among these patients, 35 received neoadjuvant therapy prior to surgery; 13 underwent chemotherapy, and 22 received neoadjuvant endocrine therapy. The presence and extent of DCIS in surgical pathological reports were automatically determined using a rule-based text-analysis algorithm (see detailed method below), and patients were categorized into three groups accordingly: (1) EIC, with a DCIS component greater than 25 %; (2) DCIS-Low (DCIS-L), with a DCIS component less than 25 %; and (3) NO-DCIS, indicating pure IDC with no evidence of DCIS. We retrospectively retrieved clinicopathological characteristics, including age at diagnosis, tumor size, grade, lymph nodes (LNs) involvement, and distant disease recurrence (DDR). The study was approved by the local Institutional Ethics Committee under the number TLV18-0426.

2.2. Automatic extraction of the present and extent of coexisting DCIS using a rule-based text-analysis algorithm

We analyzed all free-text surgical pathology reports of patients with early invasive HR + HER2-breast cancer, who had undergone surgery (lumpectomy or mastectomy) at our institute and had a known OncotypeDX RS score. Reports with a non-IDC histological subtype (e.g., ILC, Medullary, Papillary, etc.) were excluded from this analysis. We developed a rule-based text analysis algorithm to identify the presence and extent of a coexisting DCIS component within IDC. Initially, for each pathology report, we determined the presence of a DCIS component, and then we assessed whether the DCIS was extensive or non-extensive. To identify the presence of DCIS, we searched the reports for specific keywords ('DCIS', 'in situ', 'intraductal'). We acknowledged that some of the keywords may retrieve reports without a DCIS component (for instance ‘in situ’ may be found in the context of ‘lobular carcinoma in situ’ and ‘DCIS’ may be found in the context of ‘negative for DCIS’), leading to false positives. To refine our search and exclude these false positives from our DCIS-related group, we looked for additional specific keywords (e.g., 'lobular') near the original DCIS-related keyword and accordingly excluded those cases from our DCIS-related group. Additionally, we employed negation-detection rules to exclude expressions that negate the presence of DCIS. Excluded reports or reports lacking any DCIS-related keywords were classified under the NO-DCIS (pure IDC) group. Next, we evaluated the extent of DCIS within the DCIS-related reports, and categorized each report into either the EIC (extensive) or DCIS-L (non-extensive) groups. Similar to our initial approach, we searched for the term 'extensive' in proximity to the DCIS-related keywords. To reduce false positives, we introduced another set of negation-detection rules to identify when the term 'extensive' was negated (as in 'not extensive' or 'negative for extensive' etc.). Reports with such negations were excluded from the EIC group and were classified under the DCIS-L group. Additionally, reports not identified as EIC within the DCIS-related group were automatically categorized as DCIS-L. To assess the accuracy of our algorithm, it was validated by a breast medical oncologist. Our evaluation set collectively contains 93 reports, which represent 13 % of the total cases. The overall accuracy was 98 %: 100 % in both the EIC and NO-DCIS groups and 93 % in the DCIS-L group. Two cases were misclassified; both were incorrectly placed into the DCIS-L group but should have been categorized in the NO-DCIS group. In the first case, the patient underwent two breast surgeries; pathology revealed DCIS in one and pure IDC (on which the Oncotype test was conducted) in the other. In the second case, the patient had IDC with ductal papilloma, which our algorithm mistakenly identified as DCIS. The two misclassified cases were manually reassigned to the NO-DCIS group for further analysis.

2.3. Assessment of the genomic and clinical risk

Genomic risk: Genomic risk was determined by OncotypeDX RS. RS ≤ 25 was considered low-intermediate risk and RS ≥ 26 was considered high risk [20,23].

Clinical risk: The clinical risk assessment in this study was based on the Adjuvant! Online algorithm (version 8), which integrates tumor size, grade, and nodal status to predict cancer outcomes at 10 years [27,28]. Although Adjuvant! is no longer available online for clinical use, the study used a binary clinical-risk categorization model based on the Adjuvant! algorithm, as applied in the MINDACT trial (See appendix table S13 in Cardoso et al.) [29]. A low clinical risk was defined as greater than 92 % probability of breast cancer–specific survival at 10 years in women with HR + HER2-tumors who received endocrine therapy alone. For N0/N1mic patients, the clinical risk was defined as low if one of the following conditions was present: grade I and tumor size ≤3 cm, or grade II and tumor size ≤2 cm, or grade III and tumor size ≤1 cm. For N1 patients, the clinical risk was defined as low only if it was Grade I and the tumor size was ≤2 cm. Otherwise, the clinical risk was defined as high.

