Patients were randomly selected from a retrospective database of male and female patients over 18 years of age who underwent surgery for breast cancer at our institution between January 2012 and December 2017 and for whom a specimen is available in our tumour library. Each patient received comprehensive information about the study’s objectives, procedures, and potential implications, and their consent was obtained before their tissue samples were used for analysis. The procedure has been declared to the French National Institute for Health Data (N° F20220615153900).

Immunodetection of PSMA was performed to assess its cellular and neovascular expression. Immunohistochemistry was performed on paraffin embedded tumour tissues using a Ventana Discovery XT autostainer on 4 μm-thick sections. Slides were deparaffinised with EZPrep buffer at 75 °C for 8 min, and epitopes were unmasked at 95 °C for 56 min in CC1 buffer. Sections were incubated 40 min at 37 °C with PSMA antibody (ab133579, Abcam, 1/1000) or CD31 antibody (ab28364, Abcam, 1/50). Secondary antibody (Omnimap Rabbit) was incubated for 16 min at 37 °C. After washes, staining was performed with 3,3′-diaminobenzidine (DAB), and sections were counterstained with hematoxylin and bluing reagent. Whole slide images were digitized at 20 × (0,5 μm/pixel) using the VS120 scanner (Olympus). The slides were controlled by a certified pathologist. They were recorded as tiled tiff images.

The digitized slides were reviewed and analysed by an experienced pathologist. The number of positive cells per field at 20-fold magnification was used to evaluate PSMA expression. The quantification of expression was performed on the entire slide. This measurement represents an average value, reflecting the overall expression across the entire tissue sample. CD31 immunodetection was used to specifically stain endothelial cells. This marker allowed the pathologist to differentiate between tumoural PSMA expression and PSMA expression in the neovasculature. By highlighting the blood vessels, CD31 staining enabled the identification and quantification of vessels expressing PSMA, thereby ensuring assessment of neovascular PSMA expression. Tumours with no detectable PSMA expression were scored as “0,” tumours with 1 to 5 positive cells per field were scored as “1,” and tumours with more than 5 positive cells per field were scored as “2”. For tumours scored as “2,” the corresponding TIFF images were processed to calculate the percentage of positive neo-vessels: \(\:\frac\times\:100\).

Genetic status was determined by germline sequencing. These analyses were conducted using DNA extracted from a blood sample at the Biology and Genetics Laboratory of the François Baclesse Centre. Next generation sequencing (NGS) was performed using specific genetic analyses on the Illumina platform (mainly by Nextseq500-MiSeq or GAIIx). The bioinformatics pipeline included CASAVA, NextGENe, CNVseq and Alamut-HT. Patients received genetic counselling both before and after genetic testing during a face-to-face consultation.

Graph analysis and statistical analysis were performed on XLSTAT software (XLSTAT: Data analysis and statistical solutions for Microsoft Excel. Addinsoft (2017)). Clinical characteristics were compared according to PSMA expression (“0” vs. “1 or 2”) using the Fisher’s exact test. Variables associated with PSMA expression at an alpha level of 10% were entered into a logistic regression model for stepwise selection. The model maximizing Akaike’s criterion was retained. For all statistical tests, a two-tailed p value of less than 0.05 was considered statistically significant.