Optimal protocol for PTEN immunostaining; role ofanalytical and preanalytical variables in PTEN stainingin normal and neoplastic endometrial, breast,and prostatic tissues☆☆
Oscar Maiques PhD student a, Maria Santacana MsC a, Joan Valls PhDa,Judit Pallares MD, PhDa, Cristina Mirantes PhD student a, Sónia Gatius MDa,Diego Andrés García Dios PhDb, Frederic Amant MD, PhDb,Hans Christian Pedersen PhD c, Xavier Dolcet PhDa, Xavier Matias-Guiu MD, PhDa,⁎
aDepartment of Pathology and Molecular Genetics/Oncologic Pathology group, Hospital Universitari Arnau de Vilanova,Universitat de Lleida, IRBLleida, Lleida 25198, SpainbDepartment of Obstetric and Gynecology, University Hospitals Gasthuisberg, Leuven 3000, BelgiumcResearch and Development Dako, DK-2600 Glostrup, Denmark
Received 24 July 2013; revised 8 October 2013; accepted 16 October 2013
Summary In some tumors, phosphatase and tensin homolog (PTEN) inactivation may have prognosticimportance and predictive value for targeted therapies. Immunohistochemistry (IHC) may be aneffective method to demonstrate PTEN loss. It was claimed that PTEN IHC showed poorreproducibility, lack of standardization, and variable effects of preanalytical factors. In this study, wedeveloped an optimal protocol for PTEN IHC, with clone 6H2.1, by checking the relevance of analyticalvariables in normal tissue and tumors of endometrium, breast, and prostate. Pattern and intensity ofcellular staining and background nonspecific staining were quantified and subjected to statisticalanalysis by linear mixed models. The proposed protocol showed a statistically best performance (P b.05) and included a high target retrieval solution, 1:100 primary antibody dilution (2.925 mg/L), FLEXdiluent, and EnVisionFLEX+ detection method, with a sensitivity and specificity of 72.33% and78.57%, respectively. Staining specificity was confirmed in cell lines and animal models. Endometrialcarcinomas with PTEN genetic abnormalities showed statistically lower staining than tumors withoutalterations (mean histoscores, 34.66 and 119.28, respectively; P = .01). Controlled preanalytical factors(delayed fixation and overfixation) did not show any statistically significant effect on staining withoptimal protocol (P N .001). However, there was a trend of significance for decreased staining andfixation under high temperature. Moreover, staining was better in endometrial aspirates than in matched
☆☆ Conflict of interest: The study was done according to the research collaboration with Dako Denmark A/S. The sponsor did not participate interpretation of results. One author (H. C. P.) works for Dako and was involved in optimization of protocol. The other authors do not have conflict of interest.⁎ Corresponding author. Hospital Universitari Arnau de Vilanova, Av Rovira Roure, 80, 25198 Lleida, Spain.E-mail address: [email protected] (X. Matias-Guiu).
Phosphatase and tensin homolog (PTEN) is frequentlysomatically mutated or deleted in tumors [1]. Among otheractivities, PTEN antagonizes the phosphoinositide 3 kinase/alpha serine-threonine kinase (PI3K/AKT) pathway bydephosphorylating PIP3, resulting in a decreased transloca-tion of AKT to cellular membranes [2,3]. PTEN can beinactivated by several different mechanisms, including genemutation, deletion, epigenetic silencing, transcriptionalrepression, micro-RNA regulation, disruption of competi-tive endogenous RNA networks, posttranslational modifi-cations, and aberrant PTEN localization [4]. PTEN/PI3Kpathway activation may be a strong biological correlate ofmetastasis and poor prognosis in some tumors [5-7].Moreover, a new functional role for PTEN in genomicinstability has been proposed [4] suggesting that PARP(poly ADP ribose polymerase) inhibitors may be useful inpatients with PTEN mutant tumors [8]. Furthermore, PTENexpression has also been related to response to several drugsin anticancer targeted therapies [9-11]. It has been suggestedthat immunohistochemistry (IHC) may be an effectivemethod to demonstrate loss of PTEN function, but somevariability and poor reproducibility have been observed withdifferent antibodies and techniques [12-16].
In this study, we optimized a protocol for PTEN IHC andassessed the impact of analytical and preanalytical factors, tocheck the usefulness of IHC as a method for evaluation ofPTEN alterations. We assessed these variables in 3 tissues(endometrium, breast, and prostate), in which PTEN isexpressed in normal cells and frequently abnormal in tumorcells. The optimal protocol was tested in cell lines and animalmodels as well as in tumors with known molecular status ofin PTEN.
