P.K., G.L., E-A.T. and N.P. contributed equally to the manuscript
A comprehensive immunohistochemical and molecular approach to the PI3K/AKT/mTOR (phosphoinositide 3-kinase/v-akt murine thymoma viral oncogene/mammalian target of rapamycin) pathway in bladder urothelial carcinoma
Version of Record online: 29 OCT 2012
© 2012 THE AUTHORS. BJU INTERNATIONAL © 2012 BJU INTERNATIONAL
Volume 110, Issue 11c, pages E1237–E1248, December 2012
How to Cite
Korkolopoulou, P., Levidou, G., Trigka, E.-A., Prekete, N., Karlou, M., Thymara, I., Sakellariou, S., Fragkou, P., Isaiadis, D., Pavlopoulos, P., Patsouris, E. and Saetta, A. A. (2012), A comprehensive immunohistochemical and molecular approach to the PI3K/AKT/mTOR (phosphoinositide 3-kinase/v-akt murine thymoma viral oncogene/mammalian target of rapamycin) pathway in bladder urothelial carcinoma. BJU International, 110: E1237–E1248. doi: 10.1111/j.1464-410X.2012.11569.x
- Issue online: 21 DEC 2012
- Version of Record online: 29 OCT 2012
- Accepted for publication 11 July 2012
- bladder cancer
What's known on the subject? and What does the study add?
A few published studies investigating single or various PI3K/AKT/mTOR signalling components have produced inconsistent results. Moreover, PI3K regulatory subunit p85a and activated p70S6K expression levels have not been previously examined in urothelial carcinoma (UC).
The present study addresses simultaneously all key members of PI3K/AKT/mTOR signalling cascade supporting a differential implication of PI3K/AKT/mTOR pathway components in urothelial tumorigenesis. Furthermore, we propose p-4E-BP1 as a potential prognostic marker in UC, which might assist the selection of patients more likely to benefit from chemotherapy regimens based on PI3K/AKT/mTOR pathway inhibition. Finally, the present study indicates PIK3CA/AKT1 mutational status as a potential predictive marker for time-to-recurrence.
- • To perform a comprehensive simultaneous assessment of all key members of phosphoinositide 3-kinase/v-akt murine thymoma viral oncogene/mammalian target of rapamycin (PI3K/AKT/mTOR) pathway along with AKT homolog 1 (AKT1) and PIK3 catalytic alpha polypeptide (PIK3CA) mutations in bladder urothelial carcinoma (UC).
- • Published information is limited to a few studies looking into single or various combinations of members of this pathway with inconsistent results. In particular the expression status of phosphorylated (p-)p70S6 kinase (p70S6K) and p85a subunit of PI3K has not been tested in UC.
PATIENTS AND METHODS
- • Paraffin-embedded transurethral resection tissue from 113 patients with UC was investigated for the association of p85aPI3K, p-AKT, p-mTOR, p-p70S6K and p-4E-BP1 (eukaryotic initiation factor 4E-binding protein 1) expression status, as well as PIK3CA and AKT1 mutations with p-extracellular signal-regulated kinase 1/2 (ERK1/2), fibroblast growth factor receptor 3 (FGFR3), pathological features, recurrence and cancer-specific survival.
- • With the exception of p-p70S6K, all others components of the PI3K/AKT/mTOR pathway were upregulated in UCs as compared with normal urothelium.
- • p-mTOR expression strongly correlated with its upstream p-AKT and marginally with its downstream p-p70S6K. p85aPI3K and p-ERK1/2 levels were also marginally correlated.
- • PIK3CA and AKT1 mutations were distinctly uncommon and mutually exclusive, without any association with pathological features. However, the presence of AKT1 mutations was associated with increased FGFR3 levels and was restricted to p85aPI3K immunonegative cases, whereas PIK3CA mutant cases had marginally lower p85aPI3K levels.
- • The presence of PIK3CA single or combined with AKT1 mutations was associated with shorter recurrence-free survival in univariate survival analysis. An inverse relationship was established between p-4E-BP1 immunopositivity and histological grade or T category, as well as between p-p70S6K levels and T category, the latter relationship being of marginal significance.
- • p-4E-BP1 nuclear expression was marginally associated with the presence of lymphovascular invasion and adversely affected survival in multivariate, but not in univariate analysis.
- • PI3K/AKT/mTOR signalling components appear to be differentially implicated in urothelial tumorigenesis and, with the exception of p85aPI3K, are unrelated to the PIK3CA or AKT1 mutational status.
- • Our findings propose p-4E-BP1 as a potential prognostic marker in UC independent of its association with pathological features, which might assist the selection of patients more likely to benefit from PI3K/AKT/mTOR axis inhibition.
