Phosphatase and tensin homolog (PTEN) is a promising prognostic and potentially predictive biomarker in prostate cancer.
To assess the effects of preanalytic variables on an analytically validated and fully automated PTEN immunohistochemistry assay.
PTEN immunohistochemistry was performed on Ventana immunostaining systems. In benign prostate tissues, immunostaining intensity across variable conditions was assessed by digital image analysis. In prostate tumor tissues, immunostaining was scored visually.
Delay of fixation for 4 hours or longer at room temperature or 48 hours or longer at 4°C and duration of formalin fixation did not significantly alter immunostaining intensity. Intensity of staining was highest in 10% formalin compared with other fixatives. Tumor tissues with PTEN loss processed using protocols from 11 academic institutions were all evaluable and scored identically. PTEN immunostaining of needle biopsies where tissue blocks had been stored for less than 10 years was more frequently scored as nonevaluable compared with blocks that had been stored for 10 years or longer. This effect was less evident for radical prostatectomy specimens, where low rates of nonevaluable staining were seen for 23 years or more of storage. Storage of unstained slides for 5 years at room temperature prior to immunostaining resulted in equivalent scoring compared with freshly cut slides. Machine-to-machine variability assessed across 3 Ventana platforms and 2 institutions was negligible in 12 tumors, and platform-to-platform variability was also minor comparing Ventana and Leica instruments across 77 tumors (κ = 0.926).
Automated PTEN immunostaining is robust to most preanalytic variables in the prostate and may be performed on prostate tumor tissues subjected to a wide range of preanalytic conditions. These data may help guide assay development if PTEN becomes a key predictive biomarker.
Tissue-based biomarkers hold great promise for the practice of personalized medicine in oncology. In particular, immunohistochemistry (IHC) assays, which enable the detection of proteins in situ in tissue specimens, are commonly used to determine expression states or detect genomic alterations in cancer as part of routine clinical care. The results of these assays, such as estrogen receptor (ER), progesterone receptor (PR), or human epidermal growth factor receptor 2 (HER2) expression in breast cancer or programmed death ligand-1 expression in other tumor types, provide prognostic or predictive information that helps to direct the treatment course. Because these data are essential to reliably inform the clinical decision-making process, it is critical that these tests accurately reflect underlying biology.
Before an assay result can be studied in a clinical context, it must have achieved a level of performance that shows it is fit for the purpose (the context of use) the result is intended to inform. Analytic validation establishes the sensitivity, specificity, precision, and accuracy of the assay compared with a gold standard control test. Yet even when the analytic performance of the assay itself has been established and validated, the result obtained is still dependent on preanalytic factors related to the acquisition of the specimen, its transport to the laboratory, and variations in the methodology by which the biological sample is handled.1 In the context of IHC assays, the effects of these preanalytic variables, such as cold ischemic time, duration and type of tissue fixation, tissue processing technique, and paraffin tissue block and slide age, have all been documented to affect assay performance,2–8 leading to potentially clinically significant errors. Along these lines, the American Society of Clinical Oncology and College of American Pathologists jointly published guidelines9,10 for tissue handling prior to the assessment of ER and PR as well as HER2 by IHC in breast cancer. However, few other markers have been subjected to such a rigorous analysis.
Over the past several years, we have focused on the analytic validation of an IHC assay to assess phosphatase and tensin homolog (PTEN), the most commonly inactivated tumor suppressor in prostate cancer.11–17 We and others have shown that the loss of PTEN expression in prostate cancer is correlated with oncologic outcomes and adds valuable prognostic information to the standard clinical parameters, such as prostate-specific antigen, Gleason score, and extent of disease in the context of radical prostatectomy.11,12,14–17 With the increasing popularity of active surveillance for clinical management of prostate cancer, PTEN and other prognostic biomarkers may be particularly useful to identify patients for whom definitive therapy is necessary.18 In addition, recent clinical trial results suggest an emerging role for PTEN as a predictive biomarker for response to phosphoinositide 3-kinase (PI3K)/protein kinase B (AKT)–targeted therapies in metastatic, castration-resistant prostate cancer.19 Yet despite the increasingly widespread use of this biomarker in many clinical laboratories, the effect of tissue variables on PTEN IHC results has not been studied extensively.20 To address this knowledge gap, we systematically queried the effects of preanalytic variables on the PTEN IHC assay, examining time of cold ischemia before fixation, time lapsed in formalin fixation, and the effect of different fixatives and different processing protocols, as well as tissue block and slide age.