2.4. Assessment of the probability for adjuvant chemotherapy recommendation

The probability for adjuvant chemotherapy recommendation was calculated based on the genomic and clinical risk according to the algorithm suggested by the results of the TAILORx study [20], subsequent analysis by Sparano et al. [16], and the RxPONDER study [23], as previously described [30]. In essence, this model indicates that for node-negative disease adjuvant chemotherapy is considered in women of all ages with a high RS (≥26). Additionally, chemotherapy should be considered in younger women (Age ≤50) with a RS 16 or higher, based on their clinical risk; for high clinical risk tumors, chemotherapy should be considered for a RS ≥ 16 and for low clinical risk for a RS ≥ 21. For patients with positive nodes: Chemotherapy is recommended for postmenopausal patients with 1–3 nodes (N1) and a high RS (≥26) or patients with more than 3 nodes (N2/N3). Additionally, chemotherapy is recommended for all premenopausal patients with positive nodes (N1/N2/N3).

2.5. Statistical analysis

Demographic and clinicopathological data of patients in the three groups (EIC, DCIS-L and NO-DCIS) were compared, using the chi-square test for categorical variables and the Kruskal-Wallis test for continuous variables. Distribution of genomic risk and clinical risk was compared between the groups using the chi‐square test. Multivariate analysis was done using logistic regression. Hazard ratios for DDR were estimated using a univariate Cox regression analysis. All p values were two-sided and p < 00.05 was considered significant. IBM SPSS Statistics for Windows, version 28 (IBM Corp., Armonk, NY, USA) was used for all statistical analyses.

3. Results

3.1. Patient characteristics

Our cohort consisted of 717 consecutive patients diagnosed with HR + HER2- IDC, all of whom had known OncotypeDX scores and available surgical pathological reports. The rule-based text-analysis algorithm identified a coexisting DCIS component in 410 (57.2 %) patients. Of these, 33 (4.6 %) had an EIC (>25 %; EIC group), while the remaining 377 (52.6 %) patients had a non-extensive DCIS component (DCIS-L group). Additionally, 307 (42.8 %) patients showed no evidence of DCIS (NO-DCIS group) in their pathological reports (Fig. 1). Patient characteristics are summarized in Table 1. Women in the EIC group were significantly younger than those in the DCIS-L or NO-DCIS groups, with median ages of 52, 58, and 62 years, respectively (p < 0.001). Although a lower proportion of women in the EIC group had high-grade (G3) tumors compared to those in the DCIS-L or NO-DCIS groups (15.2 % vs. 28.9 % and 26.4 %, respectively), this difference was not statistically significant. Additionally, no significant differences were observed in the proportions of patients with lymph node involvement or large tumor size (>2 cm) among the three groups.

Fig. 1.

Study flow diagram and implantation of rule-based text-analysis algorithm. Abbreviations: IDC, invasive ductal carcinoma; HR + HER2-, hormone receptor positive and HER2/Neo negative.

Table 1.

Clinicopathological characteristics according to the presence and extent of coexisting DCIS. Abbreviations: LN, lymph node.

Characteristics	EIC (n = 33)	DCIS-L (n = 377)	NO-DCIS (n = 307)	p value
Age (median)	52	58	62	<0.001
Age group, n (%)				<0.001
≤50	16 (48.5)	113 (30)	64 (20.8)
>50	17 (51.5)	264 (70)	243 (79.2)
Tumor size, n (%)				0.29
≤2 cm	21 (63.6)	280 (74.3)	232 (75.6)
>2 cm	12 (36.4)	97 (25.7)	73 (23.8)
Missing data	(0)	(0)	2 (0.7)
Grade, n (%)				0.13
Grade 1	1 (3)	24 (6.4)	9 (2.9)
Grade 2	25 (75.8)	236 (62.6)	209 (68.1)
Grade 3	5 (15.2)	109 (28.9)	81 (26.4)
Missing data	2 (6.1)	8 (2.1)	8 (2.6 %)
LN involvement, n (%)				0.13
N0/N1mic	28 (84.8)	312 (82.8)	235 (76.5)
N1	4 (12.1)	63 (16.7)	69 (22.5)
N2–N3	1 (3)	2 (0.5)	3 (1)