2. Materials and methods
2.1. Case selection and tissue microarrayconstruction
Five tissue microarrays (TMAs) were constructed fromformalin-fixed, paraffin-embedded (FFPE) tissue, corre-sponding to normal prostate (PN; 30 cases), prostaticcarcinoma (PC; 30 cases), normal breast tissue (BN; 30cases), breast carcinoma (BC; 30 cases), normal endome-trium (30 cases), and endometrial carcinoma (EC; 45cases). Samples were obtained from the surgical pathology
files of Hospital Universitari Arnau de Vilanova, Lleida,Spain (HUAV). The study was approved by the local ethicscommittee. Informed consent was obtained from eachpatient. A tissue arrayer device (Beecher Instrument, SunPrairie, WI, USA) was used. All tissue samples werehistologically reviewed by 3 members of the team, andrepresentative tumor or nontumor areas were marked in thecorresponding paraffin blocks. Two selected cylinders (0.6mm in largest diameter) from 2 different tumor ornontumor areas were included for each case.
2.2. Tissues for assessment of preanalyticalvariables
Eight surgical specimens were used, obtained from thepathology files of HUAV. Three of them corresponded tohysterectomy specimens for EC, 4 were breast excisions forBC, and 1 was a partial prostatectomy for prostatichyperplasia. They were obtained from the operating room,immediately after resection, and were selected because thesize of the specimens allowed obtaining a high number ofsmall fragments to check all possible combinations of thecontrolled, preanalytical variables. The tissue was obtainedfrom the exceeding material, after having selected tissuesamples for appropriate diagnosis. The influence of addi-tional, uncontrolled, preanalytical variables was alsochecked in 4 TMAs constructed from FFPE tissuescorresponding to matched endometrial aspirates and hyster-ectomy specimens. These tissues had been collected from atotal of 100 patients with EC, from the surgical pathologyfiles of HUAV. The study was approved by the local ethicscommittee. Informed consent was obtained from eachpatient. Although preanalytical conditions were not con-trolled in these cases, this material was considered interestingfor assessing the influence of uncontrolled preanalyticalfactors because endometrial aspirates are usually immersedin formalin, immediately after being obtained whilehysterectomy specimens are usually subjected to variableconditions of delayed fixation and time fixation.
2.3. Case selection for validation
We used 2 different series of EC with known PTENmolecular status. One of them corresponded to 33 cases,from Hospital Santa Creu i Sant Pau, included in a TMA,which had been previously assessed for PTEN alterations, bymutation, promoter hypermethylation, and loss of heterozy-gosity (LOH) analyses [6]. The second series corresponded
to 32 cases, which have been previously evaluated forgenomic copy number gene alterations by high-resolutionsingle nucleotide polymorphism (SNP) microarray, obtainedfrom the surgical pathology files of HUAV (manuscriptin preparation). Whole FFPE sections were available foreach of these cases. Informed consent was obtained fromeach patient.
2.4. Cells lines
Three different human EC cell lines were used as controlsfor IHC. The Ishikawa 3-H-12 (IK) cell line was obtainedfrom the American Type Culture Collection (Manassas,VA). RL-95/2 (RL) and HEC-1-A (HEC) cells were a giftfrom Dr Reventos (Hospital Vall d’Hebron, Barcelona,Spain). Two of them have abnormalities in PTEN (IK andRL-95-2), whereas the third (HEC) contains wild-typePTEN. IK cells [17] contain 2 frameshift mutations inPTEN. RL-95-2 is derived from a grade 2 moderatelydifferentiated adenosquamous EC [18] and produces amutated PTEN protein. To generate appropriate positiveand negative controls, IK and RL-95-2 were transfected withexogenous PTEN, whereas HEC was subjected to PTENKnock-down. IK and RL 95/2 cells were transfected with aplasmid containing complementary DNA for PTEN (PRK5-PTEN) or the empty vector (PRK5) using polyethylenimidereagent. On the other hand, lentiviral-based vectors for RNAinterference-mediated gene silencing (FSVsi) consisted of anU6 promoter for expression of short-hairpin RNAs(shRNAs) and the Venus variant of YFP under the controlof an SV40 promoter for monitoring transduction efficiency.Samples were harvested by trypsinization and embedded in1% agarose in phosphate-buffered saline, fixed in neutralbuffered formalin for 12 hours, and embedded in paraffin. Atthe end, we had 6 different cell line controls. Three of themwere used as negative controls (IK, RL-95-2, that do notexpress PTEN protein, and HEC subjected to PTENsilencing by shRNA PTEN), and 3 were used as positivecontrols (HEC that expresses PTEN protein, and IK and RL-95-2 cell lines after overexpressing PTEN by transfectionwith PTEN complementary DNA).
2.5. Animal tissues
Three different animal models were used: first, PTEN+/+mice, with 2 normal copies of PTEN; second, PTEN+/−mice, which have deletion of 1 copy of PTEN and developendometrial hyperplasia in 100% of cases and EC in 20% ofthe cases [19]; finally, a novel conditional and induciblePTEN−/− mouse model, which was recently developed byour group [20], with PTEN deletion of the 2 alleles, mainlyin epithelial endometrial cells, but not in stromal cells, anddevelopment of EC in 100% of the cases. Five animals ofeach model were euthanized by cervical dislocation, anduteri were collected. FFPE tissue was subjected to IHC.
