Abstract
To determine whether very small prostate cancers present in patients who also have lymph node (LN) metastases represent a particularly aggressive disease variant compared with larger LN-positive tumours.
We identified 37 501 patients diagnosed with prostate cancer between 1988 and 2001 treated with radical prostatectomy within the Surveillance, Epidemiology, and End Results database. The primary study variables were tumour size by largest dimension (stratified into: (i) microscopic focus only or 1 mm; (ii) 2-15 mm; (iii) 16-30 mm; (iv) >30 mm), regional LN involvement, and the corresponding interaction term. We evaluated the risk of 10-year prostate cancer-specific mortality (PCSM) using the Fine and Gray model for competing risks after controlling for race, tumour grade, T stage, receipt of radiation, number of dissected LNs, number of positive LNs, year of diagnosis, and age at diagnosis.
The median follow-up was 11.8 years. There was a significant interaction between tumour size and LN involvement (P-interaction <0.001). In the absence of LN involvement (36 561 patients), the risk of 10-year PCSM increased monotonically with increasing tumour size. Among patients with LN involvement (940), those with the smallest tumours had increased 10-year PCSM compared with patients with tumours sized 2-15 mm (24.7% vs 11.8%; adjusted hazard ratio [AHR] 2.84, 95% confidence interval [CI] 1.21-6.71; P = 0.017) or 16-30 mm (24.7% vs 15.5%; AHR 3.12, 95% CI 1.51-6.49; P = 0.002), and similar 10-year PCSM as those with tumours >30 mm (24.7% vs 24.9%; P = 0.156).
In patients with prostate cancer with LN involvement, very small tumour size may predict for higher PCSM compared with some larger tumours, even after controlling for other prognostic variables. These tumours might be particularly aggressive, beyond what is captured by pathological assessment of tumour grade and stage.
