Background

The role of tumor-infiltrating lymphocytes (TILs) in ER+/HER2- breast cancer (BC) is debated. We evaluated the association of TILs and clinico-pathological (CP) features with distant disease-free survival (DDFS) in a large series of patients (pts) with ER+/HER2- BC.

Methods

A case-cohort was built by randomly selecting 17% of an initial cohort of 3986 pts who underwent surgery at IEO in the period 1998-2002, and for whom long-term follow-up data was available (680 pts). 307 more pts with an event were added to this cohort. TILs were assessed for these 987 cases on centralized H&E slides. TILs were considered both as continuous variable, and dichotomized in low (<5%) vs high (≥5%). DDFS was calculated from the date of surgery to the date of any first event or the date of last contact with the patient. Median f-up was 7.5 years (0.1-10). Differences between BC subtypes were assessed using the log-rank test. Univariable and multivariable Cox proportional hazards regression with inverse sub-cohort sampling probability weighting were used to evaluate the risk across groups. Analyses were carried out with the SAS software version 9.4.

Results

Median TILs was 2%. Higher TILs were positively associated with pN (p = 0.003), grade (p < 0.0001), peritumoral vascular invasion (p = 0.003), Ki-67 (p = 0.0001), lumB subtype (p < 0.0001), and chemotherapy (p < 0.0001), while they were inversely associated with ER (p < 0.0001) and age (p = 0.02). In multivariable regression analysis, only Ki-67 expression retained significant association with TILs. Age and ER showed a trend towards negative association with TILs. In univariate Cox regression, TILs expression (≥5% vs. <5%) was not associated with DDFS (HR 1.08, 95% CI 0.80-1.46, p = 0.62). At stratified cox exploratory analyses, we found an association between high TILs and low risk in very young women (p = 0.03) and G3 tumors (p = 0.047); high TILs were associated with worse outcome in G1 tumors (p = 0.05). TILs were not associated with DDFS in the group without chemo. Instead, in the group with chemo, high TILs were associated with better DDFS (p = 0.006), particularly for ki67≥20% (p = 0.01).

Conclusions

High TILs in ER+/HER2- BC are significantly associated with several CP features of dismal outcome. This subgroup might be more immunogenic, thus deserving the exploration of immunotherapy approaches.

Legal entity responsible for the study

Carmen Criscitiello.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.

Background

A significant percentage of BRCA1-associated breast cancers do not display a clear second hit (loss of heterozygosity, methylation, inactivating mutation) in a large part of the tumor cells. In patients with germline defects in BRCA1, microRNAs could potentially inhibit the remaining functional allele, thereby contributing to oncogenesis in a way similar to more conventional second hits. We hypothesize that overexpression of oncogenic microRNAs may play a role in downregulating the expression of the retained wild type BRCA1 allele, at least on a subclonal level.

Methods

Genome-wide microRNA expression profiling investigated the expression microRNAs in 51 BRCA1-associated breast cancers for which BRCA1 was thoroughly investigated at the (epi-)genetic level. We evaluated the association with molecular subtype and histopathological features, and with the presence of a residual functional BRCA1 allele using differential expression analysis. BRCA1 protein expression in these tumors was studied using immunohistochemistry.

Results

We identified fourteen microRNAs upregulated in tumors retaining the BRCA1 wild type allele in > 50% of tumor cells. Differential expression analysis based on histopathological features of breast tumors was highly similar to publically available data from two prior independent studies. Intratumoral heterogeneity for BRCA1 protein expression was seen in several tumors, but did not always correlate to the observed amount of retained BRCA1 wild type allele.

Conclusions

Our study revealed candidate microRNAs that are potentially active in BRCA1-associated breast tumorigenesis. miRNAs upregulated in tumors with a retained functional BRCA1 allele, lacking BRCA1 protein expression might be involved in BRCA1 regulation. This regulation can occur in a direct or indirect manner and should be further functionally validated. microRNAs involved in BRCA1 regulation have the potential to sensitize tumors to PARP inhibitor or platinum-based therapies.

Legal entity responsible for the study

Ghent University.

Funding

‘Flanders Innovation and Entrepreneurship’ in the form of a doctoral basic strategic research grant [IWT/SB/131739] and the Ghent University Special Research Fund (BOF15/GOA/011).

Disclosure

All authors have declared no conflicts of interest.