MATERIALS AND METHODS
Patient and Tissue Selection
This study, including tissue collection and IHC staining, was approved by the Johns Hopkins University (JHU; Baltimore, Maryland) Institutional Review Board. All tumor samples were obtained from radical prostatectomy or transrectal ultrasound-guided needle biopsy specimens from patients at JHU, New York University (NYU; New York, New York), or the University of Washington (Seattle).
Tissue Microarray Construction
A number of tissue microarrays (TMAs) testing different preanalytic variables were used in the current study as described below. All TMAs were assembled using a Beecher tissue puncher and array system (Beecher Instruments, Silver Spring, Maryland). For each block, two to four 0.6-mm core samples were acquired separately from tumor and benign tissue (as specified below) and assembled in the receiving TMA block.
Time in Cold Ischemic TMAs
Five tissue punches (7 mm) were obtained from grossly benign tissue in 3 different radical prostatectomy samples at JHU from 2011. For each specimen, one of the cores was placed in 10% neutral buffered formalin (NBF) fixative immediately. Three additional cores were placed in a humidified chamber and left at room temperature for 1, 2, or 4 hours, followed by fixation in 10% NBF for 24 hours. A final core was processed following standard JHU protocol: it remained at room temperature for 30 minutes and was fixed in formaldehyde for 4 to 12 hours as a control. The same protocol was followed using grossly benign cores from an additional 7 radical prostatectomy samples from the NYU School of Medicine. After fixation and tissue processing, the NYU tissue blocks were transferred to JHU, punched in triplicate, and assembled into one TMA, and then processed and stained within the past 3 years at JHU as described below.
A second TMA testing the effects of cold ischemic time was prepared at the University of Washington. Cores (1.2 mm in diameter) from the grossly benign transition zone of 5 radical prostatectomy samples were procured and subjected to 30 minutes or 1, 2, 4, 8, 16, 24, or 48 hours of cold ischemic time before fixation in 10% NBF for 12 hours. Each sample was punched in duplicate using 1-mm cores for a TMA that was processed and stained using exactly the same machine protocol described below at the University of Washington.
Time in Fixation TMAs
A 7 mm core of grossly benign tissue (1.2 mm diameter cores) was taken from each of 8 radical prostatectomy specimens from JHU in 2005, subjected to less than an hour of cold ischemic time, and then fixed in 10% NBF for 0, 4, 8, 12, or 48 hours. After fixation, tissues were washed and stored in phosphate-buffered saline solution at 4°C until processing (all samples were processed together). The same protocol was followed at the NYU School of Medicine for an additional 8 cases. All the tissue blocks were then cored and assembled into one TMA, processed, and stained within the past 3 years at JHU.
A second TMA testing the effects of duration of fixation was prepared at the University of Washington. For this TMA, grossly benign tissue cores (1.2 mm in diameter) from the transition zone of 2 radical prostatectomy samples were fixed in 10% NBF for 8, 16, 24, 48, 64, 72, 96, 120, or 128 hours. As above, samples were punched using a 1-mm core and assembled into a TMA. The TMA was sent to JHU for staining.
Effects of Tissue Fixative
Three 7 mm punches of grossly benign tissue were harvested from each of 5 radical prostatectomy samples that were subjected to less than an hour of cold ischemic time and then fixed in 10% NBF, Bouin fixative, or Hollande fixative for 12 hours. The samples were then processed and stained at JHU.
