Several reports have suggested that one or both of the trinucleotide repeat polymorphisms in the human androgen receptor (hAR) gene, (CAG)n coding for polyglutamine and (GGC)ncoding for polyglycine, may be associated with prostate cancer risk; but no study has investigated their association with disease progression. We present here a study of both hAR trinucleotide repeat polymorphisms not only as they relate to the initial diagnosis but also as they are associated with disease progression after therapy. Lymphocyte DNA samples from 178 British Caucasian prostate cancer patients and 195 control individuals were genotyped by PCR for the (CAG)n and (GGC)n polymorphisms in hAR. Univariate Cox proportional hazard analysis indicated that stage, grade and GGC repeat length were individually significant factors associated with disease‐free survival (DFS) and overall survival (OS). The relative risk (RR) of relapse for men with more than 16 GGC repeats was 1.74 (95% CI 1.08–2.79) and of dying from any cause, 1.98 (1.13–3.45). Adjusting for stage and grade, GGC effects remained but were not significant (RRDFS= 1.60, p = 0.052; RROS= 1.65, p = 0.088). The greatest effects were in stage T1‐T2 (RRDFS= 3.56, 95% CI 1.13–11.21) and grade 1 (RRDFS= 6.47, 95% CI 0.57–72.8) tumours. No differences between patient and control allele distributions were found by odds‐ratio analysis, nor were trends with stage or grade evident in the proportion of short CAG alleles. Non‐significant trends with stage and grade were found in the proportion of short GGC alleles. The (GGC)n polymorphism in this population is a significant predictor of disease outcome. Since the (GGC)n effect is strongest in early‐stage tumours, this marker may help forecast aggressive behaviour and could be used to identify those patients meriting more radical treatment. Int. J. Cancer (Pred. Oncol.) 84:458–465, 1999. © 1999 Wiley‐Liss, Inc.

Prostate cancer shows evidence of familial aggregation, particularly at young ages at diagnosis, but the inherited basis of familial prostate cancer is poorly understood. Smith et al. recently found evidence of linkage to markers on 1q, at a locus designated "HPC1," in 91 families with multiple cases of early-onset prostate cancer. Using both parametric and nonparametric methods, we attempted to confirm this finding, in 60 affected related pairs and in 76 families with three or more cases of prostate cancer, but we found no significant evidence of linkage. The estimated proportion of linked families, under a standard autosomal dominant model, was 4%, with an upper 95% confidence limit of 31%. We conclude that the HPC1 locus is responsible for only a minority of familial prostate cancer cases and that it is likely to be most important in families with at least four cases of the disease.

Genetic susceptibility to prostate cancer is likely to be controlled by many loci, with no single gene explaining a large fraction of the familial risk. Pooling of results from all available genome scans is likely to be required to obtain definitive linkage results.

Estimating the penetrance of a specific mutation using different approaches, we found that both the choice of study population and statistical method affect the magnitude of the estimates.