Background

Although previous studies have suggested a higher risk of breast cancer recurrence with increasing age, older patients with early stage breast cancer generally receive less extensive treatment. The aim of this study was to evaluate the impact of older age on risk of locoregional and distant recurrence among patients aged 70 years or older in a large population based cohort.

Methods

Patients aged 70 years and older who were diagnosed with early stage breast cancer between 2003 and 2009 and underwent surgery were selected from the Netherlands Cancer Registry. Cumulative incidences of locoregional and distant recurrence according to age categories were calculated using the Cumulative Incidence Competing Risks (CIRC) method. The Fine and Gray model was used to estimate the effect of age on recurrence which was expressed as subdistribution hazard ratio (sHR). A two-step multivariable analysis was performed, the first multivariable model adjusted for patient and tumor characteristics and the second model, the full model, additionally adjusted for treatment characteristics.

Results

Overall, 19748 patients were included. During follow-up, 832 of 7799 patients (10.7%) aged 70-74 years, 707 of 4344 patients (16.3%) aged 75-79 years and 927 of 6322 patients (14.7%) aged ≥80 years had a locoregional or distant recurrence. In univariate analysis, age 75-79 years and ≥80 years were associated with a higher risk of locoregional and distant recurrence compared to the reference age group of 70-74 years. After adjustment in the full multivariable model, the risk of distant recurrence remained significantly higher for patients aged 75-79 (sHR: 1.25, 95% CI 1.11-1.41, p = 0.001), but not for patients aged ≥80 years (sHR: 1.04, 95% CI 0.93-1.18, p = 0.483). For locoregional recurrence risk, no significant association with age was found in the multivariable analysis.

Conclusions

In this large population based study, patients aged 75-79 years were at increased risk of distant recurrence compared to patients aged 70-74 years. Possible explanations can be suboptimal treatment selection and different treatment response in older patients compared to younger patients.

Legal entity responsible for the study

Leiden University Medical Center.

Funding

ZonMw.

Disclosure

All authors have declared no conflicts of interest.

Background

The optimal time to deliver adjuvant chemotherapy has not been defined, although benefit for early chemotherapy (<20 days of surgery) was previously reported in pre-menopausal women with oestrogen-receptor (ER) negative cancers. More recently, a SEER database study demonstrated that overall survival declined with delays of ≥ 60 days.

Methods

We conducted a retrospective study of all patients who received adjuvant anthracycline and/or taxane-based chemotherapy at RMH 1993-2010. Data were collected for clinico-pathological features, surgery, radiotherapy and chemotherapy regimens. The primary endpoint was 5-year disease-free survival (DFS) in patients commencing chemotherapy <31 days after surgery (early chemotherapy) compared to ≥ 31 days (delayed chemotherapy). Secondary endpoints included 5-year overall survival (OS) by early vs delayed chemotherapy, and outcomes in subgroups of patients defined by age, nodal status and ER and HER2 receptor status.

Results

We identified 2003 eligible patients, 99.5% female, median age 50, 82.7% ductal, 57% grade 3, median tumour size 2.2cm, 57.1% with involved lymph nodes (median 1 node), 68% ER positive, 15.7% HER2 positive, 24.6% HER2 unknown. Median follow-up was 115 months. There was no difference in 5-year DFS rate in patients treated with early compared to delayed chemotherapy (81% vs 82% HR 1.13, 95% CI 0.91 -1.40, p = 0.324). No sub-group was identified in whom DFS was significantly affected by delaying chemotherapy. Five year OS rate was similar in patients who received early or delayed chemotherapy (90% compared to 91%, (HR 1.16, 95% CI 0.86-1.57, p = 0.321). For patients with triple negative breast cancer (TNBC, n = 250), mortality was twice as high in patients whose adjuvant chemotherapy was delayed compared to those receiving chemotherapy <31 days of surgery. (HR = 2.18, 95% CI: 1.11-4.30, p = 0.02). Five year OS rate was 77% (95% CI 0.82-0.94) in TNBC patients who received late chemotherapy compared to 89% (95% CI 0.82-0.94) in those receiving early chemotherapy.

Conclusions

Although delayed adjuvant chemotherapy did not affect DFS in our study, in patients with TNBC, delaying adjuvant chemotherapy beyond 30 days significantly reduced 5-year overall survival, and should therefore be avoided when possible.

Legal entity responsible for the study

The Royal Marsden NHS Foundation Trust.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.