- • PIK3CA/AKT1 mutational status may have a place in the prediction of time-to-recurrence.
v-akt murine thymoma viral oncogene (homolog 1)
eukaryotic initiation factor 4E-binding protein 1
eukaryotic translation initiation factor 4E
extracellular signal-regulated kinase
fibroblast growth factor receptor
high-resolution melting analysis
mitogen-activated protein (kinase)
mammalian target of rapamycin (complex 1) (complex 2)
phosphatase and tensin homologue
proto-oncogene serine/threonine-protein kinase
rat sarcoma viral oncogene homolog
single-strand conformation polymorphism
tuberous sclerosis complex
The pathogenesis of bladder urothelial carcinoma (UC) is driven by two divergent molecular routes. The first route is characterised by gain-of-function mutations in oncogenes, e.g. H-ras, fibroblast growth factor receptor (FGFR)3 and less often phosphoinositide 3-kinase (PI3K) and gives rise to superficial papillary tumours, which frequently recur but are rarely lethal. The second route entails the inactivation, usually via loss-of-function mutations of tumour suppressor genes, e.g. p53, phosphatase and tensin homologue (PTEN) and retinoblastoma (RB1) and leads to the formation of highly invasive and metastatic tumours . Unravelling the molecular alterations underlying the pathogenesis of bladder neoplasms has propelled the identification of suitable prognostic and predictive biological markers that can also be targeted for therapeutic purposes.
The phosphoinositide 3-kinase/v-akt murine thymoma viral oncogene/mammalian target of rapamycin (PI3K/AKT/mTOR) signalling cascade plays a pivotal role in orchestrating the regulation of cell growth, proliferation, glucose uptake and survival through multiple direct and indirect mechanisms . PI3Ks are major effectors of receptor tyrosine kinases, e.g. endothelial growth factor receptor, fibroblast growth factor receptor, type 1 insulin-like growth factor receptor, platelet-derived growth factor receptor) and transduce signals by generating phosphatidylinositol triphosphate (PIP3), which in turn recruits the serine-threonine kinase AKT to the cell membrane for phosphorylation . Full AKT activity is achieved by phosphorylation at both Thr308 in the activation loop by phosphoinositide-dependent protein kinase 1 and Ser473 by mTOR complex 2 (mTORC2). AKT regulates the activity of a wide array of cytosolic as well as nuclear targets including other kinases and transcription factors. mTOR, a serine/threonine kinase existing in two different complexes, mTOR complex 1 (mTORC1) and mTORC2, is a strategic downstream effector of this pathway. The formation of mTORC1 allows for phosphorylation of p70S6K serine/threonine kinase and initiation binding protein eukaryotic initiation factor 4E-binding protein 1 (4E-BP1), with consequent enhanced translation of 5′ terminal oligopyrimidine tract mRNAs, inactivation of Bcl-2-associated death protein (Bcl-2-associated death promoter) and release of eukaryotic initiation factor-4E required for assembly of the translation initiation complex . Aberrant activation of the pathway presents as either gain of growth-promoting function or loss of inhibitory function and may affect the cell dynamics facilitating neoplastic transformation.
Class IA PI3K proteins consist of a catalytic (p110α) and a regulatory (p85, p50, p55) subunit encoded by PIK3 catalytic alpha polypeptide (PIK3CA) and PIK3 regulatory subunit, polypeptide 1 (PIK3R1), respectively. The primary function of p85 is to bind, stabilise and inhibit the p110 subunit until RIK attraction . Oncogenic mutations of PIK3 commonly abrogate the inhibitory function of p85 subunit, thus leading to unrestricted constitutive activity of p110 . Activating mutations in PIK3CA as well as in AKT1 are frequent genomic aberrations promoting tumorigenesis through upregulation of PI3K/AKT signalling cascade. Interestingly, p85PI3K harbours both oncogenic and tumour suppressive properties, exerting a dual effect on signalling . It has been reported that FGFR3 is able to transduce mitogenic signals probably via either PI3K/AKT or mitogen-activated protein kinase (MAPK) pathways and the effects of mutant FGFR3 are cell-type specific . FGFR3 is one of the most frequently mutated genes in UC  and mutations result in overexpression of the respective protein .
Recently, we reported that activation of extracellular signal-regulated kinase (ERK) may promote aggressive behaviour in UC and is not correlated with FGFR3 overexpression . In the present investigation, we pursued our analysis in the same series of UCs of the PI3K/AKT/mTOR pathway, which is also thought to be critically involved in the oncogenesis of UC. The aims were as follows: first, to search for PIK3CA and AKT1 mutations in relation to the expression status of several AKT/mTOR pathway molecules, phosphorylated (p-) extracellular signal-regulated kinase 1/2 (ERK1/2) and FGFR3; second, to examine the relationships of PI3K/AKT/mTOR components with clinicopathological data; and third to determine the potential prognostic utility of these molecules, including PIK3CA and AKT1 mutations in UC.
PATIENTS AND METHODS
This study comprises 113 consecutive patients with newly diagnosed bladder UC, treated and followed-up at Asklepeion Voula and IKA Hospitals in Athens between 1985 and 1995, for whom sufficient paraffin-embedded transurethral resection tissue was available. Cases with upper urinary tract UC were not included in this investigation. Informed consent was obtained from all patients before enrolment in the study. The study follows the principles of the Declaration of Helsinki and was approved by the Ethical Committee of the University of Athens Medical School. All cases were reviewed by two experienced pathologists (G.L., P.K.) and assigned a histological grade and T-category according to the WHO 2004 classification . Staging was based on a combination of histological data and CT. Follow-up information was available in 75 patients. The clinicopathological characteristics of the patients are given in Table 1.
