We report a reliable method of tissue banking which does not compromise the histological evaluation of prostate samples for patient diagnosis. By flash freezing tissue sections from the prostate, conventional histological evaluation can be performed without compromising margin analysis or pathological staging. In the rare event that an area of suspicion is only identified at the border of a prostate section, the adjacent biobanked section can be retrieved from storage for further study by the pathologist. 9% of our patients have more tissue taken from the biobank after identifying suspicious areas on clinical specimens, and we believe that the ability to access the biobanked tissue is fundamental to ensuring the integrity of the histological diagnosis.
Harvesting alternate sections also ensures the procurement of a substantial mass of tissue, and therefore provides a sufficient yield of RNA for genetic studies. Furthermore, the tissue is of sufficient quality for use in high-demand genetic studies, with 73% of epithelial samples demonstrating a RIN > 7. However, the converse is equally true, and we should note that 27% of epithelial samples will be insufficient for high fidelity RNA studies.
A RIN of > 7 is generally considered suitable for gene expression studies , and while our study did not have a control arm, the user-independence of measuring RIN values permits the comparison between different studies and helps to overcome this limitation. A large report from a cooperative human tissue biobank demonstrated 'less than good' quality RNA in 40% of samples collected , while a large pancreatic cancer biobank has demonstrated RIN ≥ 7 in just 42% of samples . While it is reasonable to assume that some of these differences reflect the varying cellular content of different tissues (pancreas being more sensitive to degradation than prostate), it is also possible that the delicate tissue-handling capabilities afforded by the robotic platform are responsible for a less severe impact on the cellular response to surgery; a possible relationship between RNase release within the tissue and specimen handling intraoperatively has been suggested . It must be reiterated that a direct comparison was not performed, and obviously the ideal randomized controlled comparison study to elucidate any difference would be unethical. To date, any comparisons between less mature robotic series and traditional open radical procedures have failed to show any significant difference in RNA quality [5, 6]. Comparing our results with data from other groups is challenging, and in part is limited by the small sample sizes, with Ricciardelli et al. reporting RIN values of 8-10 from just five prostate specimens . Bertilsson and colleagues have since reported RIN scores above 9 using further modified techniques with 53 prostate samples, presumably using the same source of samples from open radical prostatectomy . Such differences even between studies from the same institution highlight the important principle that RNA integrity reflects a complex interplay between pre-processing collection methods, and tissue processing methodology. From our data in the context of limited external data, we surmise that RALP permits the collection of prostate specimens which are at least non-inferior to traditional open prostatectomy with respect to RNA integrity.
The two stromal samples excluded from our multivariate analysis reflects the sensitivity of the RIN protocol to local DNA and/or RNase contamination. The method we have described is particularly advantageous in permitting the histological identification of our banked specimens, in comparison to biopsy techniques which rely on less accurate methods of sampling . For example, Riddick et al. have described taking punch biopsies from suspicious areas of the prostate (as identified by examining the prostate for firm irregular nodules and/or colour/texture heterogeneity), and performing histopathological assessment of the surrounding excised area. Although this method demonstrated concordance with the core sample in 92% of cases, it cannot be used to target specific cell populations within the biobanked tissue. We are able to direct our biopsy cores to histological areas of interest, permitting the investigation of stromal, benign or malignant epithelial cells. Since our samples are 5 mm in diameter there is a potential for introducing alternative cell types to that identified in the corresponding slide, although this method was consistent for all of our studies, thus minimising any bias which may have been introduced; alternative strategies such as Laser Capture Microdissection may offer greater selectivity and improve cell selection, but require real-time pathology support which is not available at most institutions including our own.
While a few samples were derived from the same RALP specimen, due to the multifocality and heterogeneity of malignant prostatic tissue, it is reasonable to assume that such samples will behave independently, hence minimising any selection bias which this might have introduced. Matched pair analysis between 45 stromal and 45 benign epithelial samples taken from the same specimen, confirmed the same relationship between RIN and cell type, with stromal samples showing a significantly lower mean RIN than epithelial samples (see Table 2).
Although it appears reasonable to assume that longer ischaemia times will potentiate RNA degradation, in this study we have found no negative impact on RNA quality within the narrow warm ischaemia times of robotic prostatectomy (mean total WIT of 120 mins). In one time course degradation study of lung tissue, nucleic acid stability has been demonstrated for up to 5 hours after excision at room temperature . Analysis of non-fixed surgical specimens revealed RNA stability in fresh tissue for up to 6-16 hours at room temperature . Similar studies have demonstrated minimal RNA degradation in samples stored on ice for as long as 24-96 hours after collection [12, 20]. While gene expression studies suggest an impact of ischaemia on prostatic tissue due to marked changes in hypoxia-related genes within the first hour of surgery [14, 15], our results did not show a relationship between warm ischaemia time and RIN (a measure of the integrity of the total RNA population, therefore not discounting alterations in the transciptome). This leads to the suggestion that the onset of cellular ischaemia intraoperatively is sufficient to produce genetic responses, without necessarily compromising the RNA integrity of prostatic tissue. It is possible that the expediency of our surgical technique impairs our ability to extrapolate any relationship with ischaemia, due to our narrow range of operating times. This finding also lends further support to our method of tissue collection.
In an effort to better understand clinicopathological factors which may influence specimen RNA quality, we performed multiple linear regression analysis, and found an inverse relationship between prostate volume and RNA quality.
One explanation for the relationship with prostate volume, may be a greater degree of ischaemia in specimens with a smaller surface area: volume ratio, although this is difficult to rationalize without a relationship between ischaemia time and RIN. Bertilsson et al. described a weak correlation with blood loss (r = -0.11, p = 0.02); the group postulated a relationship between excessive surgical handling, as indicated by blood loss, and subsequent RNase release . We did not anticipate a relationship between blood loss and RNA quality, given the restricted range of this variable (mean 157 ml, standard deviation 41 ml) when using a robotic technique, compared with the open radical prostatectomy study (median 575 ml; > 50% between 500-1000 ml). However, a similar explanation may be used for our relationship with prostate volume, with smaller prostates suffering less intraoperative surgical manipulation, and hence RNase release. An interquartile range was not reported in Bertilsson's initial study, and it is possible that a restricted cohort limited their identification of this correlation. One additional explanation may be that larger prostates are composed of a greater proportion of stromal tissue, which was also shown to inversely correlate with RNA quality in this study (r = -0.34, p = 0.03)  as well as our own (B = 1.738, p < 0.001).
The study identified a higher quality of RNA associated with samples taken from tumour cells as opposed to benign cells. It is possible that this relationship is related to a greater abundance of RNA within more aggressive tumour cell populations; this may reflect greater cell turnover and/or a higher rate of transcription per cell. Our hypothesis stems from the tumour cell exhibiting a greater abundance of RNA transcripts, and hence it might be postulated that a greater proportion of intact mRNA may exist, as a function of unregulated synthesis of a limited number of malignant transcripts. However, there is no literature to support this hypothesis and as such it warrants further investigation; we are in the process of designing a future study to evaluate this.