Background

DD adjuvant CT is standard of care in high-risk early BC pts. However, the role of DD CT in HER2+ BC pts remains uncertain, particularly when T is administered. In this exploratory analysis of the GIM2 trial (Del Mastro et al, Lancet 2015), we investigated the efficacy of DD CT in the subgroup of HER2+ BC pts with or without subsequent exposure to T.

Methods

Using a 2x2 factorial design, the GIM2 trial randomized node-positive early BC pts to receive 4 cycles of (fluorouracil)epirubicin/cyclophosphamide (F)EC every 2 (DD) or every 3 (standard interval [SI]) weeks followed by 4 cycles of DD or SI paclitaxel (P). The same number of cycles (4 (F)EC and 4 P) and doses (FEC 600/90/600 mg/m2, P 175 mg/m2) were used in all treatment arms. After the approval of adjuvant T, protocol was amended in April 2006 to mandate use of T for 1 year after CT completion in all HER2+ pts. The efficacy of DD CT in terms of disease-free survival (DFS) and overall survival (OS) was compared between HER2+ pts with or without subsequent exposure to T and those with HER2-negative /HER2-unknown (HER2-/ukn) status.

Results

Out of 2,003 pts randomized to DD or SI CT in the GIM2 study, HER2 status was positive in 452 (22.6%) pts, negative in 1,243 (62.0%) and unknown in 308 (15.4%). Among 452 pts with HER2+ disease, T was administered to 132 (29.2%) pts. Overall median follow-up was 8.1 years (interquartile range: 7.0-9.3). There was no significant interaction between T therapy and the effect of DD CT, (p_interaction=0.603 for DFS and p_interaction=0.776 for OS); however, among pts treated with T, the effect of DD CT appeared to be smaller as shown in the table.

Table: 90O

Conclusions

In HER2+ early BC pts, DD adjuvant CT appears to have a role only in pts without subsequent exposure to T.

Clinical trial identification

NCT00433420.

Legal entity responsible for the study

GIM (Gruppo Italiano Mammella) Study Group.

Funding

Bristol-Myers Squibb, Pharmacia, and Dompè Biotec.

Disclosure

M. Lambertini: Advisory role: Teva; Speakers’ honoraria: Theramex. E. de Azambuja: Honoraria: Roche-Genentech; Research grants: Roche-Genentech (to the institution); Travel grants: Roche-Genentech, GlaxoSmithKline outside the submitted work. L. Del Mastro: Personal fees: Novartis Pharma AG, Roche-Genentech, Ipsen, AstraZeneca, Takeda, Eli Lilly outside the submitted work. All other authors have declared no conflicts of interest.

Background

Biomarker/s able to detect minimal residual disease (MRD) after breast cancer (BC) surgery are needed to aid optimal systemic therapy. We aimed at investigating the feasibility of using circulating tumor DNA (ctDNA) to early detect MRD in plasma samples collected during the follow-up of patients (pts) with early BC.

Methods

Forty BC pts undergoing surgery with curative intent and regular follow-up for a minimum of 13 years were included in a 1:3 case-control study. Preliminary experiments demonstrated that heparinase I digestion does not affect the quality of DNA extracted from heparin-collected blood and that pre-amplification overcomes limitations due to small (<0.5 ml) plasma aliquots. Mutational analysis of archival BC tissues was performed by Ion AmpliSeq^TM targeted sequencing and identified Single Nucleotide Variations (SNV) were validated and tracked in plasma by digital Polymerase Chain Reaction.

Results

One or more circulating SNVs were identified in 27/40 cases prior to surgery. During follow-up (range 80-200 months), 6 pts relapsed locally, 4 at distant sites, and 17 remained disease-free (DF). ctDNA was undetectable in 16/17 DF cases, whilst it was detectable in 9/10 recurrent pts, and anticipated overt metastases/loco-regional recurrences with a median lead time of 20 (range 10-47) months. Based on these results, we started a prospective study on ctDNA tracking in triple negative (TN) BC pts. At the time of writing, 67 pts with early stage TNBC have been enrolled in a longitudinal trial and followed up (range 10-74 months). Before surgery, ctDNA was detectable in 64% of cases and its clearance was associated with DF status. In additional 12 TNBC pts, ctDNA was analyzed prior and during neoadiuvant chemotherapy (NAC): ctDNA was detectable in 81% of the cases at baseline, and its dynamics during and after NAC reflected tumor response, and anticipated overt metastases with a lead time up to 13 months.