Effects of Processing Protocols and Unstained Slide Age
Eleven needle biopsies of grossly identified prostate tumor (confirmed at frozen section) were punched from each of 2 radical prostatectomy samples subjected to less than 1 hour of cold ischemic time. Each core was placed in 10% NBF and sent overnight to each of the 11 institutions (Supplemental Table 1; see supplemental digital content, containing 1 table and 1 figure at www.archivesofpathology.org in the March 2019 table of contents) as part of a study by the Inter-Prostate SPORE Biomarker Study, which was a project comparing preanalytic variables related to biomarker expression from 11 institutions, for tissue processing with each institution's standard tissue processing protocol for clinical prostate biopsy specimens. After processing, paraffin blocks were prepared at each institution and sent to Memorial Sloan Kettering Cancer Center (New York, New York) for microtomy and slide preparation. Unstained slides from each case were stored for 5 years at room temperature and transferred to JHU for immunostaining. In addition, fresh unstained slides from the same cases were cut from the original paraffin blocks for immediate immunostaining at JHU.
Effects of Paraffin Block Age
A cohort of 277 Gleason score 6 or 7 prostate cancer needle biopsies (fixed and processed at JHU between 2014 and 2015) was prospectively immunostained for PTEN within 2 weeks of tissue embedding. A separate, previously published set of 278 Gleason score 7 biopsies was retrospectively recut from paraffin blocks of varying ages (2–15 years) and immediately immunostained for PTEN.21 Finally, a previously published cohort of 508 radical prostatectomy specimens was examined where blocks were between 9 and 23 years old at the time of recutting and immunostaining.
For all experiments except those designed to compare the Ventana (Roche-Ventana Medical Systems, Tucson, Arizona) and Leica (Leica Biosystems, Buffalo Grove, Illinois) immunostaining platforms as described below, PTEN IHC was performed on the Ventana Discovery Ultra, Benchmark Ultra, or Discovery XT automated staining platform using CC1 antigen retrieval buffer (Roche-Ventana) for 32 minutes at 100°C, followed by incubation with a rabbit anti-human PTEN antibody (1:100 dilution; clone D4.3 XP, Cell Signaling, Danvers, Massachusetts) for 32 minutes at 36°C, followed by the Optiview HRP multimer secondary detection system (Roche-Ventana).
Effects of Automated Immunostaining Platform
For immunostaining on the Leica Bond platform (Leica Biosystems), antigen retrieval was performed with Epitope Retrieval Solution 2 (Leica Biosystems) for 20 minutes, followed by incubation with the D4.3 PTEN antibody for 15 minutes and the Bond Polymer Refine Detection kit (Leica Biosystems) for 8 minutes. For comparison of the Ventana and Leica systems, a TMA containing 77 primary prostate tumors from radical prostatectomy, sampled with three to six 0.6-mm cores, was used, containing a total of 253 individual tumor spots that were evaluable by both methods.
All TMA slides were scanned at 40× magnification on the Aperio Scanscope AT Turbo (Leica Biosystems) for analysis. Experiments testing the effects of preanalytic variables were designed differently because of pragmatic considerations for each study and thus different scoring methodologies were used for each. To test the effect of fixation conditions (time to fixation, time in fixation, type of fixation), benign tissues from the same patient were subjected to variable fixation conditions prospectively. Because of the difficulty in prospectively procuring multiple 7 mm tissue cores of tumor from the same patient during gross examination to subject to varying times to fixation, time in formalin, and type of fixation, we used benign tissue to assess the effects of these parameters. We used quantitative analysis of PTEN intensity in benign glands to determine whether the preanalytic variable affected PTEN immunostaining. Quantification was performed with FrIDA TMAJ software (an open-source software using ImageJ available at http://tmaj.pathology.jhmi.edu)22,23 where the epithelium was traced using an annotation tool for each tissue core. Applying an automatic quantification algorithm, we measured brown (positive) pixels within a previously defined range, with a hue value of 0.1, a hue width of 0.5, and color saturation of at least 0.03. The intensity of the positive pixels was then averaged among all the positive pixels in the epithelium using all available cores from the same specimen to derive mean intensity.
