What molecular signatures distinguish low‑PSA, high‑grade prostate cancers from PSA‑secreting tumors?

1. AR suppression and loss of PSA as a defining axis

At the core of the low‑PSA, high‑grade phenotype is dissociation of canonical AR signaling from tumor growth: tumor cells either downregulate AR and its transcriptional outputs (including KLK3/PSA) or become functionally AR‑indifferent, producing little secreted PSA despite aggressive histology; this AR/PSA heterogeneity has been documented in primary and metastatic lesions and correlates with therapeutic resistance ^{[2] [6]}. Transcriptomic profiling in The Cancer Genome Atlas and cell‑line models shows that very low PSA tumors cluster with mCRPC signatures, implying that low PSA is a surrogate for reduced AR output and lineage plasticity rather than simply low tumor burden ^{[1] [2]}.

2. Transcriptomic signatures: enrichment for mCRPC and neuroendocrine programs

RNA‑level analyses reveal that genes upregulated in very low‑PSA, high‑risk cases are significantly enriched for mCRPC gene sets and treatment‑emergent small‑cell/neuroendocrine prostate cancer (t‑SCNC/NEPC) signatures, indicating transdifferentiation toward neuroendocrine‑like states or other de‑differentiated lineages that do not express PSA ^{[1] [3]}. Experimental isolation of PSA‑neg/low populations from LNCaP and LAPC9 lines further showed stem‑like, tumor‑propagating features and gene expression profiles associated with castration resistance, supporting the clinical transcriptomic observations ^[2].

3. Recurrent genomic alterations associated with the phenotype

Genomic studies highlight frequent alterations that favor lineage plasticity and aggressive behavior in low‑PSA cases: focal PTEN loss and dysregulated PI3K signaling are common in aggressive prostate cancers and have been implicated in poor outcomes and AR pathway interplay, while targeted sequencing of low‑PSA mCRPC cohorts demonstrates variants and signature scores consistent with t‑SCNC biology ^{[7] [3]}. While specific universal driver mutations unique to low‑PSA, high‑grade tumors have not been exhaustively cataloged in the cited reports, the existing data links canonical poor‑prognosis alterations (e.g., PTEN loss) to this clinical subgroup ^{[7] [3]}.

4. Clinical correlations: prognosis, metastasis and therapeutic resistance

Multiple large clinical cohorts and registry analyses report that men with high‑grade tumors and very low presenting PSA have higher prostate‑cancer specific mortality and biochemical recurrence and are likelier to harbor nodal or metastatic spread—findings that support a biologically more aggressive, de‑differentiated disease state with reduced PSA secretion ^{[5] [4] [1]}. Practical consequences include poorer responses to conventional androgen‑deprivation and AR pathway inhibitors and the need to consider alternative diagnostic and therapeutic strategies for these patients ^{[1] [3]}.

5. Diagnostic and therapeutic implications: PSMA, imaging and biomarker strategy

Despite low PSA secretion, many aggressive tumors upregulate non‑secreted targets such as PSMA and show distinct cell‑free or tissue genomic signatures that can be exploited diagnostically and therapeutically: PSMA expression increases with higher grade and castration resistance and supports PSMA‑PET imaging and targeted therapies even when serum PSA is low, while liquid biopsy and CTC/exosome profiling may capture AR‑independent genomic alterations ^{[8] [9] [10]}. These avenues are promising but require standardization; relying on serum PSA alone risks underestimating disease aggressiveness in this subset ^{[11] [10]}.

6. Caveats, unanswered questions and need for prospective validation

The literature consistently warns that definitions of “low‑PSA” vary across studies, that many mechanistic inferences derive from mCRPC cohorts or cell lines rather than well‑annotated, treatment‑naïve primary tumors, and that prospective validation of candidate genomic/transcriptomic classifiers is still lacking—therefore, while AR suppression, mCRPC/NEPC transcriptional enrichment, and recurrent alterations like PTEN loss form a coherent molecular picture, the field requires standardized biomarkers and longitudinal cohorts to translate signatures into clinical decision tools ^{[3] [2] [11]}.