Conclusions

Post-surgical ctDNA can anticipate the diagnosis of new disease manifestations, including loco-regional recurrences, which are amenable of treatment with a curative intent. Additional results on the longitudinal trial currently ongoing in TNBC pts will be presented at the meeting.

Legal entity responsible for the study

Maria Grazia Daidone, PhD.

Funding

Associazione Italiana per la Ricerca Contro il Cancro (AIRC).

Disclosure

All authors have declared no conflicts of interest.

Background

Half of the distant recurrences (DR) occur after the first five years from diagnosis. The Clinical Treatment Score post-5 years (CTS5) intends to predict DR after 5 years of endocrine therapy (ET). This study uses the IDEAL study cohort, consisting of patients disease free after 5 years of standard adjuvant ET and randomized between 2.5 or 5 years of extended therapy with letrozole, to validate the CTS5 as a prognostic tool for late DR and, if it is validated, to examine it as a predictive tool for choosing between 2.5 years or 5 years of extended ET.

Methods

The CTS5 categorizes patients that had been disease free for 5 years, into low (<5%), intermediate (5-10%) and high (>10%) risk groups for developing DR between 5 and 10 years after diagnosis. To validate the CTS5, both the discrimination and calibration are examined.

Results

In this analysis, 1591 patients from the IDEAL study cohort were included. Overall, they were younger, had more lymph node involvement, higher grade tumours and received more chemotherapy than patients in the ATAC and 1BIG 1-98 cohorts used to train and test the CTS5. Consequently, more patients were allocated to the high risk group than in the training and testing cohort. For the prognostic value of the CTS5 score, we showed that there were significantly more DR between 5 and 10 years after diagnosis in the high risk group (HR 4.9, 95% CI 2.3-10.1) and the intermediate risk group (HR 2.2, 95% CI 1.03-4.8) compared to the low risk group (log-rank p < 0.001). The expected risk scores for the low, intermediate and high risk groups, calculated using the CTS5 scores, were 3.6%, 7.2% and 19.4%, respectively. The observed late DR using K-M curves were 1.4% (95% CI 0.7-2.1), 5.6% (4.7-6.5) and 10.6% (9.4-11.8) for the low, intermediate and high risk groups, respectively. The CTS5 risk score systematically overestimates the risk of late DR. Since the algorithm could not be validated, it was not deemed relevant to examine its predictive abilities for extended ET within the risk groups.

Conclusions

The CTS5 is able to significantly discriminate patients into three different risk categories. However, the calibration could not be validated, as the absolute risk for DR was significantly lower than the expected DR. The CTS5 should therefore not be used as prognostic tool for distant breast cancer recurrences.

Clinical trial identification

BOOG 2006-05.

Legal entity responsible for the study

CJH van de Velde.

Funding

The original IDEAL trial was supported by Novartis by means of an independent educational grant (CFEM345DNL03). This IDEAL substudy received no additional funding.

Disclosure

All authors have declared no conflicts of interest.

Background

TAM causes several side effects that impact the quality of life and adherence to treatment in patients with BC. Due to the complexity of TAM metabolism, serum biomarkers predicting TAM side effects have not been identified. We hypothesized that TAM levels may be associated with side effects, thus identifying pts eligible to tailored supportive interventions.

Methods

CANTO cohort (NCT01993498) is a French multicenter prospective longitudinal study including 12000 women with stage I-III BC aimed at characterizing long-term toxicities of BC treatment. CANTO COMPLETE is a pre-defined sub study of adherence to endocrine therapy evaluating TAM serum concentrations in premenopausal pts. Here, we analyzed the association between continuous TAM serum levels and side effects at 1 year of TAM initiation in pts with serum defined adherence to TAM (> 60 ng/mL). TAM level was determined by liquid chromatography tandem mass spectrometry on 200 µL of serum. Selected TAM-related toxicities were assessed by increase in body weight ≥5%, CTCAE v4.0 and patient reported outcomes (EORTC C30 and hospital anxiety and depression scale [HADS]). Univariate and multivariate linear and logistic regressions were fitted to measure the association between TAM levels and selected toxicities.