|Patients age, years||68 (44–89)|
|Follow-up, months||40.1 (3.5–82.8)|
|Tumour size, mm||1.5 (0.1–25)|
|Low malignant potential||29 (25.6)|
|Low-grade carcinomas||54 (47.8)|
|High-grade carcinomas||30 (26.5)|
|Median (range) follow-up: 42.3 (12–86.7) months|
|Dead from disease||61 (78.5)|
|Median (range) follow-up: 16.5 (3.5–55.1) months|
|Presence||33 (38), median recurrence time: 6 months|
|Superficial tumours (Ta–T1)**:|
|Maximal transurethral resection (TUR)||79 (100)|
|Intra-vesical chemotherapy||44 (55.6)|
|Muscle-invasive tumours (T2–T4):|
|Intra-vesical chemotherapy||23 (69.7)|
Although 33.6% (38/113) of patients were lost to follow-up there was not any significant difference between the characteristics of those patients and those included in survival analysis (P > 0.10, data not shown). Moreover, the absence of survival data was not correlated with histological grade and T-category (P > 0.10, data not shown).
For genomic DNA isolation, sections (10 µm) were cut from paraffin-embedded tissues and treated with xylene/ethanol. DNA was extracted using standard protocols.
For PCR, 200 ng of genomic DNA was amplified in a 25 µL reaction mixture. The profile used in the Progene Techne thermal cycler was: 2 min at 94 °C; 30 s at 94 °C, 40 s at 58 °C, 40 s at 72 °C for 40 cycles; 7 min at 72 °C. The sequences of the primers are shown in Table 2.
|Variable||No. tumours||Total mutations, n (%)||No. of mutated tumours (%)|
|LMP||23||2 (7.7)||1 (3.9)||1 (3.9)|
|low grade||49||4 (8.2)||1 (2)||3 (6.1)|
|high grade||26||2 (8.7)||1 (4.4)||1 (4.4)|
|Ta–T1||73||7 (4)||2 (2.7)||5 (6.%)|
|T2–T4||25||1 (9.6)||1 (4)||0|
|absence||53||1 (2.4)||0||1 (2.4)|
|presence||33||3 (9.6)||1 (3.2)||2 (1.4)|
|Primers for PCR|
|PIK3CA exon 20||F-CTC TGG AAT GCC AGA ACT AC|
|R-ATG CTG TTT AAT TGT GTG GAA G|
|PIK3CA exon 9||F-ATC ATC TGT GAA TCC AGA|
|R-TTA GCA CTT ACC TGT GAC|
|AKT1 exon 4||F-GGG TCT GAC GGG TAG AGT G|
|R-TCT TGA GGA GGA AGT AGC GT|
|Primers for HRMA|
|PIK3CA exon 9||Forward: 5′-ATC ATC TGT GAA TCC AGA-3′|
|Reverse: 3′-TTA GCA CTT ACC TGT GAC-5′|
|PIK3CA exon 20||Forward:5′-CTC TGG AAT GCC AGA ACT AC-3′|
|Reverse:3′-ATG CTG TTT AAT TGT GTG GAA G-5′|
|AKT1||Forward:5′-GGG TCT GAC GGG TAG AGT G-3′|
|Reverse:3′–TCT TGA GGA GGA AGT AGC GT-5′|
The PCR products were screened for mutations in exons 9 and 20 of the PIK3CA by single-strand conformation polymorphism (SSCP) as previously described . DNA extracted from human colon cancer cell lines HCT116 and SW480 were used as positive and negative controls for the analysis of PIK3CA exon 20 respectively, as well as negative controls for exon 9. The results were verified by independent PCR reactions at least twice.
PIK3CA exon 9 and 20, and AKT1 were screened for mutations using high resolution melting analysis (HRMA) on a Light Cycler 480 (Roche Diagnostics, GmbH, Germany) in duplicate, according to the manufacturers protocol.. The sequences of the primers for PIK3CA exon 9 and 20 and AKT1 are shown in Table 2.
AKT1 mutations observed by HRMA, were identified by Pyrosequencing using the Pyromark Gold Q24 Reagents kit with the Q24 Pyrosequencer (Qiagen GmbH, Hilden, Germany) according to the manufacturer's protocol.
The PCR products with an abnormal SSCP band pattern were sequenced using the Big Dye terminator cycle sequencing kit (Applied Biosystems, CA, USA). The sequencing products were analysed on an ABI Prism 310 Genetic Analyser (Applied Biosystems). PCR primers were also used for sequencing analysis to confirm the presence of mutations. Results were verified by forward and reverse sequencing of at least two independent PCR products.
Immunostaining was performed on paraffin-embedded 4-µm sections of formalin-fixed tumour tissue using the two-step peroxidase conjugated polymer technique (DAKO Envision kit, DAKO, Carpinteria, CA). The primary antibodies used are listed in Table 3. In negative controls primary antibodies were substituted with non-immune serum.