2.6. Immunohistochemistry
TMA blocks and whole-section FFPE blocks weresectioned at a thickness of 3 μm, dried for 1 hour at 65°C,deparaffinized, rehydrated, and subjected to target retrieval inthe pretreatment module, PTLINK (Dako, Glostrup, Den-mark) at 95°C for 20 minutes in target retrieval solution, pH 9(Dako). Endogenous peroxidase was blocked using peroxi-dase-blocking reagent (Dako) followed by incubation withprimary PTEN antibody clone 6H2.1 (Dako), whichrecognizes human and mouse PTEN protein. Incubationtime for primary antibody was 40 minutes in whatevercondition. Secondary EnVision FLEX/HRP (20 minutes) orFLEX+/HRP (20 + 15 minutes) was used to amplify signal.Detection was done using 3,3′-diaminobenzidine tetrahy-drochloride as chromogen (Dako). Slides were counter-stained with hematoxylin, dehydrated, and mounted.Appropriate negative controls were also tested, includingsame tissues with omission of primary antibody and tumorsamples known to show PTEN silencing.
2.7. Optimization of PTEN IHC
We applied 12 different protocols, resulting fromcombination of 3 analytical variables (primary antibodydilution, primary antibody diluent, and detection method), asshown in Fig. 1. Three dilutions were selected (1/50, with afinal antibody concentration of 5.85 mg/L, 1/100, with a finalantibody concentration of 2.925 mg/L and 1/150, with a finalantibody concentration of 1.95 mg/L). Evaluation of primaryantibody diluent (FLEX Antibody Diluent, K8006 versusDako Antibody Diluent, S3022) and detection method(EnVision FLEX versus EnVision FLEX+) was also done.
2.8. Preanalytical variables
Samples fixed immediately after surgical resection (b5minutes) were considered as reference [21] (Fig. 2). Fourdifferent conditions of delayed fixation (30 minutes, 2 hours,8 hours, and 12 hours) were combined with 3 differenttemperature conditions (4°C, room temperature, and 36°C).Four different time fixation conditions (3, 12, 24, and 72hours) and 3 different time dehydration conditions (2, 3, and4 hours) were evaluated. The impact of preanalyticalvariables was also assessed in TMA constructed frommatched endometrial aspirates and hysterectomy specimensfrom 100 EC patients.
2.9. Assessment of antigen stability after storageof paraffin blocks
Stability of PTEN antigen was assessed in 3 BC samplesthat had been subjected to optimal conditions (fixation timeof b5 minutes). FFPE blocks were stored at differenttemperature conditions (4°C, room temperature, and 36°C).
Fig. 1 Analytical variables.
Fig. 2 Preanalytical variables.
525A protocol for PTEN immunostaining
The blocks were retrieved at the moment (0 months), 3, and 6months, and sections were obtained for IHC.
2.10. Scoring of IHC
Two different scoring systems were used and comparedbetween them. First, we used a histoscore (Hscore) that takesinto consideration the intensity of the staining and thepercentage of positive cells, by applying the formula Hscore =1× (% light staining) + 2× (%moderate staining) + 3× (% strongstaining). This Hscore ranges from 0 to 300. The reliability ofsuch scores for the interpretation of IHC staining in TMAs hasbeen reported previously [26]. A second scoring system wasalso used, with the following categories: 0 (negative staining), 1(lower staining than normal tissue), 2 (staining similar to normaltissue), and 3 (staining higher than normal tissue). Scoring wasperformed by 3 of the authors. Discrepancies were solved byexamining problematic cases under the microscope.
2.11. Statistical analysis
Linear mixed models were used to assess the effects ofany experimental factor on PTEN staining. For eachexperimental design, SEs were used to statistically assessthe main effect of each variable but also their pairedinteractions. Parametric and nonparametric tests, such asStudent t and Mann-Whitney U tests, were used to study theassociation between PTEN IHC expression and the presenceof PTEN genetic alterations. A t test and a Wilcoxon test forpaired data were used to compare PTEN IHC betweenendometrial aspirates and hysterectomy specimens. Finally,the accuracy of PTEN IHC level as predictor for thepresence/absence of genetic alterations was checked afterestablishing an optimal cut point, using the same criterion asthe one used for regression models, based on minimizing thesum of squared residuals. Then, we performed a sensitivity/specificity analysis to estimate the discriminating ability of
such dichotomization and to control false-positive and false-negative cases. All analyses were performed using the freepackage R and a threshold for statistical significance at 0.05.