Results

A cohort of 989 pts with serum defined adherence to TAM was identified. Median age at diagnosis was 45 yrs (min.-max. 22 - 57), 43.7% were stage I, 64.6% received (neo)adjuvant chemotherapy, 89.4% had Charlson score 0, 33.8% had BMI≥25 kg/m², 20.9% were active smokers and 52.4% had education ≥ college. Median serum TAM was 119 ng/ml (P25-P75 96-152). In univariate and multivariate analyses, TAM levels were not associated with worse side effects, in specific: weight gain; gynecological, musculoskeletal or neurologic toxicities (including cognitive); fatigue or insomnia; and anxiety or depression.

Conclusions

In premenopausal women adherent to adjuvant TAM, TAM serum levels were not associated with TAM-related side effects at 1 year post treatment initiation. As next steps, we will analyze the association of tamoxifen metabolites and side effects.

Clinical trial identification

NCT01993498.

Legal entity responsible for the study

INSERM UMR 981.

Funding

Institut National du Cancer (to Fabrice André and Barbara Pistilli), Susan G. Komen (to Ines Vaz-Luis) and Fondation ARC (to Ines Vaz-Luis and Gwenn Menvielle).

Disclosure

All authors have declared no conflicts of interest.

Background

Dual HER2 blockade with trastuzumab (T) and lapatinib (L) is approved for patients (pts) with T-resistant HER2+ metastatic breast cancer, although little is known regarding the cardiotoxicity of this combination. Thus, we report cardiac data on 4,190 pts treated with 1 year of adjuvant T or concomitant T+L in ALTTO (BIG 206/N063D/EGF106708) trial.

Methods

ALTTO randomized 8,381 pts into 4 arms to investigate the benefit of L and T in early HER2+ breast cancer. Our dataset consists of pts randomized to the T arm and concomitant T+L arm. Eligible pts must have had a baseline left ventricular ejection fraction (LVEF) ≥50%, no serious cardiac illness and cumulative doses of doxorubicin ≤360mg/m² or epirubicin ≤720mg/m². Signs/symptoms of heart failure and LVEF were assessed tri-monthly during treatment, every 6 months until year 2, and yearly thereafter until year 10. Cardiac events (CEs) were defined as symptomatic heart failure NYHA class II, III and IV associated with a significant LVEF drop; asymptomatic CE (significant drop in LVEF without symptoms); and cardiac death. Acute recovery was defined as ≥ 2 consecutive LVEF assessments of ≥ 50% after a CE. The distribution of CEs and end-points are described by arm. A logistic regression model by arm was used to assess the odds of CEs and risk factors for its occurrence.

Results

Pts characteristics were equally distributed among the 2 arms, except for diabetes (more common in T-arm; p = 0.024). With a median follow-up of 6.5 years (range 5.6-7.1 years), 197 (9.3%) CEs occurred in T vs 166 (7.9%) CEs in T+L arm. Median time to develop a CE was 6.4 months (range 3.6-11.7 months) in T vs 7.1 months (range 2.9-16.6 months) in T+L arm. Most CEs occurred during treatment (73.2%) and were asymptomatic (74%). Acute recovery was reached in 83.6% and 84.1% of pts in T and T+L arms, respectively. Time to recover from symptomatic CEs (T: 5.6 months; T+L: 4.2 months) was longer than for asymptomatic CEs (T: 3.1 months; T+L: 2.9 months). A 2^nd LVEF drop to < 50% occurred in 29.9% of pts who recovered from a CE. T was completed by 84% and 82% of pts in T and T+L arms, respectively, while L was completed by 68% of pts in T+L arm. The main reason for discontinuation of L was safety (60%), especially non-cardiac safety (82%). Five cardiac risk factors were identified (see Table).

Table: 199O

Conclusions

Although dual HER2 blockade does not increase the rate of cardiotoxicity compared to T alone for 1 year, identification of risk factors prior to start of therapy and close collaboration with cardiologists is essential to ensure its proper management and treatment continuity.

Clinical trial identification

NCT00490139, EudraCT Number: 2006-000562-36.

Legal entity responsible for the study

Breast International Group (BIG).

Funding

GSK/Novartis provided funding for the main trial.

Disclosure

D. Eiger: Scholarship via ESMO`s clinical research fellowship program, made possible thanks to a sponsorship provided by Novartis. M. Piccart: Honoraria: Roche; Research funding to IJB: Roche, Novartis. E. de Azambuja: Honoraria, advisory board: Roche/GNE; Travel grants: Roche/GNE, GSK/Novartis; Research grant (my institution): Roche/GNE, Novartis, AstraZeneca. All other authors have declared no conflicts of interest.