|Protein||No of cases||Clone||Company||Catalog no.||Raised in||Positive controls||Antigen retrieval method||Dilution and incubation time for immunohistochemistry|
|p-mTOR (Ser2448)||72||monoclonal||Cell Signaling Technology, USA||#2976||Rabbit||Human breast cancer||Citrate buffer, pH 9||1:50, overnight|
|p-p70S6K (Thr421/Ser424) [specific for p70 subunit]||72||polyclonal||Santa Cruz Biotechnology, USA||sc-7984-R||Rabbit||Human colon carcinoma||Citrate buffer, pH 6||1:250, overnight|
|p-4E-BP1 (Thr37/46) 236B4||70||monoclonal||Cell Signaling Technology, USA||#2855||Rabbit||Human breast cancer||Citrate buffer, pH 6||1:800, overnight|
|AKT-pS473 [phosphorylation site specific]||74||Polyclonal, Clone 14-5||Dako, Dakopatts||M3628||Rabbit||Mouse kidney||Citrate buffer, pH 9||1:20, 1 h|
|p85a (B-9) [specific for a-subunit of p85]||67||monoclonal||Santa Cruz Biotechnology, USA||sc-1637||Mouse||Normal tonsillar tissue||Citrate buffer, pH 9||1:50, overnight|
Two pathologists (E.T., P.K.) viewed the first 10 cases for each antibody together to obtain consensus for the evaluation, without knowledge of the clinical information. The remaining cases were then examined by one pathologist (E.T.) and of those another 10 cases were checked for reproducibility by the second pathologist (P.K.). Nuclear and cytoplasmic immunoreactivity was recorded separately. For statistical analysis only the predominant pattern, i.e. cytoplasmic for p-mTOR and p-AKT and nuclear for p-p70S6K, p85aPI3K and p-4E-BP1, was considered. A Histo-score (H-score) based on the percentage of stained neoplastic cells (labeling index) multiplied by staining intensity was calculated, as previously described . In addition, the expression levels of p-ERK1/2 and FGFR3 in 107 and 93 of cases respectively were available from our previous investigation .
In the basic statistical analysis, p-mTOR, p-p70S6K, p-AKT and p85aPI3K H-scores were treated as continuous variables, whereas p-p4EB1 as categorical, i.e. positive or negative consistent with previous studies . Correlations among the immunohistochemical expression of the investigated proteins as well as with clinicopathological parameters, AKT1/PIK3CA mutational status, p-ERK1/2 and FGFR3 expression were tested using non parametric tests (Kruskal–Wallis anova, Mann–Whitney U-test, Spearman's rank correlation coefficient, Fisher's exact test and shi-square as appropriate).
Survival analysis was performed using death from disease as the endpoint for cancer-specific survival (CSS) and recurrence as the endpoint for recurrence-free survival (RFS). The effect of various parameters on clinical outcome was assessed by comparing groups using the log-rank test. Multivariate analysis was performed using Cox's model. All results with a two-sided P≤ 0.05 were considered statistically significant.
PIK3CA AND AKT1 MUTATIONAL ANALYSIS (TABLE 2)
In all, 98 samples from patients with UC were screened for the presence of activating mutations at exon 4 of AKT1 by SSCP and/or HRMA. A transition GA at nucleotide 49 (c.49G>A), leading to a glutamic acid to lysine substitution at hotspot codon 17 (p. E17K) was found in three cases (3%; Fig. 1).
Mutations in hotspot exons 9 and 20 of PIK3CA gene were detected by SSCP and HMRA and identified with sequencing analysis. Four different mutations were detected among the tested samples (4.1%); two transitions, a AG at nucleotide 3140 (c. 3140A>G) leading to a histidine to arginine substitution at hotspot codon 1047 (p.H1047R) and a GA at nucleotide 3103 (c. 3103G>A), leading to an alanine to threonine substitution at codon 1035 (p.A1035T) and two transversions a CT at nucleotide 3137 (c. 3137C>T), leading to an alanine to valine substitution at codon 1046 (p.A1046V) and a GC at nucleotide 3145 (c. 3145G>C) leading to a glycine to arginine substitution at codon 1049 (p.G1049R). These three last mutations have not been reported before in bladder UC according to the Catalogue Of Somatic Mutations In Cancer (COSMIC) database (Fig. 1). Furthermore, one case had a mutation in exon 9 identified as a transition GA at nucleotide 1624 (c. 1624 G>A), which leads to a glutamic acid to lysine substitution at hotspot codon 542 (p. E542K).
PI3K/AKT/MTOR PATHWAY IN UROTHELIAL TUMOURS AND NORMAL UROTHELIUM (FIG. 2)
p-mTOR immunoreactivity was cytoplasmic and/or membranous in 63/72 cases (87.5%) and in 22/72 cases (30.5%) also nuclear. Nuclear p-p70S6K and p-4E-BP1 immunoexpression was seen in 71/72 cases (98.6%) and 53/70 cases (75.7%) respectively, with occasional cases (9/72, 12.5% for p-p70S6K and 15/70, 21.4% for p-4E-BP1) displaying also cytoplasmic positivity. Tumour infiltrating lymphocytes and endothelial cells expressed both proteins and served as internal positive controls in each case. In all, 60 cases (60/67, 89.5%) coexpressed p-p70S6K and p-mTOR, with seven p-p70S6K positive cases being negative for p-mTOR (7/67, 10.4%). However, five of these seven cases had cytoplasmic p-AKT. Six cases positive for p-4E-BP1 but negative for p-mTOR were positive for pERK1/2. The percentage of coexpression of all three components of the mTOR cascade rose to 66.15% (43/65 cases). None of our cases was triple negative for p-mTOR pathway components, whereas a single case positive for p-mTOR but double negative for p-p70S6K and p-4E-BP1 expressed p-AKT.
p-AKT immunoreactivity was cytoplasmic (65/74, 87.8%) with 18 cases (18/74, 24.3%) also showing nuclear immunoexpression. Cytoplasmic p-AKT was coexpressed with all three mTOR pathway components in 39/65 cases (60%). p85apI3K nuclear expression was recorded in 35/67 cases (52.2%), with 16 cases (16/67, 23.9%) also displaying cytoplasmic immunoexpression. p85apI3K was coexpressed with pAKT in 31/67 (46.3%) cases. Coexpression of all five molecules under study was seen in 27/62 cases (43.5%).