3.2. Association between the presence and extent of coexisting DCIS, and the genomic risk of recurrence

The genomic risk for disease recurrence was assessed by OncotypeDX RS. When comparing patients with coexisting DCIS component (EIC and DCIS-L groups combined) and pure IDC (NO-DCIS group), no differences were found between the groups in RS distribution (p = 0.6, Fig. 2A). However, when evaluating the EIC and DCIS-L groups separately and comparing the three groups (EIC, DCIS-L, and NO-DCIS), the distribution of RS significantly varied between the groups (p = 0.03, Fig. 2A). The proportion of patients with low/intermediate RS ($\le$ 25) was higher in the EIC group compared to the DCIS-L or the NO-DCIS groups (97 % vs. 80 % and 80 %, respectively). Conversely, only 3 % of the EIC group had a high RS (>25) compared to 20 % in the DCIS-L and the NO-DCIS groups. Moreover, in a multivariate analysis, the presence of EIC was found to be an independent predictive variable for RS, when adjusted for age, tumor size, grade, LNs involvement and the year of breast cancer diagnosis (Table 2). Both the DCIS -L group and the NO-DCIS group were significantly correlated with a higher probability of high RS compared to the EIC group (DCIS-L: OR 12.3, 95 % CI 1.5–100.1; p = 0.019. NO-DCIS: OR: 13.1, 95 % CI 1.6–107.3; p = 0.016). Additionally, high grade (G3) and large tumor size (>2 cm) were identified as significant independent predictors for the likelihood of high RS, when adjusted for other variables (Table 2).

Fig. 2.

Genomic and clinical risk distribution by the presence and extent of coexisting DCIS. A, Genomic risk distribution according to DCIS (any coexisting DCIS) and NO-DCIS groups (upper) and according to EIC, DCIS-L and NO-DCIS groups (lower). Low/Intermediate RS, RS ≤ 25; High RS, RS > 25. *p = 0.03. B, Clinical risk distribution according to genomic risk. Low/Intermediate RS, RS ≤ 25; High RS, RS > 25. **p < 0.001. C, Clinical risk distribution according to the presence and extent of coexisting DCIS.

Table 2.

A multivariate analysis of clinicopathological factors influencing the genomic risk of recurrence. The genomic risk was assessed by OncotypeDX RS. Abbreviations: LNs, lymph nodes; RS, OncotypeDX Recurrence Score. * The year of breast cancer diagnosis was included in the model as a continuous variable.

Probability of high RS	OR (95 % CI)	p value
Coexisting DCIS
EIC	1
DCIS-L	12.3 (1.5–100.1)	0.019
NO-DCIS	13.1 (1.6–107.3)	0.016
Year of diagnosis*	0.9 (0.9–1.01)	0.14
Age group
≤50	1
>50	0.9 (0.6–1.4)	0.63
Tumor size
<2	1
>2	2.1 (1.5–3.6)	<0.001
Grade
G1	1
G2	4.5 (0.6–34)	0.14
G3	24.1 (3.2–181.8)	0.002
LNs involvement
N0/Nmic	1
N1	0.8 (0.5–1.4)	0.46
N2/N3	0.34 (0.03–3.7)	0.37

3.3. Association between the presence and extent of coexisting DCIS, the clinical risk of recurrence, and chemotherapy recommendation

The clinical risk for disease recurrence was assessed using tumor size, grade, and nodal status. As expected, the clinical risk was significantly correlated with the genomic risk of recurrence (Fig. 2B; p < 0.001). When exploring the correlation between the presence and extent of coexisting DCIS component and the clinical risk of recurrence, no significant differences were found between the groups (Fig. 2C). The likelihood of a chemotherapy recommendation was assessed using an algorithm that reflects contemporary guidelines, based on factors such as age, nodal status, genomic risk, and clinical risk (see Methods section). Patients in the EIC group had a lower likelihood of receiving a chemotherapy recommendation based on this algorithm compared to those in the DCIS-L or the NO-DCIS groups (24 % compared to 31 % and 29 %, respectively; Supplementary Table 1).

3.4. Association between the presence and extent of coexisting DCIS, and distance disease recurrence (DDR)

The median follow-up for DDR was 76.8 months. Univariate Cox regression analysis demonstrated a significantly higher hazard ratio for DDR in patients with high RS compared to low/intermediate RS (HR = 2.14, 95%CI: 1.23–3.75, p = 0.007; Fig. 3). Similarly, a higher hazard ratio for DDR was found for patients with high clinical risk compared to low clinical risk (HR = 2.12, 95%CI: 1.2–3.7, p = 0.009; Fig. 3). However, no significant correlation was found between the presence and extent of coexisting DCIS and the incidence of DDR (HR = 0.6, 95 % CI: 0.19–2.11, p = 0.46 for the DCIS-L compared to the EIC groups; and HR = 0.9, 95 % CI: 0.28–3.03, p = 0.88 for the NO-DCIS compared to the EIC groups; Fig. 3).