3. Results
3.1. Optimization of PTEN IHC
Staining pattern and intensity of cell cytoplasm and nucleusas well as background staining were considered. Specificstainingwas determined as the cytoplasmic and nuclear stainingdetected in appropriate cells in the positive control cases andabsent when primary antibody was omitted. Backgroundstaining was the nonspecific staining that was seen in similarintensity in the different tissue components, including extra-cellular elements of connective tissue, and also seen in negativecontrol cases. Results of the effects of the 3 different analyticalvariables (primary antibody dilution, primary antibody diluent,and detection method) are shown in Table 1.
3.1.1. EC and endometrial normal tissueA significant interaction by linear mixed model was seen
between the detection method and primary antibodyconcentration either at the nuclear or at the cytoplasmiclevel in EC (P b .0001) and at the cytoplasmic (P = .002) andnuclear levels (P = .0044) in normal tissue [28]. There wasalso an effect of the diluent type on background (P b .0001).Moreover, the intensity of staining was higher when we usedhigh concentrations of the antibody in combination withEnVisionFLEX+ (Supplementary Figures 1 and 2). Theintensity of the nonspecific background staining was lowerwhen using S3022 diluent (Supplementary Figures 3 and 4).
3.1.2. BC and breast normal tissueAt the cytoplasmic and the nuclear levels, there was a
relevant interaction regarding the detection method and theprimary antibody concentration (P = .0009 and P = .007,respectively, for BC; P b .001 and P = .0004 for normaltissue). An interaction between type of diluent and thedetection method (P b .0001 both in BC; P b .0001 and0.00021 in normal tissue) was also observed. Backgroundstaining was dependent on each analytical variable. Stainingincreased at high antibody concentration in combination withthe EnVisionFLEX+ (Supplementary Figures 5 and 6).However, the primary antibody concentration did notinfluence staining levels as much as the EnVisionFLEX+detection method. On the other hand, the staining of thenucleus and the cytoplasm in BCwas higher whenwe used theEnVisionFLEX+ method combined with S3022 diluent.Moreover, statistically significant differences were seenwhen different diluents were used, depending on the detectionmethod and primary antibody concentration that was used.There were some differences between BC and normal tissueregarding background staining (Supplementary Figures 7, 8).However, at the cytoplasmic level in normal tissue, weobserved high Hscores when using S3022 diluent incombination with EnVisionFLEX+ detection method. A
Table 1 Results of the effects of combination of the 3 different anal
NOTE. The table is organized according type of tissue (normal tissue or tumorAbbreviation: NS, nonsignificance.
decrease in the intensity background staining in normal tissuewas seen when using S3022 diluent instead of FLEX diluent.
3.1.3. PC and prostatic normal tissueIn tumor tissue, there was an effect of the 3 analytical
variables independently (P b .0001 each variable) atcytoplasmic level. At nuclear level, there was an interactionbetween the detection method and primary antibody diluentused (P b .001) and a second interaction between the diluentand primary antibody concentration (P = .05). FLEX diluentshowed lower background than S3022 diluent. SupplementaryFigures 9 and 10 show an increase of the cytoplasmicimmunostaining in an independent manner when using highprimary antibody concentration, S3022 diluent and EnVision-FLEX+ detection method. At the nuclear level, there was anincrease in staining at high concentrations combined withFLEX diluent. In normal tissue, 2 statistically significantinteractions were obtained at the cytoplasmic level; first,combination of primary antibody concentration and thedetection method (P = .04), and second, combination of typeof primary antibody diluent and the detection method (P b.0001). At the nuclear level, there was an interaction betweenprimary antibody concentration and the detection method thatwas used (P = .003) and a second independent effect regarding
tissue) and organ (endometrium, breast, and prostate).
527A protocol for PTEN immunostaining
type of primary antibody diluent used (P b .0001). At thebackground level, there was only an effect of primary antibodydiluent (P b .0001). As seen in Supplementary Figures 11 and12, a positive effect was observed at cytoplasmic level as aresult of an increase of primary antibody concentration incombination with the EnVisionFLEX+ detection method.Similarly, a positive effect was seen when combiningEnVisionFLEX+ detection method with S3022 diluent.Finally, FLEX diluent offered lower background stainingthan S3022 diluent in both normal and tumor tissue.
3.1.4. Optimal protocolIn summary, evaluation of the effects of the analytical
variables allowed us to obtain several conclusions: (1)FLEX diluent was better than S3022 diluent, in terms ofreducing background and improving staining qualityparticularly in breast tissue, but the impact in endometrialand prostatic cells was not significant; (2) the use of thelinker significantly improved the intensity of PTEN stainingby reducing false-negative staining and resulting in a muchbetter definition of cytoplasmic/nuclear staining. (3) Higherconcentration of primary antibody (1/100 dilution) wasassociated with fewer false-negative stains. After optimiza-tion, the proposed protocol included a high target retrievalsolution (20′), a primary antibody dilution of 1/100, FLEXdiluent, as primary antibody diluent, and EnVision FLEX+as detection system. Interestingly, FLEX diluent workedbetter in prostatic tissue, but S3022 got better staining inbreast and endometrial tissue. This protocol gave theclearest signal to noise of PTEN immunostaining. Repre-sentative pictures of staining with the optimal protocol areseen in Fig. 3.