Normal urothelium had lower p85apI3K, p-AKT and p-mTOR H-scores than the UC cells (p85apI3K 13 cases, P < 0.001; p-AKT 20 cases, P= 0.052; and m-TOR 15 cases, P < 0.001, Mann–Whitney U-test). It is to be noted that p-mTOR staining was restricted to superficial cells. For p-p70S6K, there was no significant difference in H-scores between normal and neoplastic urothelial cells (13 cases, P > 0.10, Mann–Whitney U-test), whereas normal urothelium less often displayed p-4E-BP1 immunoreactivity than the UC cells (P= 0.034, Fischer's exact test; Fig. 3).
The correlations among these proteins are shown in Table 4. A significant correlation was only established between p-mTOR and cytoplasmic p-AKT, whereas the correlation between p-mTOR and p-p70S6K attained a marginal significance.
|p-mTOR cytoplasmic expression||p-p70S6K nuclear expression||p-p4E-BP1 nuclear expression||p-AKT cytoplasmic expression||p-85PI3K nuclear expression|
|p-p70S6K nuclear||R = 0.2243|
|p-p4E-BP1 nuclear||R =−0.023||R = 0.023|
|P > 0.10||P > 0.10|
|p-AKT cytoplasmic||R = 0.528||R = 0.150||R =−0.032|
|P < 0.001||P > 0.10||P > 0.10|
|p-p85PI3K nuclear||R = 0.206||R = 0.076||R =−0.233||R = 0.269|
|P > 0.10||P > 0.10||P > 0.10||P > 0.10|
|p-ERK cytoplasmic||R = 0.097||R = 0.189||R =−0.018||R = 0.074||R = 0.299|
|P > 0.10||P > 0.10||P > 0.10||P > 0.10||P= 0.086|
|p-ERK nuclear||R =−0.121||R = 0.087||R =−0.101||R =−0.022||R = 0.235|
|P > 0.10||P≥ 0.10||P > 0.10||P > 0.10||P < 0.10|
|FGFR3||R =−0.116||R = 0.150||R = 0.170||R = 0.031||R =−0.112|
|P > 0.10||P > 0.10||P > 0.10||P > 0.10||P > 0.10|
ASSOCIATIONS BETWEEN PI3K/AKT/MTOR PATHWAY COMPONENTS AND CLINICOPATHOLOGICAL FEATURES (TABLE 5, FIGS 3,4)
|Median (range), positive/negative, n|
|LMP||17.5 (0–120), 17/18||180 (120–200), 18/18||4 (0–60), 13/16||20 (0–60), 16/18||2 (0.5–30), 7/16|
|Low grade||10 (0–180), 29/33||190 (7.5–297), 33/33||2 (0–75), 29/33||12.5 (0–160), 29/35||2 (0.1–100), 21/31|
|High grade||20 (0–100), 17/21||170 (0–200), 20/21||0.5 (0–60), 11/21||40 (0–170), 20/21||7.5 (0–45), 7/20|
|Ta–T1||10 (0–180), 44/49||185 (7.5–297), 50/50||2 (0–75), 40/48||16.25 (0–160), 45/52||2 (0.25–100), 27/46|
|T2–T4||10 (0–80), 19/23||160 (0–200). 21/22||0.75 (90–60), 13/22||40 (0–170), 20/22||0.5 (0.1–45), 8/21|
|Total||10 (0–180), 63/72||180 (0–297), 71/72||2 (0–75), 53/70||20 (0–170), 65/74||2.25 (0.1–100), 35/67|
|Absence||10 (0–100), 26/34||170 (0–200), 32/33||1 (0–60), 25/33||16.25 (0–170), 28/35||0 (0–20), 15/33|
|Presence||20 (0–140), 20/21||180 (10–200), 21/21||2 (0–75), 15/21||10 (0–160), 21/22||1 (0–30), 13/21|
p-4E-BP1 positivity was inversely associated with tumour histological grade (Fisher's exact test, low malignant potential vs low grade vs high grade, P= 0.014). Moreover, an inverse correlation emerged between p-4E-BP1 positivity or p-p70S6K expression and tumour T-category (Mann–Whitney U-test Ta/T1 vs T2-T4, P= 0.060 for p-p70S6K and chi-square test P= 0.028 for p-4E-BP1), whereas p-4E-BP1 positivity prevailed in papillary in contrast to solid tumours (Fisher's exact test, P= 0.006) and marginally in cases with lymphovascular invasion (Fisher's exact test, P= 0.057). All other relationships with clinicopathological characteristics were not significant.