Fig. 3.

Cox Regression analysis of distant disease recurrence (DDR). A, DDR according to genomic risk, assessed by OncotypeDX RS. B, DDR according to the clinical risk. The clinical risk was calculated based on tumor size, grade, and lymph node involvement. C, DDR according to the presence and extent of coexisting DCIS. Abbreviations: RS, OncotypeDX Recurrence Score.

4. Discussion

In our study, we explored the impact of the presence and extent of DCIS component within IDC on the genomic and clinical risk of disease recurrence in women with early HR + HER2-breast cancer. Our results revealed that patients with EIC (>25 %) demonstrated a significantly lower likelihood of high genomic risk (RS > 25) compared to those with pure IDC or with non-extensive coexisting DCIS (OR 12.3 and OR 13.1 for the NO-DCIS and the DCIS-L group compared to the EIC group, respectively). Nevertheless, no significant difference was found in the clinical risk of recurrence between the groups. However, patients with EIC were less likely to receive an algorithm-based recommendation for adjuvant chemotherapy.

The association between favorable clinicopathological characteristics and therefore favorable biological behavior of tumors with coexisting DCIS compared to pure IDC is well established [7,31,32]. Similar to previous reports, our results indicate that a DCIS component is correlated with younger age and a lower tumor grade [7,8,11,31]. In addition to traditional clinicopathological factors, gene expression profiles have become a significant factor in assessing tumor prognosis and defining treatment. Using the OncotypeDX RS test with a slightly different RS risk cutoff than used in our work, Zeng et al. demonstrated that a DCIS component is associated with a lower genomic risk of recurrence [26]. However, their initial analysis grouped together all tumors with a DCIS component, suggesting a general trend in which the presence of DCIS is associated with a lower genomic risk. While they demonstrated that within the subgroup of tumors with coexisting DCIS, patients with a very high DCIS proportion (>50 %) were more likely to have a lower genomic risk compared to those with lower proportions (<50 %), this analysis did not include the 'pure IDC' group [26]. The exclusion of this group limits the ability to draw definitive conclusions about the influence of the extent of DCIS on genomic risk. Employing the same method of genomic risk assessment, we showed that only the subpopulation of patients with EIC (>25 %) exhibited a lower genomic risk compared to those with pure IDC. In contrast, no significant differences in genomic risk were found for patients with lower proportions of DCIS (<25 %). Thus, our findings highlight that the mere presence of a DCIS component does not necessarily impact the genomic risk of recurrence; rather, it is the extent of the DCIS that is more significantly associated with genomic risk and prognosis.

Numerous studies have highlighted the significant impact of the extent of DCIS within IDC on tumor biology and prognosis. Wang et al. discovered a negative correlation between the extent of DCIS within IDC and Ki67 expression levels, such that tumors with a high proportion of DCIS are associated with reduced Ki67 expression levels compared to those with a lower DCIS proportion [11]. Zeng et al. showed that a higher proportion of DCIS within IDC is correlated with lower expression levels of the invasion-related gene MMP11, and the CD68 gene, which is a marker for tumor-associated macrophages [26]. Both high levels of MMP11 and CD68 were found to be correlated with worse breast cancer prognosis [33,34]. Considering that KI67, MMP11, and CD68 are all evaluated in the OncotypeDX RS test, this could partly explain our findings regarding the association between the presence of EIC and a lower genomic risk. Furthermore, Kole et al. demonstrated a significant correlation between the extent of DCIS and overall survival. Patients with EIC had improved OS rates compared to those with non-extensive DCIS components (<25 %) or pure IDC (5-year OS rates of 90.0 %, 88.5 %, and 85.5 %, respectively) [7]. Our results could offer a potential explanation for Kole et al.’s observation, suggesting favorable gene expression profiles in tumors with an EIC component, which could lead to better breast cancer outcomes.