3.2. Specificity of staining by applying theoptimal protocol
3.2.1. Cell linesPTEN-positive staining was seen in the HEC as well as in
IK and RL overexpressing PTEN. Negative PTEN stainingwas seen in IK, RL, and PTEN shRNA HEC (SupplementaryFigure 13).
Fig. 3 Representative pictures of PTEN immunostaining by using
3.2.2. Animal modelTissues from wild-type PTEN mice showed intense
epithelial and stromal staining. Tissues from heterozygousPTEN+/− mice showed patchy, decreased staining. Tissuesfrom the conditional, inducible PTEN−/− mice showednegative staining in epithelial cells with intact expression ofstromal cells (Supplementary Figure 14).
3.2.3. ECs with known molecular statusFrom the initial series of 33 EC with known molecular
status, only 29 cases were informative (Table 2). The Hscoreof the cases with known PTEN molecular alterations (15cases) ranged from 0 to 160 (mean Hscore, 34.66), whereasthe Hscore of cases with absence of PTEN molecularalterations (14 cases) ranged from 0 to 250 (mean Hscore,119.28) (P = .01).
3.2.4. ECs characterized with SNP arraysFrom a series of 32 EC, previously assessed for gains and
losses in the PTEN region, by SNP arrays, 28 cases wereinformative. Eight cases showed LOH, 10 cases exhibitedgene polysomy, and 11 cases had no copy numberabnormalities. Tumors with decreased PTEN copy numberhad Hscore values that ranged from 0 to 150 (mean, 30),whereas the Hscore ranged from 0 to 170 (mean, 65) in thosewith normal copy number and ranged from 0 to 200 (mean,66) in those cases with increased copy number. Differenceswere not statistically significant (P = .56).
3.3. Effects of preanalytical variables
3.3.1. Delayed fixation combined with temperatureImmunostaining, as assessed by the proposed protocol,
was not influenced by preanalytical variables. The onlyfactor that showed a mild variation in results was temperaturebecause there was a trend of significance showing certaindecrease in staining in association with fixation under hightemperature (36°C) (P = .08). The influence of preanalyticalvariables in the tumor stroma was higher than in epithelialcells. Staining in stromal cells was lower when the samplewas subjected to delayed fixation at high temperatures
the optimal protocol. Magnification ×40, A, EC; B, BC; C, PC.
(P values around .001 in all cases) (Table 2). Backgroundnonspecific staining was not modified by any preanalyticalfactor (Fig. 4).
3.3.2. Time fixation combined with dehydration inalcohol absolute
PTEN staining, as assessed by the proposed protocol, didnot show any change in BN, BC, PN, PC, when time offixation was modified (Fig. 5). Similar results were obtainedin normal endometrium and EC, although there was a trendof significance in samples exposed to prolonged dehydrationin alcohol absolute, which showed higher staining, both atthe cytoplasmic and nuclear levels (P = .042 and P = .040,respectively).
3.3.3. Comparison between matched endometrialaspirates and hysterectomy specimens
Hscores in the aspirates ranged from 0 to 230 (mean, 80),whereas in the hysterectomy specimens, Hscores rangedfrom 0 to 250 (mean, 40) (P = .002).
3.4. Assessment of antigen stability after storage
There was a statistically significant interaction betweenslide storage temperature and time of storage, only whenstorage temperature was very high (36°C) (P b .00001, .018,and .029, respectively). The effect of temperature, time, andtheir interaction had statistical significance (P b .00001, .05and .004, respectively).
3.5. Comparison of 2 different scoring methods
Statistical analysis revealed that the Hscore systemwas better than the 4-tiered system in distinguishingEC with known molecular abnormalities (P = .01 and P =.92, respectively).
4. Discussion
PTEN is a tumor suppressor gene located at 10q23 [1].PTEN acts as both a lipid and a protein phosphatase. The
former activity is dephosphorylation of the 3 positions ofphosphatidylinositol 3,4,5-triphosphate, a second messengerof phosphatidylinositol 3 kinase (PI3K) [2,3]. PTEN antag-onizes PI3K activity and negatively regulates the PI3K/Aktpathway [4]. Germline PTEN mutations are found in Cowdensyndrome [22]. PTEN somatic mutations can be detected in asignificant proportion of primary glioblastomas [23,24] aswell as cancers of prostate [15,25], endometrium, ormelanoma [14,26-31]. Occasionally, PTEN mutations havebeen reported in sporadic BC (5%) [16], thyroid cancer (7%)[32], lung cancer (9%), and ovarian cancer (9%) [27]. LOH at10q23 can be found in approximately 50% of human prostatecancers, whereas homozygous deletions of PTEN can bedetected in approximately 10% of these cases [15,25]. It hasbeen questioned whether loss of 1 PTEN allele (haploinsuffi-ciency) is sufficient for tumorigenesis or whether inactivationof the second allele might occur through epigenetic rather thanmutational events [33].