ASSOCIATIONS BETWEEN PI3K/AKT/MTOR PATHWAY COMPONENTS AND P- ERK1/2 OR FGFR3 EXPRESSION
Nuclear and cytoplasmic p-ERK1/2 was coexpressed with all three mTOR pathway components in 32/49 cases (65.3%). The correlations among PI3K/AKT/mTOR signaling components, p-ERK1/2 and FGFR3 are presented in Table 4. There was marginal correlation between p85aPI3K and pERK1/2 levels.
ASSOCIATIONS BETWEEN THE PI3K/AKT/MTOR PATHWAY COMPONENTS, PERK1/2 AND FGFR3 EXPRESSION WITH AKT1 AND/OR PIK3CA (EXON 20 AND 9) MUTATIONS STATUS
The presence of AKT1 mutations was marginally correlated with increased FGFR3 expression (P= 0.053). The respective correlations between AKT1 and/or PIK3CA (exon 9 and 20) mutational status and the pathological features, most PI3K/AKT/mTOR pathway components or p-ERK1/2 expression were not significant (P > 0.10). Notably, all p85aPI3K immunopositive cases showed no AKT1 mutations, the latter being restricted to p85aPI3K immunonegative cases. PIK3CA mutated cases had marginally lower p85aPI3K H-scores (P= 0.066).
In univariate survival analysis for CSS the H-score of the investigated proteins and the presence of AKT1 or PI3KCA mutations were not significant. Results of log-rank test are summarised in Table 6. Only histological grade, T-category, tumour configuration, tumour size and patients' age adversely affected CSS (Table 6). On the other hand, the presence of PIK3CA mutations, single or combined with AKT1 mutations, adversely affected recurrence time (P= 0.05 for both associations), as all mutated cases had recurred (Fig. 5).
|Variable||Univariate survival analysis for CSS (log-rank test), P||Univariate survival analysis for RFS (log-rank test), P||Multivariate survival analysis* for CSS, P|
|AKT1 mutation (n= 63) (absence vs presence)||0.572||–||>0.999|
|PI3KCA mutation (n= 63) (absence vs presence)||0.581||0.05||>0.999|
|AKT1/ PI3KCA mutation (n= 63) (absence vs presence)||0.425||0.05||>0.999|
|p-mTOR cytoplasmic score (n= 46) (<10 vs ≥10)||0.764||0.116||0.714|
|p-p70S6K nuclear score (n= 45) (<270 vs ≥270)||0.441||0.503||0.248|
|p-4E-BP1 nuclear score (n= 46) (0 vs ≥1)||0.513||0.206||0.048|
|pAKT cytoplasmic score (n= 44) (<20 vs ≥20)||0.691||0.217||0.898|
|p-p85PI3K nuclear score (n= 45) (0 vs ≥1)||0.538||0.237||0.408|
|Histological grade (LMP vs LG vs HG)||<0.001||0.001|
|T- category (Ta–T1 vs T2–T3)||<0.001||<0.001|
|Tumour configuration (papillary vs solid)||0.001||0.623|
|Age (years) (<68 vs ≥68)||0.025||0.613|
|LVI (absence vs presence)||0.177||0.163|
Due to the few events encountered in our cohort we adjusted separate (Cox's) hazard ratio (HR) models of survival for each of the immunohistochemically investigated proteins and the mutational status of the tumours. In each model, histological grade, T-category, patient's age and tumour size were introduced to evaluate the adjusted effect of the investigated parameter on patients' CSS in each run. Results of these analyses are shown in Table 6. The presence of p-4E-BP1 expression emerged as an independent adverse prognosticator (HR 9.45, P= 0.048) along with histological grade (P= 0.006) and T-category (P= 0.003) (Fig. 6).
The PI3K/AKT/mTOR signalling axis has attracted intense scientific interest in bladder UC because PTEN, a tumour suppressor gene the protein product of which is responsible for the degradation of PIP3 and, hence, the inhibition of the pathway, is deregulated in almost one-third of UCs . Therefore, bladder UC has been considered a suitable candidate for the therapeutic incorporation of inhibitors (e.g. rapamycin) targeting various components of PI3K/AKT/mTOR cascade. Several lines of experimental evidence support this argument documenting that rapamycin induced a decrease in cell viability, cell migration and angiogenesis in UC xenografts in animal models [15–19], while preventing progression from carcinoma in situ to invasive UC and suppressing tumour recurrence . Overall, these findings indicate the importance of mTOR pathway activity for UC in vitro. Although, it is noteworthy that the expression status of p-AKT and p-mTOR showed no association with the sensitivity of UC cell lines to rapamycin , which underscores the need to evaluate a wide spectrum of molecular targets lying upstream and downstream of mTOR, for the selection of tumours more likely to benefit from mTOR inhibition. Although a few studies have invesigated single or various combinations of PI3K/AKT/mTOR pathway components [9,16,17,19,22–24], a comprehensive simultaneous assessment of all key members of this pathway (i.e. p85a, pAKT, p-mTOR, p-p70S6K and p-4E-BP1) and AKT-1 or PIK3CA mutations in relation to FGFR3 and pERK 1/2 expression status in a well-characterised set of patients withUC has not been performed. In particular, the expression status of p-p70S6K and the regulatory p85a subunit of PI3K have not been explored in UC.
