Prostate‐specific membrane antigen ( <scp>PSMA)</scp> ‐ <scp>PET</scp> ‐avid prostate cancers show stronger <scp>PSMA</scp> immunohistochemistry staining on biopsy cores

Abstract

Prostate-specific membrane antigen (PSMA) is a type II transmembrane glycoprotein, which is heavily expressed in most prostate cancers. Immunohistochemistry (IHC) staining for PSMA is a long-established tool for pathologists in identifying prostate cancer, as well as a target for molecular imaging with PSMA-positron emission tomography (PET). This has generated considerable research interest in examining whether PSMA molecular expression as assessed with IHC is associated with clinical outcomes, or imaging findings. This is a pertinent clinical issue, as the adoption of PSMA-PET as the ‘gold standard’ for staging newly-diagnosed high-risk prostate cancer comes with a significant financial burden, with costs estimated at $1203 Australian dollars (AUD) per scan [1]. Given that an estimated 4–10% of patients will not have a PSMA-avid lesion within the prostate [2], and its low sensitivity for detecting nodal disease in intermediate-risk patents [3], a significant number of resources are used on negative scans. PSMA IHC staining could therefore be a potential tool for predicting the presence of PSMA-PET-avid disease. It is widely available to pathologists and has a low cost per patient ($30 AUD at our institution). To date, a number of studies have assessed PSMA IHC on and whole-mount radical prostatectomy specimens and consistently found associations between PSMA expression and PSMA-PET findings [4, 5]. However, the correlation between IHC staining on biopsy cores and imaging findings is not as established, with few studies assessing this relationship [6, 7]. We therefore undertook a retrospective audit of a prospectively collected database of our own biopsies to assess for a relationship between biopsy IHC and PSMA-PET findings. All transperineal biopsies conducted by a single surgeon (N.L.) between November 2021 to July 2025 were scrutinised. A total of 222 transperineal prostate biopsies were taken in 197 individual patients. All patients had a template biopsy, with additional cognitive-fusion targeting of any Prostate Imaging-Reporting and Data System (PI-RADS) score ≥3 lesions, and/or any areas of focal avidity if a PSMA-PET was performed prior to initial biopsy. PSMA IHC was performed on patients with newly diagnosed prostate cancer with International Society of Urological Pathology Grade Group (ISUP GG) ≥2 or with ISUP GG 1 but high-risk features (three or more cores of ≥5 mm). PSMA-PET was routinely performed on all men with ISUP GG ≥2. In total, 74 men had PSMA IHC done on their biopsy specimens. IHC staining was performed on formalin-fixed, paraffin-embedded tissue, cut at 4 μm thickness. PSMA (clone ID6; NovaCastra/Leica, Nussloch, Germany) was used at 1:50 dilution on a Benchmark Ultra autostainer (Ventana Medical Systems, Inc., Tucson, AZ, USA) with OptiView diaminobenzidine (DAB) detection. All slides were reviewed by an experienced uropathologist (A.R.) along with at least one other pathologist and assigned an ISUP GG. PSMA IHC was performed after the main clinical review and interpreted by a single pathologist (A.R.). The core with the highest ISUP GG was utilised for IHC staining, as it was assumed to be most representative of the tumour. If multiple cores had the highest ISUP GG, the IHC staining was performed on the core with highest volume of tumour. Slides were assigned a visual IHC score from 1 to 3+ based on predominant PSMA expression patterns as quantified using a standard four-tiered visual scoring system (0 = negative, 1 + = weak, 2 + = moderate, 3 + = strong) [6] (Fig. 1). The PSMA-staining pattern was described as either luminal membrane, cytoplasmic, or both. A biopsy core was considered PSMA-PET concordant when there was a PSMA-PET-avid lesion (PRIMARY score ≥3) in the same anatomical location where clinically significant prostate cancer (ISUP GG ≥2) was found on a biopsy core. Patients with and without concordant lesions had their data compared, using the Mann–Whitney U tests. These tests were sele