A relationship between PTEN status and prognosis hasbeen proposed in several cancers. In PC, PTEN genomicdeletions have been associated with poor prognosis, andalternative mechanisms of posttranscriptional down regula-tion may play a role [34]. Deletion or reduced expression ofPTEN is present in BC. Approximately 50% of patients withBC harbor a mutation in PTEN or have lost at least 1 copy ofPTEN. In addition, the complete loss of PTEN protein ismore frequent in metastatic than in primary BC [16].Moreover, a role for PTEN has been proposed as predictorof response to mammalian target of rapamycin (mTOR) orphosphoinositide 3-kinase (Pi3K) inhibitors and also inresistance to HER2 or EGFR inhibitors [11,35]. PTEN isfrequently abnormal in EC. LOH at the PTEN region occursin 40% of cases, and somatic mutations, in 37% to 61%. Theprognostic significance of PTEN mutations in EC iscontroversial. Several studies have shown that EC withPTEN mutations has genomic instability [5,29,30].
PTEN function can be altered by genetic mutations, whichresult in either a heterozygous or homozygous loss. Epigeneticsilencing, transcriptional repression, micro-RNA regulation,disruption of competitive endogenous RNA networks, post-translational modifications, and the aberrant localization ofPTEN can cause subtle or dramatic losses of PTEN function.Furthermore, numerous PTEN-interacting proteins can posi-tively or negatively regulate PTEN function [1]. PTEN is notthe only gene abnormal in the PI3K signaling pathway.Mutations in PIK3CA (p110) and PIK3RI (p85α) havebeen detected in some tumors [2,3]. Obviously, alterations ofthese genes may be associated with activation of the PI3Ksignaling in cases in which PTEN expression is retained.
Accurate identification of functional PTEN loss isregarded as a possible important parameter in comprehensiveevaluation of tumors, as a potential diagnostic, prognostic,and predictive biomarker. However, although IHC is seen asa good method to assess functional PTEN loss in tumors,there is a general belief that it has been associated with poorreproducibility. This has, in large part, been due to different
Fig. 4 Impact of preanalytical variables on PTEN staining in EC (delayed fixation and temperature) (×20).
antibodies used, different protocols, various visualizationreagents, and use of manual or automated systems as well asthe variation of scoring systems between studies. However,to date, mutation analysis has been considered to be thestandard reference for detection of PTEN mutations andsubsequent loss of protein function [6,30]. One of thepossible reasons of poor performance of PTEN IHC in someseries may be the variation in preanalytical variables such asischemic time, fixation time, and fixation temperature. Goodperformance of an antibody in series of cases with goodcontrol of preanalytical variables does not guarantee asimilar performance in series of cases with differentconditions of delayed fixation or overfixation, unless thestaining protocol is optimized to get similar results insamples under different preanalytical conditions, as it hasbeen done in the present study.
In this study, we have developed an optimal protocol forPTEN IHC, using clone 6H2.1, which, in a previous study ofour group, had better performance in comparison with theother 3 commercially available antibodies [12]. We decidedto use normal and tumor tissue from 3 different types oforgans (endometrium, breast, and prostate). For validation,we used 3 cell lines, which were specifically selected aspositive and negative controls. We took also advantage of thefact that the clone 6H2.1 identifies both the human and themurine PTEN protein, for using 3 different animal models totest the results of our protocol. Finally, the protocol was testedin 2 validation series, composed of EC that had been assessedfor PTEN alterations by different approaches. Afterwards, wechecked the possible influence of preanalytical factors [21].Finally, we assessed the usefulness of 2 different scoringsystems. Specific attention was given to the impact ofanalytical and preanalytical variables to cytoplasmic andnuclear expression of PTEN. Although initially interpreted asa cytoplasmic protein, it is now clear that PTEN may beexpressed in the nucleus [12-16]. The ratio of cytoplasmic/nuclear expression is variable and context specific. It has beensuggested that a high nuclear/cytoplasmic ratio is seen innormal cells, whereas low nuclear/cytoplasmic ratio ischaracteristic of cancer cells [32].
The process of implementing IHC tests in pathology hasbeen simplified by the availability of standardized reagents,instruments, and protocols from commercial manufacturers.Our optimized protocol provides us a better sensitivitywithout decreasing specificity. We achieved these results byestablishing criteria for primary antibody dilution (1/100),primary antibody diluent (FLEX), and detection method(EnVisionFLEX+), in agreement with the reagents providedby the manufacturers of the PTEN antibody, clone 6H2.1.Finally, we observed that the Hscore method combiningintensity of staining and percentage of positive cells providesthe best specificity.