Activating mutations of PIK3CA are clustered in two exons (9 and 20) encoding for the helical and kinase domain and result in augmented enzymatic activity. In UC the reported incidence of PIK3CA mutations ranges from 13% to 25% . The lower frequency of PIK3CA mutations (5%) recorded in the present study could be attributed to ethnic variation and population-based differences, intratumoral heterogeneity as well as the presence of contaminating normal cells within a tumour, as microdissection was not performed for tumour enrichment of the samples. Most of the observed mutations (80%) were clustered in exon 20 of PIK3CA, whereas only one of them was located in exon 9. Interestingly, we detected three mutations that have not previously been found in bladder UC. An intriguing finding was the marginally lower p85aPI3K H-scores displayed by PIK3CAmut cases, which may relate to the primary inhibitory effect of p85 upon p110. In agreement with Juanpere et al. no relationship was elicited between PIK3CA mutations, grade or stage. As far as AKT1 mutations are concerned, p.E17K was the only alteration detected in 3% of the cases. This frequency is consistent with currently published data in UCs [25,26]. In contrast to previous studies , AKT1 mutations in the present series were not restricted to high-grade UCs but were equally distributed among low malignant potential, low-grade and high-grade carcinomas or T categories. There was also no relationship between the presence of AKT1 mutations and p-AKT overexpression but AKT1 mutated cases had higher levels of FGFR3 protein. In this context, pAKT overexpression has been reportedly correlated with FGFR3 mutations . A novel finding of the present study was the adverse prognostic effect of PIK3CA mutations alone or combined with AKT1 mutations on recurrence rate.
PIK3CA and AKT1 mutations were not found simultaneously in the present cohort in accordance with two previous studies in bladder UC [9,27], indicating that alterations in the two genes are mutually exclusive in carcinogenesis. Relevant to this notion may be the present finding that AKT1mut genotype was restricted to p85aPI3K immunonegative cases. As far as the mutations in the two examined exons of PIK3CA gene are concerned, they are thought to act synergistically when simultaneously present in the same p110 molecule, as they may trigger gain-of-function through two different and independent mechanisms .
In the present cohort, immunohistochemistry showed that a significant percentage of cases (60%) coexpressed p-mTOR along with its two downstream targets and its upstream p-AKT, denoting that activation of p-AKT/mTOR axis is a common event in UC and disputing the earlier assumption made by Schultz et al. regarding down-regulation of this pathway in UC. It is noteworthy that the rate of p-p70S6K positivity (98.6%) far exceeded that of p-mTOR (87.5%), implying that other pathways may also contribute to p70S6K phosphorylation, presumably by MAPK through ribosomal S6 kinase p90rsk  or directly by AKT through tuberous sclerosis complex (TSC)1/2 . This explanation is reinforced by the present finding that most of p-p70S6K positive/p-mTOR negative cases expressed cytoplasmic pAKT and fits well with the marginally significant correlation between p-p70S6K and p-mTOR levels recorded in the present series. In this context, a reciprocal interplay is known to exist between mTOR and p70S6K, in the sense not only mTORC1 modulates phosphorylation of S6K at Thr389 but is also phosphorylated by the latter at Ser2448 in response to both mitogen- and nutrient-derived stimuli . The strong positive correlation between p-AKT and p-mTOR established in the present cases lends further support to mTOR being an important effector of AKT in bladder UC, as reported in prostate cancer . The interconnection between the mTOR cascade and PI3K/AKT pathway involves at least two mechanisms, i.e. direct phosphorylation of mTOR by AKT itself, as well as via phosphorylation and inactivation of mTOR inhibitor TSC2 [33,34]. Schultz et al. also failed to elicit a correlation between p-AKT and pS6-the downstream target of p70S6K or total 4E-BP1, in harmony with the present findings, whereas two other studies have reported a positive albeit weak correlation of p-mTOR with pS6 and p-4E-BP1 [22,23]. It thus seems that a one-to-one strong relationship between any two components in PI3K/AKT/mTOR pathway is rather unlikely to be obtained, as their activation depends on the equilibrium of a complex interaction of multifactorial oncogenic events [19,22]. Along this line, in vitro data have substantiated the role of ERK signalling in regulating 4E-BP1 , which accounts for the expression of pERK1/2 in all the present cases with p-mTOR negative/p-4E-BP1 positive phenotype. We also report for the first time a marginal correlation between p85apI3K and pERK1/2 expression levels. Given that p85aPI3K H-scores were lower in PIK3CAmut cases, this finding is in broad agreement with the reported association between wild PIK3CA, AKT1 or FGFR3 genotypes and pERK1/2 overexpression  and consistent with the interaction of PI3K/AKT and the proto-oncogene serine/threonine-protein kinase (RAF)/MAPK/ERK kinase (MEK)/ERK cascades through rat sarcoma viral oncogene homolog (RAS) .
Regarding the expression of the molecules under study in normal urothelium, the rate of p-4E-BP1 immunopositivity and the levels of p85apI3K, p-AKT and p-mTOR immunoreactivity were considerably higher in neoplastic urothelial cells than in the benign cohort, arguing in favour of an overall up-regulation of PI3K/AKT/mTOR signalling in urothelial tumorigenesis, including p85aPI3K and with the exception of p-p70S6K. Reported results on this issue have been quite inconsistent. Thus, p-mTOR  and p-4E-BP1  expression levels were higher in UCs, concurring with the present findings, but pS6 was underexpressed [23,24]. AKT, on the other hand, was reported to be down-regulated by Schultz et al. but up-regulated by Sun et al. in UC samples. The concept of p-4E-BP1 being essential for neoplastic transformation is corroborated by experimental data showing that transfer of 4E-BP1 phosphorylation site mutants into breast cancer cells suppressed their tumorigenicity  and overexpression of its partner eukaryotic translation initiation factor 4E (eIF-4E) is capable of inducing malignant transformation of mammalian cells .
