Relevance of immunohistochemical expression of p57kip2 in epithelial ovarian carcinoma- A systematic literature review
© Madhuri et al.; licensee BioMed Central Ltd. 2012
Received: 20 June 2012
Accepted: 26 September 2012
Published: 21 December 2012
Epithelial Ovarian Cancer (EOC) is the second most common gynaecological cancer and accounts for more deaths than all gynaecological cancers combined. Despite extensive research, progress has been slow in understanding the pathobiology. EOC is identified as a heterogeneous malignancy with various histological subtypes. It is now well known that these different histological subtypes show differences in terms of presentation, response to treatment, immunohistochemical (IHC) reactivity and molecular profiling. Cell cycle deregulation is key in cancer development and there is some evidence in the literature that this is relevant to the problem of EOC and the development of drug resistant disease. The need to identify prognostic markers has led to several gene profiling studies using tumour tissue with equivocal results. p57kip2 is one such cell cycle regulator and its functions are being explored as recent research has shown that it is more than just a negative regulator of the cell cycle.
The aim of this review is to evaluate the literature around the IHC expression of p57kip2 in EOC.
Systematic review of the literature focussing on clinical outcome and immunohistochemical expression in epithelial ovarian cancer.
Four papers are discussed in this review and have shown great variation in IHC expression of p57kip2 in EOC. These studies incorporated different histological subtypes of EOC. However they all suggest that p57kip2 has a significant role in prognosis and its therapeutic indication needs to be studied. Multicentre collaborative studies on individual histological subtypes might provide more data and help to increase the number of cases especially for rarer tumours.
KeywordsEpithelial ovarian cancer Pathobiology Histology Immunohistchemistry Cell cycle Cyclin dependant kinase p57kip2
In 2008, epithelial ovarian cancer (EOC) accounted for 225,000 new cases worldwide and causes 140,100 deaths . EOC is the second most common gynaecological cancer and accounts for more deaths than all gynaecological cancers combined. In the UK, there were 6,537 new cases of ovarian cancer diagnosed in 2008 and 4,440 women died of ovarian cancer in 2008 . Treatment for EOC includes surgery and chemotherapy. Although there have been some advances in treatment and management of this disease, more than 75% of cases at presentation are advanced Stage 3/Stage 4 with a very poor 5 year survival (20-30%) .
Despite extensive research, progress has been slow in understanding the pathobiology of the cause and progression of EOC. In their landmark paper on the hallmarks of cancer, Hanahan and Weinberg suggested six essential cell physiology alterations that led to cancer development and followed this with an updated review considering 2 further enabling characteristics including genetic instability and tumour promoting inflammation together with avoiding immune destruction [4, 5]. In particular, and in relation to the present review, the evasion of growth suppressors and the concomitant drive on proliferation involving disruption of the activities of RB and/or p53 have been the subject of many research studies. Hanahan and Weinberg included a focus on therapeutic targeting of the hallmarks of cancer involving cyclin-dependent kinase inhibitors, which is particularly relevant to this review .
Cell cycle deregulation is key in cancer development and there is some evidence in the literature that this is relevant to the problem of EOC and the development of drug resistant disease . Despite good initial response to platinum-based chemotherapeutic agents, prognosis with EOC is poor as there are few effective chemotherapeutic agents available for management of this disease. Moreover, there is a need to identify patients who are likely to relapse whilst undergoing chemotherapy treatment with paclitaxel and carboplatin. The need to identify prognostic markers has led to several gene profiling studies using tumour tissue with equivocal results. One of the criticisms levelled at such studies has been the possibility for contamination of tumour tissue with stromal components. However, it would seem that less effort appears to have gone into identification of molecular markers of prognosis using immunohistochemistry (IHC) of tissue sections and correlation with clinical outcome. This approach has the advantage in that the tumour component of the tissue can be clearly identified.