In summary, we have optimized a protocol for PTEN IHCafter having checked the influence of several analytical andpreanalytical variables. Although the protocol provided goodresults in samples subjected to different controlled conditions
of delayed fixation and overfixation, differences were seenwhen comparing uterine aspirates and matched hysterectomyspecimen, subjected to uncontrolled conditions of delayedfixation and overfixation. For that reason, it is recommendedthat PTEN IHC should be performed in samples subjected tooptimal fixation.
Supplementary data
Supplementary data to this article can be found online athttp://dx.doi.org/10.1016/j.humpath.2013.10.018.
Acknowledgments
The study was done according to the research collaborationwith Dako Denmark A/S. The research team was alsosupported by grants FIS PI100922, Fundación MutuaMadrileña AP75732010, 2009SGR794, RD12/0036/0013,Fundación Asociación Española contra el Cancer, programade intensificación de la investigación, Instituto Carlos III,Verelst Baarmoederkankerfonds, Leuven, and EuropeanNetwork for Individualized Treatment of Endometrial Carci-noma. F. A. is senior researcher for the research fundFlandersb. Tumor samples were obtained with the support ofXarxaCatalana de Bancs de Tumors, the TumorBanc Platformof RTICC, and Red de Biobancos (RD09/0076/00059).
X. M. and F. A. were responsible for the study design.O. M., M. S., and H. C. P. worked on optimization ofprotocol. O. M. and M. S. performed immunohistochemicaltechniques. X. M., M. S., and O. M. interpreted immuno-histochemical techniques. D. A. G. D. and F. A. performedand interpreted aSNP assays. X. M., J. P., and S. G. reviewedthe pathologic data. C. M. and X. D. obtained material fromthe cell line and animal models. J. V. performed statisticalanalysis. X. M. and O. M. were involved in drafting themanuscript. All the authors reviewed the Results andDiscussion sections.
References
[1] Steck PA, Pershouse MA, Jasser SA, et al. Identification of a candidatetumour suppressor gene, MMAC1, at chromosome 10q23.3 that ismutated in multiple advanced cancers. Nat Genet 1997;15:356-62.
[2] Chu EC, Tarnawski AS. PTEN regulatory functions in tumorsuppression and cell biology. Med Sci Monit 2004;10:235-41.
[3] Lee JO, Yang H, Georgescu MM, et al. Crystal structure of the PTENtumor suppressor: implications for its phosphoinositide phosphataseactivity and membrane association. Cell 1999;3:323-34.
[4] Song MS, Salmena L, Pandolfi PP. The functions and regulationof the PTEN tumour suppressor. Nat Rev Mol Cell Biol 2012;13:283-96.
[5] Terakawa N, Kanamori Y, Yoshida S. Loss of PTEN expressionfollowed by Akt phosphorylation is a poor prognostic factor forpatients with endometrial cancer. Endocr Relat Cancer 2003;10:203-8.
[6] Velasco A, Bussaglia E, Pallares J, et al. PIK3CA gene mutation inendometrial carcinoma: correlation with PTEN and K-RAS alterations.HUM PATHOL 2006;37:1465-72.
[7] Pradeep ST, Gayatry M, Sarah C, et al. Loss of LKB1 and PTEN tumorsuppressor genes in the ovarian surface epithelium induces papillaryserous ovarian cancer. Carcinogenesis 2013 [in press].
[8] Mendes-Pereira AM, Martin SA, Brough R, et al. Synthetic lethaltargeting of PTEN mutant cells with PARP inhibitors. EMBO MolMed 2009;1:315-22.
[9] Nagata Y, Lan KH, Zhou X, et al. PTEN activation contributes totumor inhibition by trastuzumab, and loss of PTEN predictstrastuzumab resistance in patients. Cancer Cell 2004;6:117-27.
[10] Berns K, Horlings HM, Hennessy BT, et al. A functional geneticapproach identifies the PI3K pathway as a major determinant oftrastuzumab resistant in breast cancer. Cancer Cell 2007;12:297-9.
[11] Liang K, Esteva FJ, Albarracin C, et al. Recombinant humanerythropoietin antagonizes trastuzumab treatment of breast cancercells via Jak2-mediated Src activation and PTEN inactivation. CancerCell 2010;18:423-35.
[12] Pallares J, Bussaglia E, Martinez-Guitarte JL, et al. Immunohisto-chemical analysis of PTEN in endometrial carcinoma: a tissuemicroarray study with a comparison of four commercial antibodies incorrelation with molecular abnormalities. Mod Pathol 2005;18:719-27.
[13] Djordjevic B, Hennessy BT, Li J, et al. Clinical assessment of PTENloss in endometrial carcinoma: immunohistochemistry outperformsgene sequencing. Mod Pathol 2012;25:699-708.