Among the studied molecules, p-4E-BP1 immunopositivity prevailed in low grade/stage cases, in agreement with Sun et al. and, in a broad sense, with Schultz et al.. A similar inverse relationship, although of marginal significance, was recorded between p-p70S6K levels and T-category in the present series. Considering that normal urothelium had significantly less p-4E-BP1 immunoreactivity, the present findings advocate that its up-regulation is critical for the development of superficial low-grade UCs but becomes less influential for the acquisition of an invasive phenotype. Along this line, down-regulation of mTOR pathway has been theoretically attributed to the hypoxic-tolerant phenotype of aggressive UCs . mTOR, p85aPI3K and AKT activation, on the contrary, which also appears to take place at an early stage of urothelial tumorigenesis is sustained throughout the neoplastic process, as no difference existed between low- and high-grade/stage cases. Published studies have yielded conflicting findings as to the relationship of AKT/mTOR pathway with the pathological features of UCs, especially for pS6 and p-AKT, some investigators reporting an inverse relationship  and others a positive [22,23,39] or even no relationship [9,16]. The overall impression gained from these findings is the diverse role of PI3K/AKT/mTOR signalling components during the development of the neoplastic process in UCs.
Perhaps one of the most important findings emerging from the present investigation is the adverse prognostic significance of p-4E-BP1 expression established in multivariate analysis along with tumour grade and T-category. Analogous findings have been reported in other solid tumours . The fact that p-4E-BP1 attained statistical significance only in multivariate analysis adjusted for tumour grade and T-category obviously reflects its inverse relationship with these parameters masking its intrinsic adverse prognostic effect. The latter is also illustrated in its association with the presence of lymphovascular invasion. The prognostic relevance of various PI3K/AKT/mTOR pathway members in UC has been dealt with in four reports [17,22–24] and remains a matter of debate. One initial report assigned a favourable prognostic effect to pS6 over-expression , which subsequently was proposed as an adverse prognosticator by Sun et al. and Park et al. along with p-mTOR and p-AKT  or p-4E-BP1 . Our results conform to those of the latter group , although we were not able to reproduce the prognostic significance of p-p70S6K. A similar dissociation between p-4E-BP1 and p-p70S6K for prognosis has also been seen in breast cancer  and astrocytomas .
The oncogenic effects of 4E-BP1 phosphorylation are entangled with the release of eIF-4E with subsequent preferential enhancement of the synthesis of growth promoting or oncogenic proteins (including angiogenic factors, FGF, myc, matrix metalloproteinase 9, cyclin D1) thereby facilitating angiogenesis, invasion and metastasis . Given that activation of 4E-BP1 represents the convergence point of several oncogenic pathways besides mTOR , it would be expected to provide a better reflection of tumour aggressiveness than the triggering genetic alteration upstream. This may explain why p-mTOR proved inferior to p-4E-BP1 as a prognosticator in the present study. Modulation of 4E-BP1 function gains further importance in light of recent data substantiating its usefulness as a determinant of the sensitivity of tumour cells to PI3K/AKT signalling inhibitors .
The controversy surrounding the clinicopathological significance of various PI3K/AKT/mTOR signalling components in UC may lie in technical issues, including the type of antibodies used (against phosphorylated or total molecules), staining dilutions and quantitation methods, the relative proportion of cases within various grades and stage categories and the use of various thresholds for the categorisation of immunoreactivity. Also, in the present study nuclear distribution of p-4E-BP1 and p85aPI3K prevailed over the cytoplasmic one, in contrast to previous studies [22,23]. Given that these proteins shuttle between the nucleus and the cytoplasm , it is not surprising that some authors have relied upon the nuclear compartment while others upon the cytoplasmic compartment for interpretation of the results. We have most recently reported that the nuclear p-4E-BP1 immunoreactivity is an adverse prognosticator in human astrocytomas . It therefore seems that the subcellular localisation of these proteins may be of biological importance.
In summary, PI3K/AKT/mTOR signalling components appear to be differentially implicated in urothelial tumorigenesis and, with the exception of p85aPI3K, are largely unrelated to the PIK3CA or AKT1 mutational status, which appears to be of potential importance only for recurrence. More importantly, the present data bring forward nuclear p-4E-BP1 immunoexpression as a molecular marker of prognostic value in these tumours. Validation of these findings in larger scale, preferably prospective studies allowing for subgroup analysis is warranted before p-4E-BP1 is included in the arsenal of molecular predictive and prognostic markers in UC, to assist the selection of patients more likely to benefit from therapy with PI3K/AKT/mTOR inhibition.
CONFLICT OF INTEREST
- 12World Health Organization Classification of Tumours. Pathology and Genetics, Tumours of the Urinary System and Male Genital Track. Lyon: IARC Press, 2004: 89–115, , .