For experiments where tumor tissue was available for study (processing protocols comparison, slide age study, block age study, immunostaining platforms study), PTEN IHC was visually scored dichotomously using our previously published and analytically validated scoring system by a pathologist blinded to the preanalytic conditions.13,15 A tissue core was considered to have PTEN protein loss if the intensity of cytoplasmic and nuclear staining was markedly decreased or entirely negative across any fraction of sampled tumor cells compared with surrounding benign glands and/or stroma, which provide internal positive controls for PTEN protein expression. Nuclear and cytoplasmic staining were scored together, as we have never seen cases with convincing loss of nuclear staining with intact cytoplasmic staining. If the tumor core showed PTEN protein expression in all sampled tumor glands, the tumor was scored as PTEN intact. Finally, a tumor was scored as having ambiguous PTEN IHC results if the intensity of the tumor cell staining was light or absent in the absence of evaluable internal benign glands or stromal staining.
Statistical tests were performed using GraphPad Prism 7 (GraphPad Software, La Jolla, California). For fixation parameter analysis, repeated-measures analysis of variance (ANOVA) was used to compare means among groups of the same specimens among variables. The χ2 test was used to compare proportion of nonevaluable specimens among groups of differing tissue block ages. The κ statistic was used to examine correlation between immunostaining results on different automated platforms.
Effect of Duration of Cold Ischemia Prior to Fixation on PTEN Immunostaining
To determine if duration of tissue cold ischemia affected IHC detection of PTEN protein, we evaluated benign prostate tissue from radical prostatectomy cases (n = 10) procured at JHU and NYU by TMA. A total of five 7 mm cores of benign tissue from each specimen were sampled and subjected to varying periods of cold ischemia (0, 1, 2, 4, or >4 hours) at room temperature prior to formalin fixation, standard tissue processing at JHU, incorporation into a TMA, and PTEN immunostaining (Figure 1, A through D). Duration of cold ischemic time did not correlate with the digitally quantified benign epithelial signal intensity for PTEN immunostaining (Figure 1, E) across different time periods (Figure 1, F) (P = .11 by repeated-measures ANOVA). For validation, we examined PTEN immunostaining in a separate group of benign prostate tissues from 3 radical prostatectomy samples procured at the University of Washington, where periods of up to 48 hours of cold ischemia at 4°C were tested (Figure 1, G through J). Though the number of samples was small and the mean signal intensity differed in this group of specimens compared with the previous group, likely because of differing tissue preparation and imaging conditions, no significant differences in immunostaining intensity for differing durations of cold ischemic time up to 48 hours were observed (P = .81 by repeated-measures ANOVA; Figure 1, K).
Effect of Duration of Formalin Fixation on PTEN Immunostaining
In order to investigate the effects of formalin fixation duration on PTEN immunostaining, we used TMAs of benign prostatic tissue procured at radical prostatectomy at JHU and NYU (n = 16). From each prostate, a total of six 7 mm cores of benign tissue were each subjected to less than an hour of cold ischemia and subsequently fixed in 10% NBF for 0, 4, 8, 12, 24, or 48 hours of fixation, followed by standard tissue processing at JHU and PTEN immunostaining (Figure 2, A through D). The digitally quantified benign epithelial signal intensity for PTEN immunostaining did not differ significantly across groups (P = .30 by repeated-measures ANOVA; Figure 2, E). In a second experiment using tissues from the University of Washington with 8, 16, 24, 48, 64, 72, 96, 120, or 128 hours of formalin fixation (data not shown), similar results were obtained (P = .28 by repeated-measures ANOVA), indicating that virtually any duration of formalin fixation is adequate prior to tissue processing for PTEN immunostaining, at least for a 4 mm core of tissue.
Effect of Different Fixatives on PTEN Immunostaining
To determine the effect of different tissue fixatives on PTEN immunostaining, we used benign 7 mm cores of prostate tissue from 5 different radical prostatectomy specimens that were fixed in Bouin fixative, Hollande fixative, or 10% NBF for 12 hours prior to processing and immunostaining (Figure 2, F through N). By digital image analysis, the intensity of PTEN immunostaining in the benign glands was significantly stronger for the samples fixed in formalin compared with Bouin or Hollande fixatives (P = .03 and .004, respectively; Figure 2, O).