[14] Kapucuoglu N, Aktepe F, Kaya H, Bircan S, Karahan N, Ciris M.Immunohistochemical expression of PTEN in normal, hyperplastic andmalignant endometrium and its correlation with hormone receptors,bcl-2, bax, and apoptotic index. Pathol Res Pract 2007;203:153-62.
[15] Lotan TL, Gurel B, Sutcliffe S, et al. PTEN protein loss byimmunostaining: analytic validation and prognostic indicator for ahigh risk surgical cohort of prostate cancer patients. Clin Cancer Res2011;17:6563-73.
[16] Perren A, Weng LP, Boag AH, et al. Immunohistochemical evidenceof loss of PTEN expression in primary ductal adenocarcinomas of thebreast. Am J Pathol 1999;155:1253-60.
[17] Croxtall JD, Elder MG, White JO. Hormonal control of proliferation inthe Ishikawa endometrial adenocarcinoma cell line. J Steroid Biochem1990;35:665-9.
[18] Way DL, Grosso DS, David JR, Surwit EA, Christian CD.Characterization of a new human endometrial carcinoma (RL95-2)established in tissue culture. In Vitro 1983;19:147-58.
[19] Podsypanina K, Ellenson LH, Nemes A, et al. Mutation of PTEN/MMAC1 in mice causes neoplasias in multiple organ systems. ProcNatl Acad Sci U S A 1999;98:10320-5.
[20] Mirantes C, Eritja N, Dosil MA, et al. An inducible Knockout mouse tomodel the cell-autonomous role of PTEN in initiating endometrial,prostate and thyroid neoplasias. Dis Model Mech 2013;6:710-20.
[21] Engel KB, Moore HM. Effects of preanalytical variables on thedetection of proteins by immunohistochemistry in formalin-fixed,paraffin-embedded tissue. Arch Pathol Lab Med 2011;135:537-43.
[22] Agrawal S, Eng C. Differential expression of novel naturally occurringsplice variants of PTEN and their functional consequences in Cowdensyndrome and sporadic breast cancer. Hum Mol Genet 2006;15:777-87.
[23] Hu X, Pandolfi PP, Li Y, Koutcher JA, Rosenblum M, Holland EC.mTOR promotes survival and astrocytic characteristics induced byPten/AKT signaling in glioblastoma. Neoplasia 2005;7:356-68.
[24] Sano T, Lin H, Chen X, et al. Differential expression of MMAC/PTENin glioblastoma multiforme: relationship to localization and prognosis.Cancer Res 1999;59:1820-4.
[25] McMenamin ME, Soung P, Perera S, Kaplan I, Loda M, Sellers WR.Loss of PTEN expression in paraffin-embedded primary prostatecancer correlates with high Gleason score and advanced stage. CancerRes 1999;59:4291-6.
[26] Dolcet X, Llobet D, Pallares J, Rue M, Comella JX, Matias-Guiu X.FLIP is frequently expressed in endometrial carcinoma and has a rolein resistance to TRAIL-induced apoptosis. Lab Invest 2005;85:885-94.
[27] Tashiro H, Blazes MS, Wu R, et al. Mutations in PTEN are frequent inendometrial carcinoma but rare in other common gynecologicalmalignancies. Cancer Res 1997;57:3935-40.
[28] Llobet D, Pallares J, Yeramian A, Santacana M, Eritja N, Velasco A,Dolcet X, Matias-Guiu X. Molecular pathology of endometrialcarcinoma; practical aspects from the diagnostic and therapeuticalview points. J Clin Pathol 2009;62:777-85.
[29] George L, Mutter, Ming-Chieh L L, et al. Altered PTEN expression asa diagnostic marker for the earliest endometrial precancers. J NatlCancer Inst 2000;92:924-31.
[30] Levine RL, Cargile CB, Blazes MS, van Rees B, Kurman RJ, EllensonLH. PTEN mutations and microsatellite instability in complex atypicalhyperplasia, a precursor lesion to uterine endometrioid carcinoma.Cancer Res 1998;58:3254-8.
[31] Kanamori Y, Kigawa J, Itamochi H, et al. PTEN expression isassociated with prognosis for patients with advanced endometrialcarcinoma undergoing postoperative chemotherapy. Int J Cancer2002;100:686-9.
[32] Gimm O, Perren A, Weng LP, et al. Differential nuclear andcytoplasmic expression of PTEN in normal thyroid tissue, and benignand malignant epithelial thyroid tumors. Am J Pathol 2000;156:1693-700.
[33] Alimonti A, Carracedo A, Clohessy JG, et al. Subtle variations inPTEN dose determine cancer susceptibility. Nat Genet 2010;42:454-8.
[34] Berger AH, Knudson AG, Pandolfi PP. A continuummodel for tumoursuppression. Nature 2011;476:163-9.
[35] Fujita T, Doihara H, Kawasaki K, et al. PTEN activity could be apredictive marker of trastuzumab efficacy in the treatment of ErbB2-overexpressing breast cancer. Br J Cancer 2006;94:247-52.