Background

Inflammatory breast cancer (IBC) is an aggressive form of breast cancer with elevated metastatic potential. In the past, we have identified a gene expression profile that characterizes IBC, suggesting that a specific molecular biology underpins this devastating disease. Here, we explore the genomic alterations underlying IBC.

Methods

Mutation and copy number variation (CNV) profiles for 756 genes were assembled from 2.920 primary tumor samples (subtype distribution: 63% HR+, 18% HER2+, and 19% TNBC) including 101 profiles from patients with IBC before therapy and 468 profiles from metastatic breast cancer samples. Differences in the frequency of genomic aberrations between patients with and without IBC, stratified per subtype, were investigated. Genomic perturbation differences for pathways and mutational signature (MS) profiles were compared between patients with and without IBC.

Results

Seventy-six genes showed evidence of extensive genomic alterations in samples from patients with IBC as compared to those without IBC (i.e. false discovery rate < 10%), whereas only 3 genes reveal the opposite pattern. The 10 top scoring genes according to the odds ratio include: MYC, CYP2D6, VEGFA, CCND1, GNAQ, ZNF703, PTPN11, MCL1, CDK4, and CCDC6. Analysis of MS profiles revealed differences for signature 2, -11, -20, -23 and -24. When comparing to metastatic breast cancer samples, mutations in genes involved in homologuous recombination and in the NOTCH pathway were more prevalent in IBC.

Conclusions

These data suggest that IBC is characterized by an extensive mutational burden that results, amongst others in the activation of NOTCH signaling as well as deficient homologuous recombination. The analysis of MS profiles identifies 3 mutational processes in IBC that are associated with transcriptional strand-bias for mutations involving a cytosine (i.e. C>T and C>A), which is indicative for transcription coupled nucleotide excision repair and suggests that mutations involve the complementary guanine base.

Legal entity responsible for the study

University Antwerp.

Funding

Has not received any funding.

Disclosure

All authors have declared no conflicts of interest.

Background

We previously identified a VEGFA metagene in TNBC which is associated with poor prognosis and eliminates the good prognostic effect of TILs (PMID 21978456, 28750120). But high pretreatment values of this signature predicted pCR in the neoadjuvant GeparQuinto trial (OR 2.40, P = 0.006) showing an interaction with bevacizumab treatment (P = 0.020). The main cellular sources of the transcripts that comprise the VEGFA metagene are unknown since mRNA profiling of bulk biopsies contains signals from different cell types. The individual genes have been reported to be expressed both in endothelial cells, fibroblasts, and epithelial cancer cells.

Methods

We analyzed single-cell sequencing RNA-Seq (sc-RNA-Seq) data of n = 772 single cells from TNBC and 52,698 single cells from lung cancer and non-malignant lung to identify the cellular source of the metagene expression. We then established and validated single marker IHC assays for the VEGFA metagene and applied them to tissue micro arrays (TMA).

Results

sc-RNA-Seq from TNBC revealed co-expression of core genes of the VEGFA metagene (NDRG1, VEGFA, and DDIT4) in different cell types with highest expression in basal-like cancer cells. Strong co-expression in a subset of carcinoma cells was also seen in lung cancer sc-RNA-Seq, but only limited expression in endothelial cells. An IHC assay revealed strong para-necrotic expression in line with functions in cellular stress and hypoxia. Whole-slide-IHC showed good correlation to mRNA expression even in core biospies (Kappa=0.864, P < 0.001). However, comparison of whole-slide and TMA demonstrated a profound loss of sensitivity, while specificity remained high. Loss of sensitivity was mainly due to heterogeneity in expression and the small sized TMA cores from pretreatment biopsies. Nevertheless, still a predictive value for pCR was detected in the TMA dataset (P = 0.025).

Conclusions

The cellular source of the VEGFA metagene are peri-necrotic carcinoma cells. Thus, necrosis detection may help in response prediction and as stratification factor in trials. Since the signature indicates an immunosuppressive environment it should also be further studied in the context of combination therapies of anti-angiogenic and immune treatments.

Legal entity responsible for the study

University of Frankfurt.

Funding

H.W. & J. Hector-Stiftung, Mannheim, Germany.

Disclosure

All authors have declared no conflicts of interest.