Effect of Tissue Processing Protocol on PTEN Immunostaining
To address the effect of variations in tissue processor protocols on PTEN immunostaining, tumor tissues from 2 radical prostatectomies with homogeneous PTEN protein loss were examined. Eleven needle biopsies from each tumor specimen were obtained, fixed in 10% NBF, and shipped overnight to 11 different academic centers for tissue processing using the standard prostate biopsy tissue processing protocol for each institution prior to slide preparation and immunostaining (Supplemental Table 1). Because these samples were predominantly tumor tissue with only rare intermixed benign glands, we evaluated the effect of tissue processor protocols by blindly visually scoring each tumor sample for PTEN status and assessing the consistency of scoring for a given sample across protocols and the overall number of ambiguous or nonevaluable samples for each protocol. Ambiguous PTEN immunostaining implies weak to absent PTEN staining in both the cancer cells and the background stromal cells and benign glands, which constitute the internal control for PTEN expression in prostate tissue. Thus, ambiguous immunostaining indicates a probable effect of preanalytic variables in specimen processing. Despite notable differences in processing protocols, PTEN immunostaining was evaluable and visual scoring performed identically across both samples by a blinded reviewer (Figure 3, A through C).
Effect of Block Age on PTEN Immunostaining
We next assessed the effect of paraffin block age on PTEN immunostaining by examining the percentage of prostate tumor specimens scored as ambiguous for PTEN status, stratified by tissue block age. To establish a baseline, Gleason score 6 or 7 prostate cancer needle biopsies (fixed and processed at JHU between 2014 and 2015) were prospectively immunostained for PTEN within 2 weeks of tissue embedding. The percentage of cases with ambiguous or uninterpretable PTEN immunostaining in this cohort was less than 1% (2 of 277). To determine whether older specimen age is associated with an increased rate of ambiguous immunostaining results, we examined data from a recently published study of 278 Gleason score 7 biopsies where unstained slides were retrospectively recut from paraffin blocks of varying ages and immediately immunostained for PTEN. Combining data from this study with the prospective samples described above, we observed some variation in the proportion of cases with ambiguous immunostaining over time (Figure 3, D). The percentage of cases with ambiguous results using tissue blocks that were less than 5 years or 5 to 10 years old at the time of staining was 2% (6 of 399) and 4% (4 of 101), respectively (Figure 3, E). In contrast, tissue blocks that were between 11 and 15 years old at the time of cutting and staining showed 14% (8 of 55) of cases with ambiguous results. These data indicate a probable effect of tissue block age on PTEN immunostaining, specifically when the blocks are in excess of 10 years old at the time of cutting (P = .02, χ2 test).
To examine the variable of tissue block age in the context of radical prostatectomy specimens, where even older blocks were available for analysis, we next examined the proportion of ambiguous PTEN immunostaining in a previously described cohort of prostate tumors on TMA. In this cohort of 508 tumors where the donor tissue blocks were between 9 and 23 years old at the time of staining, the percentage of ambiguous specimens was generally surprisingly low (2% [2 of 85] for 9–13 years old and 5% [12 of 249] for 19–23 years old; Supplemental Figure 1, A), indicating a lesser effect of tissue block age on interpretation of PTEN immunostaining in radical prostatectomy samples compared with needle biopsy samples. Interestingly, specimens 15 to 17 years old showed a dramatic spike in percentage of ambiguous staining results (16% [13 of 82]; P = .002 for comparison with 9–13-year-old group and P = .001 for comparison with 19–23-year-old group). This period corresponded precisely with the introduction of microwave formalin fixation for radical prostatectomy processing at JHU. Initially, this was done without formalin injection of the prostate, and these samples showed a high rate of ambiguous staining results compared with other fixation conditions, an effect likely related to fixation and not to block age given the improved results for blocks from 19 to 23 years old (Supplemental Figure 1, B).
Effect of Unstained Slide Age on PTEN Immunostaining
We used the same 2 needle biopsy samples with PTEN loss described above for the tissue processing study to assess the effect of storage of unbaked, unstained slides for 5 years at room temperature prior to immunostaining. Across 11 different processing protocols, we compared PTEN immunostaining performed on a freshly cut block of tumor with PTEN loss with unstained slides from the same case that had been stored at room temperature for 5 years at the time of immunostaining. All cases were evaluable and PTEN scoring was identical between stored and fresh slides when performed by a blinded reviewer (Figure 3, F through I).