In accordance with contemporary guidelines, the decision to recommend adjuvant chemotherapy for patients with early HR + HER2-breast cancer is based on both clinical factors and gene expression profiles, such as the OncotypeDX RS test [35,36]. We demonstrated in our cohort that patients with EIC had a lower likelihood of receiving an algorithm-based chemotherapy recommendation (Supplementary Table 1). Gordo et al. showed that among patients who received neoadjuvant chemotherapy, those with a coexisting DCIS component were less likely to respond to the treatment. Interestingly, these patients exhibited longer DFS and higher overall survival rates despite the lower chemotherapy response, suggesting that reduced tumor aggressiveness might lead to both a lower chemotherapy response and better outcomes [37]. However, the impact of the extent of DCIS on chemotherapy response and outcomes was not assessed in the above-mentioned work. Our findings suggest that these results may be attributed to the subpopulation of patients with a high DCIS proportion, and implies a reduced benefit from adjuvant chemotherapy treatment for this unique population.

Although, as expected, the genomic and clinical risks were significantly correlated with DDR in our cohort, we could not detect a significant correlation between the presence and extent of DCIS and DDR. One explanation might be the rarity of EIC among the early HR + HER2-breast cancer population (about 5 % of cases) [[7], [8], [9]], which reduces the statistical power of this analysis. Therefore, further research on larger cohorts is needed to assess the correlation between EIC, genomic risk of recurrence, and DDR.

Our study has several limitations. Firstly, it is a single institution retrospective study. Secondly, while in our study the presence and extent of coexisting DCIS within pathology reports was determined automatically, reducing investigator bias, the initial assessment of DCIS extent within tumor tissue is conducted by pathologists and therefore may be subject to cognitive biases and inter-pathologist discordance [38]. Nevertheless, the use of a rule-based algorithm can potentially be a limitation due to its automatic approach. However, our human validation set demonstrated an overall accuracy of 98 % for the algorithm. Additionally, the selection of women for the Oncotype DX test may have introduced a bias by excluding those at very high or very low clinical risk, who might have had clear treatment recommendation regardless of test results. However, such bias, if present, would likely impact all DCIS groups, not just the EIC group. Finally, as mentioned above, although the overall cohort is relatively large, the EIC group is fairly small due to the rarity of this condition, which might affect the statistical power of our analyses.

In conclusion, our work demonstrated a significant correlation between the presence of EIC and a favorable genomic risk, suggesting a better prognosis and reduced benefit from adjuvant chemotherapy for this population. These findings add to the growing body of evidence indicating that the presence of a DCIS component is associated with reduced tumor aggressiveness and a favorable prognosis. As no differences in genomic risk were found in non-extensive DCIS cases, our work suggests that the extent of DCIS, and not merely its presence, is an important factor. More research on larger cohorts is needed to better define the correlation between EIC, genomic risk, and prognosis.

Funding

This research did not receive any specific grant from funding agencies in the public, commercial, or not-for-profit sectors.

Data sharing

The datasets generated during and/or analyzed during the current study are available from the corresponding author on reasonable request.

Ethical approval

The study was approved by the local Institutional Ethics Committee and was performed in line with the principles of the Declaration of Helsinki.

Consent to participate

As data were aggregative and anonymous no informed consent was required by the institutional Committee.

CRediT authorship contribution statement

Yael Bar: Writing – review & editing, Writing – original draft, Investigation, Formal analysis, Data curation, Conceptualization. Kfir Bar: Writing – review & editing, Writing – original draft, Data curation, Conceptualization. Didi Feldman: Writing – review & editing, Data curation. Judith Ben- Dror: Writing – review & editing, Data curation. Meishar Shahoha: Writing – review & editing, Data curation. Shir Lerner: Writing – review & editing, Methodology, Formal analysis. Shlomit Strulov Shachar: Writing – review & editing. Ahuva Weiss-Meilik: Writing – review & editing. Nachum Dershowitz: Writing – review & editing. Ido Wolf: Writing – review & editing. Amir Sonnenblick: Writing – review & editing, Writing – original draft, Conceptualization.

Declaration of competing interest

Y.Bar reports consulting fees from Eli lilly, Roche, Stemline and Gilead and travel fees from Gilead. A.Sonnenblick reports consulting fees from Eli lilly, Teva, Pfizer, Novartis, Roche, Gilead, MSD, Astra-Zenca, Progenetics and Rhenium; Travel fees from Neopharm, Celgene, Medison, Roche and MSD and grant support from Novartis and Roche, all outside the submitted work. I.Wolf reports research grants from MSD, BMS, Roche and Novartis; S. Strulov Shachar reports consulting fees from: Pfizer, Novartis, Roche, Medison, MSD, AstraZeneca, Eli Lilly, ProGenetics, Gilead and Stemline and travel fees from Pfizer, Roche, Gilead and AstraZeneca, all outside the submitted work. The rest of the authors declare that they have no conflict of interest.

Appendix A. Supplementary data

The following is the Supplementary data to this article: