In this study we have simultaneously examined the immunohistochemical expression of p16 and p27 in RP specimens of hormone naïve PC patients and reported the absence of a prognostic role for p16 and p27 in primary PC, as we observed no correlations with most important, firmly established clinico-pathological parameters, including pre-operative PSA, tumor stage and grade, even at the univariate level. In multivariate analysis including p16, p27, pre-operative PSA level, tumor stage and Gleason grade, only pT stage retained its importance in predicting PSA recurrence after RP. In fact, the risk of PSA relapse was approximately 3.3 fold greater in patients with advanced pT stage than in men with early pT stage disease.
Both histological markers have been previously studied in early PC, often with conflicting results regarding their putative prognostic relevance. An initial immunohistochemical study of p16 expression in 88 early PC patients reported a significant univariate association of p16 with a higher pre-treatment PSA level and a sooner time to PSA relapse after RP , although p16 was not associated with Gleason grade or stage. Unlike our study population, this cohort followed a higher threshold of detection (> 5%) for characterization of p16 nuclear staining in tumor cells. Also, a significant portion of the cohort (34 patients or 39%) was treated with neo-adjuvant androgen ablation, which seemed to enhance p16 expression. In multivariate analysis adjusted for tumor grade, pre-treatment PSA, and stage, overexpression of p16 did not contribute prognostic information .
In contrast, in another study including 104 patients, high level of p16 protein expression, quantitated by immunofluorescence flow cytometry, was an independent predictor of BF, although no significant association was found between p16 and standard clinico-pathological variables including serum pre-treatment PSA . This finding was not sufficiently explicable even by the authors themselves as they observed no statistically significant association between p16 expression and BF in the same cohort when classic immunohistochemistry was used as a method of assessment .
A larger study of 206 patients with clinically localized PC evaluated p16 immunohistochemistry in areas of high-grade intraepithelial neoplasia (HGPIN) and of cancer in the same specimen. In the cancers, p16 overexpression, defined as either > 1% or > 5% nuclear staining, was associated with stage and disease relapse but did not correlate with age, pre-treatment PSA concentration, or Gleason score. In a multivariate model, overexpression of p16INK4A in HGPIN was an independent predictor of disease relapse. Although 38 of the 206 patients in this group received neo-adjuvant hormone therapy (NHT) prior to surgery, this did not seem to influence the prognostic value of p16 expression when they were excluded from the analyses . In our study we did not concurrently evaluate the p16 staining status of HGPIN in our tumor samples, thus our results are not directly comparable with these of Henshall et al. Even in this case, they do not mention whether p16 expression in areas of cancer alone independently correlated with relapse-free survival.
Numerous studies have examined the prognostic significance of p27 immunohistochemical expression in RPs, with both negative and positive results. In an early work, low p27 expression, classified as < 50% of cells p27 positive, correlated with a number of prognostic morphologic features (including Gleason score, positive surgical margins, seminal vesicle involvement, lymph node metastasis and tumor aneuploidy) but did not correlate with sub-clinical biochemical failure, concurring with our results . Similarly, Erdamar et al. did not find any association between the mean labeling index (LI) of p27Kip1 expression in cancers (LI: 43.5 +/-3.7%, defined as the percentage of p27-positive cells among epithelia of the same category) and Gleason score, stage or disease progression after RP . In another study, decreased p27Kip1 staining (defined as < 25% of nuclei stained positive for p27Kip1) correlated with higher Gleason grade and was an independent predictor of treatment failure in the node-negative cohort. However, p27 was not an independent prognostic factor when 24 of 113 patients who underwent pre-operative NHT were excluded from the analysis . Yang et al. found that absence of p27kip1 expression was the strongest predictor of biochemical relapse in patients with clinically organ confined disease . In another cohort, < 10% reactivity for p27Kip1 proved to be the only independent prognostic factor for the PSA recurrence-free survival of 95 and 86 patients respectively [25, 26]. These studies do not directly contradict our results, given the differences in cut-off values examined for positivity. A negative staining reaction as a predictor of recurrence-free survival did not achieve statistical significance at other cut-off values (</≥ 40 and </≥ 60% positivity) calculated, which were much closer to the one used in our study .
Cote et al. demonstrated the prognostic value of decreased p27Kip1 expression (cut-off value of 10%) for both the recurrence-free and overall survival of 96 PC patients undergoing RP. However, only stage C patients were evaluated .
At a cut-off level much lower than ours (30% positive cells), patients with low p27 expression showed a higher risk of biochemical relapse than the others, which maintained its predictive value in a multivariate analysis along with stage. However, only a small number of patients (47) with available follow up were examined .
Expression of p27 below a median value of 64% in tissue microarrays of 104 patients treated with RP was associated with high stage, elevated pre-operative PSA and time to biochemical failure. However, neither p27 alone nor combined PTEN/p27 expression retain their significance in multivariate analysis . An artifact of loss of p27 expression during tissue processing in at least part of the samples of this study remains a possibility and might explain the different results supported by our and other groups using individual sample slides.
A negative association of p27 (Kip1) expression with tumor stage and grade has been reported in a group of 30 PC patients undergoing transurethral resection of the prostate (TURP) or RP. However, differences in sample size (only 30 PCs), surgical technique (80% perineal prostatectomy), immunohistochemical evaluation (whole tissue area), follow up (only 21 months) and cases of biochemical recurrence (only 7%) do not allow direct comparison between this study and ours . Likewise, the reported high frequency of cytoplasmic p27 localization in high grade tumors compared to low-grade ones  cannot be confirmed by our results, as we performed only nuclear p27 IHC assessment, consistent with the majority of previous studies.
In concordance with our results, lack of association between expression of p27 or its ubiquitin ligase, Skp2, and time to PSA recurrence was reported in a prospective study of 162 African-American patients with clinically localized PC with a cut point for p27 set at < 40% . The most recent study of p27 expression after RP in 100 cases of clinically localized PC, also failed to detect any association of the former with biochemical recurrence, although p27 positivity defined as ≥ 40% correlated well with a lower median pre-operative PSA and a lower Gleason score . When p27 expression was examined in a cohort of 53 patients with pT2 stage disease, no correlation with Gleason score was revealed and p27 had no prognostic value in predicting biochemical relapse .
Another Greek team retrospectively evaluated the prognostic relevance of p27 in 94 patients undergoing RP. They observed a statistically significant univariate correlation of p27(Kip1) expression, at a level below 30%, with high pre-operative PSA values and an increased likelihood of BF after RP. However these data were not confirmed in multivariate analysis .
In the present analysis, we did not observe the strong correlation between decreased p16 or p27 expression and PSA recurrence-free survival reported by a part of existing relevant studies. Possible explanations include the selection of patients representative of a specific stage in one study. Also, the inclusion of patients who received NHT prior to their RP does not consist a homogenous population let alone that this variable is known to influence time to PSA relapse after surgery. It is not clear whether race or different ethnic groups of patients contributed to the difference in the correlation with treatment outcome results.
Most importantly, different methodologies, pathologic material, and classification schemas for the evaluation of positivity were used to define p27 expression levels in other studies. With regard to the cut-off level for positivity, our intention was to achieve a high sensitivity for detection of p16 and p27 expression aberrations, based on the underlying biology. This explains why any positive p16 expression was considered as p16 overexpression whereas a high cut-off positivity of 70% was used to establish an early detection of p27 loss in PC tissues.
Several limitations of the present work should be acknowledged. The small number of patients included in our study does not permit to draw safe conclusions as 70 patients is a small number of patients in which to identify an association with outcomes unless the marker examined is very robust. Lack of compartmental p16, p27 evaluation and absence of representation of lowest Gleason scores (2, 3, 4) in the study population might have also influenced the validity of our results. The group of patients studied in this report may be too homogeneous without sufficient events to be significant in the time period examined. Moreover, the prognostic values of these markers might have been better established if they had been compared to predicted outcomes of validated nomograms. Finally, our study was not intended to be all-inclusive of current prognostic markers such as surgical margins, seminal vesicle invasion, tumor marker ploidy status and proliferation indices.