Intermachine Reproducibility of PTEN Immunostaining
Finally, we assessed whether performing PTEN immunostaining on different autostainer instruments would affect the interpretation of the stain in a given specimen. Using the identical reagents and machine instructions, machine-to-machine variability was assessed across 3 Ventana platforms (Benchmark Ultra, Discovery Ultra, and Discovery XT) and 2 institutions (JHU and Memorial Sloan Kettering Cancer Center) using a TMA containing PTEN-intact and PTEN-loss tumor samples from 12 radical prostatectomy specimens with focal PTEN loss, with identical blind scoring in all tumor components on all platforms (Figure 4, A through I). We also evaluated whether immunostaining results differed across differing automated immunostaining platforms by comparing results using the Ventana Benchmark and Leica Bond platforms in a TMA containing tumor samples from 77 radical prostatectomies. We found 98% (247 of 253) agreement between evaluable spots for blinded PTEN scores across the 2 platforms (κ = 0.926; 95% CI, 0.868–0.985; Figure 5, A through D).
The American Society of Clinical Oncology/College of American Pathologists guideline recommendations for ER, PR, and HER2 IHC brought to the forefront the importance of determining the reproducibility of an IHC stain in the context of varying and frequently unrecorded preanalytic conditions.9,10 Similarly, Food and Drug Administration clearance of tissue-based biomarkers for use as integral biomarkers in prospective clinical trials requires investigation of effects of preanalytic parameters, as does the Cancer Therapy Evaluation Program approval for retrospective biomarker studies in clinical trials. Despite the critical importance of these validation steps, remarkably few IHC biomarkers have been subjected to this rigorous analysis. PTEN loss is among the most common genomic alterations associated with clinical outcomes in prostate cancer, and it is a promising biomarker for risk stratification prior to active surveillance.11–16,24–26 In addition, recent trials suggest promise for PTEN as a predictive biomarker for response to drugs targeting the PI3K/AKT pathway.19 Previously, we rigorously analytically validated a simple and economical IHC assay to assess PTEN status in prostate cancer,11–17 beginning with a manual assay in 2011 and moving to an automated assay in 2013. Here, we have comprehensively investigated the effect of tissue fixation and storage conditions, as well as machine-to-machine and interobserver reproducibility, on PTEN immunostaining. Though PTEN is not yet an integral biomarker, these results provided needed information about specimen handling for future clinical trials where PTEN status is assessed by IHC.
Overall, we found that automated PTEN immunostaining is robust to the effects of most preanalytic variables. PTEN immunostaining may be performed on prostate tumor tissues subjected to less than 4 hours of cold ischemic time at room temperature or less than 48 hours at 4°C. Specimens must be fixed in 10% NBF for any duration up to 128 hours and processed using any standard tissue processing protocol. For needle biopsy specimens, archival paraffin blocks selected for retrospective immunostaining should be less than 10 years old, or less than 20 years old for radical prostatectomy specimens, and unstained slides may be stored at room temperature for up to 5 years prior to immunostaining.
For samples with 4 hours or less of cold ischemic time at room temperature, we found that the quantified variability in PTEN immunostaining intensity by duration of cold ischemic period was less than the overall variability among samples. Similarly, cold ischemic periods of up to 48 hours at 4°C did not significantly alter PTEN staining intensity in benign glands. How the duration of cold ischemia affects immunostaining is an important issue. However, this question is relevant primarily for radical prostatectomy specimens, because transrectal needle biopsy specimens are immediately placed into fixative at the time of biopsy in most circumstances. For radical prostatectomies, the vast majority of specimens are placed into fixative within 4 hours of excision at our institutions. Thus, the effects of prolonged cold ischemic periods are likely to be relevant only during weekends and holidays when specimens may be held prior to fixation because of lack of pathology staffing in the laboratory. In these cases, storage at 4°C for up to 48 hours appears to be a viable option to preserve the PTEN epitope. With the current popularity of robotic-assisted laparoscopic radical prostatectomies, extended periods of warm ischemia may also be relevant as a preanalytic variable, and this was one variable we did not investigate in the current study. The prostatectomy specimen often remains in the patient's body after vascular clipping for extended periods during laparoscopic surgeries, and this could potentially affect immunostaining as well. Unfortunately, it is quite difficult to control for this variable beyond simply documenting it, because it occurs prior to specimen receipt in pathology.