EOC is identified as a heterogeneous malignancy with various histological subtypes. It is now well known that these different histological subtypes show differences in terms of presentation, response to treatment, IHC reactivity and molecular profiling. Molecular studies have demonstrated this genetic heterogeneity. However, problems have been posed in previous studies that grouped together all subtypes when evaluating the expression of the various cell cycle regulators . Recognition of this has now led researchers down the path of exploring the individual cell cycle regulators in various subtypes of cancers. p57kip2 is one such cell cycle regulator and its functions are being explored as recent research has shown that it is more than just a negative regulator of the cell cycle. The aim of this review is to evaluate the literature around the IHC expression of p57kip2 in EOC. Firstly, p57kip2 and its role in tumorigenesis will be discussed with emphasis on EOC followed by evaluating the reported IHC expression of p57kip2 in EOC to understand the role of p57kip2 better, and finally correlating the IHC expression with reported clinical outcomes.
Cell cycle and p57kip2
Changes in cell cycle regulation can lead to uncontrolled cell proliferation and oncogenesis. The progression of the cell cycle through the various phases is tightly regulated by cyclins and cyclin-dependent kinases (CDKs) and simultaneously counterbalanced by cyclin dependent kinase inhibitors (CDKIs). The transition from the G1 to the S phase is regulated by the CDKIs. The CDKIs are further subdivided into the INK4 family consisting of p16ink4a, p15ink4b, p18ink4c and p19ink4d and the CIP/KIP family consisting of p21 waf1, p27 kip1 and p57kip2. The CIP/KIP family regulates the cell cycle both positively and negatively by interacting with Cyclins A, D & E. Hence p57kip2 levels are highest in the G0 & G1 cell cycle phases in keeping with its role as a CDKI. Its levels also steadily decrease while progressing from G1 through the S cell cycle phase and over expression of p57kip2 induces G1 arrest in cultured cells.
In humans, the p57 kip2 encoding gene (correctly called the CDKN1C gene) is located on chromosome 11p15.5 and is maternally expressed and paternally imprinted . p57kip2 is distributed in the nucleus, however, cytoplasmic distribution has been reported in cancer tissue . It is an important negative regulator and was previously considered to be a tumour suppressor. Recent findings of its involvement throughout the regulation of cell cycle processes suggest a multifaceted existence of the p57kip2 protein .
p57kip2 and embryology
p57kip2 has been found to be essential for normal development. It is found in derivatives from all the three germ cell layers and appears to be irreplaceable by other CDKIs. This has been demonstrated in mouse models wherein p57 kip2 devoid mice die within 2 weeks of age showing dysregulation in cell proliferation, differentiation and apoptosis as compared to other CIP/KIP family proteins which do not appear to be lethal suggesting that the p57kip2cellular function far exceeds that of its CDKI counterparts . Rodent data suggests that p57kip2 is highly cell-specific with a temporal and spatial p57kip2 expression during embryonic and fetal development and continuing on into puberty and adulthood. The highest and maximum widespread tissue distribution occurs during organogenesis when p57kip2 is present in all major organs with its expression peaking at key differentiation stage for each organ. Levels subsequently appear to be undetectable in several organs after birth. Adult human tissue expressing significant p57kip2 mRNA levels include skeletal muscle, brain, heart, lung, kidney, pancreas, testis and placenta with lower expression noted in stomach and ovaries .
Altered p57kip2 and cancer
Abnormal p57kip2 function includes its participation in cancer initiation and progression. Since p57kip2 was discovered as a CDKI, research revolved around describing its ability to inhibit proliferation hence rendering it a tumour suppressor gene status . p57kip2 has also been shown to influence cell differentiation thus positively inhibiting cancer progression and preventing their maintenance in an undifferentiated state. Evading cell death is a significant ability of cancer cells. Conflicting reports have been put forth relating to the apoptotic role of p57kip2. Both pro and anti-apoptotic functions have been described depending on cellular context, regulatory tumour cell pathways and tumour microenvironment. However, this evidence is from mouse models with structural differences compared to humans. Clinical studies suggest that absence of p57kip2 abrogates the pro-apoptotic function of anti-cancer therapy resulting in poor prognosis due to drug resistance development. Vlachos et al. demonstrated sensitization of cancer cells to apoptotic agents such as cisplatin, etoposide and other agents by selective p57kip2 expression .
