Prognostic factors for carcinoma of the prostate

Prognostic factors for prostatic carcinoma should be significant, independe nt and clinically important. They should be of practical use, and their det ermination should be affordable in everday practice. Prognostic factors may be grouped into patient-related, tumor-related a nd treatment-related. The y should meet certain requirements, such as possession of a clear biologica l significance, an adequate sample size (possibly more than 150 patients), no patient population bias, an adequate statistical test, such as Cox regre ssion analysis, as well as optimized cut-off values and reproducibility. Fr om a pathologist's view, prognostic factors with established values are gra de, margin involvement, capsular penetration, seminal vesical involvement, metastases and invasion of fat in needle biopsies. In contrast to this, fac tors with little value are, among others, zone location or nuclear shape. I f these guidelines for assessment of prognostic factors are not met, the pr ognostic factors grow exponentially, as an individual patient can only belo ng to one prognostic group. If one considers all th ree categories of progn ostic factors together, the clinical stage matters most despite all uncerta inties. The same holds true for grading; particularly, the well-differentia ted grades on biopsy cores have the drawback of being reflected in the spec imen only infrequently. The use of biomarkers to give a better prognostic i nformation is also disappointing, as only PSA and PAP have a reliable value among 28 biomarkers. It is of note that new biomarkers are continuously be ing discovered and examined, such as cyclin A or D. Due to these deficienci es in all three categories of prognostic factors for prostatic carcinoma, p rognostic indices in the form of nomograms were constructed. But, if these indices are employed to answer the most important question at the time of d iagnosis, i.e., 'is this man a candidate for surveillance?', neoadjuvant tr eatment plus irradiation, neoadjuvant treatment plus radical prostatectomy, perineal radical prostatectomy, because of a low probability of extracapsu lar extension or positive lymph nodes, adjuvant therapy after local treatme nt with curative intent as opposed to progression-based treatment or immedi ate systemic treatment, let alone intermittent endocrine manipulation, are not reliably possible. The outcomes of the few available studies based on p rognostic factors should be studied carefully. If considered, a valuable ne w way of estimating artificial neural networks is a possibility to come to practical terms. Copyright (C) 1999 S. Karger AG, Basel.