One important limitation of our cold ischemia study is that we used 7 mm cores to test different cold ischemic periods rather than the entire radical prostatectomy specimen because of clinical constraints on specimen use. Accordingly, these small cores were fixed quickly when placed in formalin because of rapid formalin diffusion into small tissue samples. In contrast, a larger radical prostatectomy specimen might have regions that are not exposed to formalin because of slow penetration for several extra hours, which could further impact immunostaining intensity. However, given that benign glands and stroma can be used as internal positive controls in all samples, it should be possible to distinguish which PTEN immunostaining samples remain interpretable even under these conditions.
Prolonged formalin fixation has been reported to lead to weak or absent staining in certain tissues because of excessive cross-linkage and contact with contaminating substances, which may lead to irreversible damage to antigen epitopes.2,27 In our series, the samples showed no significant difference in the intensity of the PTEN signal regardless of the period of formalin fixation, which suggests that even excessive cross-linkage in the tissue can be accommodated by the epitope retrieval step during the immunostaining process. Interestingly, we did have anecdotal evidence that insufficient formalin fixation could be detrimental to PTEN immunostaining interpretation. During our study of the effects of tissue block age on PTEN immunostaining, we found that the frequency of uninterpretable or ambiguous PTEN immunostaining results was dramatically higher during a period when we altered our tissue fixation protocol, including a microwave fixation step but without formalin injection of the radical prostatectomy specimen, which was subsequently instituted. Again, the importance of the internal control benign glands and stroma for PTEN immunostaining interpretation cannot be understated, as it enabled us to identify these presumably poorly fixed specimens because the internal control staining was markedly attenuated.
Interestingly, we saw some effects of fixative type on PTEN immunostaining. The vast majority of prostate needle biopsies and all radical prostatectomy specimens in the United States are fixed in NBF, which showed a higher intensity of PTEN staining in benign glands compared with other fixatives in our study. Bouin and Hollande solutions are compound fixatives known for increasing nuclear details and enhancing chromosome study, and these are less frequently used in genitourinary pathology specimens compared with gastrointestinal specimens. For the purposes of PTEN immunostaining, we recommend fixation in NBF to maximize PTEN intensity in internal control benign glands and stroma.
The processing of a tissue after fixation entails dehydration in graded organic ethanol solutions, clearing in xylene, and infiltration with paraffin wax. Most fluid transfer or enclosed processors use graded temperatures and pressures to facilitate dehydration and infiltration. The processing protocol can vary widely, which may influence the quality of the tissue for immunostaining. For example, in our study of 11 processing protocols at academic institutions, the total duration of the protocol varied dramatically, from 70 to 570 minutes (Supplemental Table 1). Such variations, including insufficient dehydration, contaminants in the xylene, and variations in the temperature of the wax used, may be deleterious for future immunostaining. When subjected to widely varying processing protocols from 11 different academic institutions, all samples remained identically interpretable for PTEN status; thus, variations in processing protocols are unlikely to influence PTEN status determination in a clinically significant fashion.
Finally, we also examined the effects of paraffin block age and unstained slide age on PTEN immunostaining interpretation. In needle biopsies, prospective, immediate recutting and staining of paraffin blocks was associated with the lowest proportion of ambiguous immunostaining results at only 1% of specimens. Retrospective recutting of the block at more than 1 year after tissue collection was associated with a slightly increased rate of ambiguous results, averaging around less than 5% and not differing dramatically for periods out to about 10 years. After this point, there was an increase in the number of unevaluable specimens, though the majority remained interpretable. In the setting of radical prostatectomies, we did not see a clear trend toward increases in proportions of ambiguous PTEN immunostaining results with increased time elapsed since specimen collection even out to more than 20 years. This difference between radical prostatectomy and needle biopsy specimens may be due to differences in initial fixation conditions or alternatively to tissue oxidation within the paraffin block, which may be less of an issue in prostatectomy samples than needle biopsies because of lower surface area to volume ratios in the former. Despite the robust results reported herein for PTEN, recent work by our group has suggested that many RNA-based biomarkers are altered with prolonged room temperature tissue block storage and unstained slide storage.28 Additional unpublished data from our laboratory indicate that this may be true for a limited set of proteins as well. In view of these data, we suggest that formalin-fixed, paraffin-embedded slides and blocks be stored at −20°C to preserve the maximum number of analytes for future studies, though this may be practical only in the research setting.
