High stromal hyaluronan level is associated with poor differentiation and metastasis in prostate cancer

Several epithelial tumours accumulate hyaluronan (HA) which promotes cancer cell invasion and metastasis. We analysed the expression of HA and its receptor CD44 and their prognostic value in 166 prostate cancer patients followed up for a mean of 13 years; standard deviation (S.D.) 2.7; range 8.7-21.4 years. HA was detected with a specific biotinylated probe prepared from cartilage aggrecan and link protein, and CD44 with an antibody recognising all forms of CD44. The peri- and intratumoral stroma from half of the patients strongly expressed immunohistochemically detectable HA in < or = 15% of the stromal area; the tumours in the remaining half expressed HA in > 15% of the area. The staining of cancer cells for HA was scored positive or negative, and for CD44 the median value of 80% of positive tumour cells was used as a cut-off point. The expression of HA in cancer cells was weakly associated with perineural infiltration of the tumour (P = 0.03) and high Gleason score (P = 0.002). There was also a significant inverse relationship between the expression of HA and CD44 in cancer cells (P < 0.001). The high level of HA in the peri-and intratumoral stroma was related to metastasis, high T-category, high Gleason score, perineural infiltration and high mitotic activity of the tumour (for all P < 0.001). There was a significant inverse relationship between the expression of CD44 in cancer cells and high level of strong expression of HA in the tumour stroma (P < 0.001). A low fraction of CD44-positive cells was related to a high TM-category, high Gleason score and rapid cell proliferation (for all P < 0.0001; M/V P value = 0.0013). In the univariate survival analysis, the high level of strong expression of HA in tumour stroma predicted an unfavourable outcome in the entire series (P = 0.003) and also in the M0 tumours (P = 0.07), while in T1-2 M0 tumours the prognostic value did not reach the level of statistical significance (P = 0.1). A low fraction of CD44-positive cells predicted a poor outcome in the entire series (P < 0.001) and also in M0 tumours (P = 0.003). Cancer cell-associated HA expression had no prognostic value in any tumour categories. In the multivariate analysis of prognostic factors, HA expression in the cancer cells or in the tumour stroma had no additional value to the standard prognostic factors TM-classification, Gleason score and CD44 expression. Our results show that stromal HA accumulation is related to several malignant features and adverse clinical outcome in prostate cancer. However, further studies based on uniformly treated patient cohorts are needed to establish the clinical significance of these findings in current clinical practice.