Our study has several limitations that merit discussion. First, we performed the validations largely using one automated staining platform, the Ventana system, throughout this study, with a smaller study of equivalence performed on the Leica system. Future work may more extensively examine the effects of preanalytic variables on additional common platforms, such as Dako (Agilent, Santa Clara, California), or more extensively investigate the Leica system. Second, because the study of block age effects was retrospective, for practical reasons we could not compare immunostaining on the same case when performed at the time of embedding and then later after archival storage for a number of years. Finally, we did not incorporate all possible preanalytic variables, such as warm ischemic time during robotic surgery.
In conclusion, the PTEN immunostaining protocol described herein is quite robust to most preanalytic parameters, including tissue fixation and storage conditions. Fixation in nonformalin fixatives such as Bouin or Hollande and prolonged (more than 10 years) storage of prostate biopsy paraffin blocks are associated with diminished staining. Ultimately, based on our published scoring system,11–17 PTEN immunostaining in tumor samples must always be interpreted in the presence of immunostaining of internal positive control stroma and glands, which provide a critical reference to assess any detrimental effects of preanalytic variables. Given its high level of analytic and preanalytic validation, the PTEN immunostaining protocol is poised to provide prognostic and potentially predictive information in prostate cancer, and results from future clinical trials of integrated biomarkers will determine its clinical utility.
Funding for this research was provided in part by a Transformative Impact Award from the CDMRP (W81XWH-13-2-0070) and NCI Cancer Center support grant P30 CA006973. Additional funding was provided by the NIH/NCI Prostate SPORE Pathology Cores P50CA58236 and P50CA097186; Defense Prostate Cancer Research Program awards W81XWH-14-2-0182, W81XWH-14-2-0183, and W81XWH-14-2-0185; the Prostate Cancer Biorepository Network (PCBN); NIH/NCI Prostate SPORE grant P50-CA92629; NIH/NCI Cancer Center support grant P30 CA008748; and the Prostate Cancer Foundation. H.I.S. was additionally supported by NIH/NCI Prostate SPORE grant P50-CA92629, NIH/NCI Cancer Center support grant P30 CA008748, and the Prostate Cancer Foundation.
The Inter-SPORE Prostate Biomarkers Study Team was a multi-institutional effort involving investigators from 11 different academic institutions, including Jonathan Said, MD (UCLA); Patricia Troncoso, MD (MD Anderson Cancer Center); Victor Reuter, MD, and Samson Fine, MD (Memorial Sloan Kettering Cancer Center); Jeffrey Simko, MD (University of California, San Francisco); Mark Rubin, MD (Dana Farber Cancer Institute); Rajal Shah, MD (University of Michigan); Michael Pins, MD (Northwestern University); Gustavo Ayala, MD (Baylor University); Robert Jenkins, MD, PhD, and John Cheville, MD (Mayo Clinic); Bruce Trock, PhD, and Angelo De Marzo, MD, PhD (Johns Hopkins University); and Beatrice Knudsen, MD, PhD, and Lawrence True, MD (University of Washington).
Supplemental digital content is available for this article in the March 2019 table of contents.
Drs De Marzo and Lotan contributed equally to this article.
Drs Lotan and De Marzo have received research support from Ventana to support other projects. Drs Trock and De Marzo have received research support from and consulted for Myriad Genetics. Dr Trock has also consulted for GenomeDx Biosciences and received a research grant from MDxHealth. Dr De Marzo served as a consultant for Cepheid Inc, is on the advisory board for Myriad Genetics, and has sponsored research from Jansson R&D and Myriad. The other authors have no relevant financial interest in the products or companies described in this article.