Cancer metastasis is the ability of the cells to travel and invade neighbouring tissues. This involves regulation of cytoskeletal dynamics, particularly through the Rho family and a negative feedback loop has been identified . Nan et al. showed that decreased p57kip2 expression was associated with increased invasive and metastatic tumour potential in hepatocelluar carcinoma . Angiogenesis is another adaptive process of solid tumours to supply themselves with nutrients and oxygen for their further growth. Matsuura et al. demonstrated increased VEGF (vascular endothelial growth factor) expression in placentae of p57kip2 null embryo mice suggesting a reciprocal correlation between lack of p57kip2 and VEGF expression though further evidence is awaited .
Epigenetic control is important in p57kip2 regulation frequently via promoter methylation of CpG islands . Transcriptional and translational down-regulation of p57kip2 has been demonstrated by Pateras et al. in various cancers with decreased expression associated with aggressiveness . Histone modifications of p57kip2 repression helps participate in tumorigenesis in methylation dependent and in an independent manner. Several studies have shown that p57kip2 gene reactivation after demethylating agent treatment in many cancers demonstrates the epigenetic mechanism of downregulation [6, 10].
p57kip2 and pathology
p57kip2 has been extensively studied in Beckwith-Wiedemann Syndrome (BWS) which is a heterogenous disease. Clinical manifestations of pre-eclampsia have been displayed by mutant mice defective for the maternal p57kip2 allele. Absence of p57kip2 in complete hydatidiform moles which is a trophoblastic proliferation abnormality suggests that p57kip2 is a very sensitive marker to differentiate between complete and partial hydatidiform mole since p57kip2 is a paternally imprinted and maternally expressed gene, it is found in partial moles but absent in complete molar pregnancies .
The role of p57kip2 has been studied to a small extent in cardiovascular and renal pathology as well. Several IHC studies have been carried out evaluating p57kip2 expression in lung, oral, gastrointestinal tract (oesophageal, gastric, colorectal), hepatocellular and pancreatic malignancies . Despite studies highlighting the importance of p57kip2 in the female genital tract pathology, few studies have highlighted the role of p57kip2 as a prognostic molecular marker for EOC as compared to studies reported on p27kip1 and p16 .
Data currently available are limited and conflicting and the role of p57kip2 as an IHC marker of prognosis requires clarification.
Based on the above considerations, we conducted a systematic review with following meta-analysis to clarify the association. In this meta-analysis, we examine studies with IHC data following p57kip2 use. Furthermore, any correlation with clinical outcome is also evaluated to assess the efficacy and usefulness of p57kip2 as a potential prognostic marker.
Thorough literature search of MEDLINE, EMBASE, and COCHRANE electronic databases was performed by 2 independent reviewers to identify relevant studies. This included bibliographies to identify all additional studies. Hand searching of the grey literature was also carried out to identify any relevant conference abstracts and unpublished MD/PhD theses. The search was not limited to any time duration. To enable assimilation of all relevant published research, all search terms were expanded and all sub-categories were included. The exact syntax of search terms included ovarian carcinomas as well as cell membrane protein and other mesh terms.
Initial scanning of the title and abstract was performed to exclude any unrelated studies. Reading of the full text of all remaining articles was performed to determine the extent of information contained in them on the subject of interest.
Inclusion into the current review was based on the following criteria for all retrieved studies:
Tumour site of study performed was ovary. Papers were reviewed to determine whether IHC assessment of p57kip2 had been performed. The studies reported the results of the IHC scoring and an attempt had been made to correlate this with clinical outcomes. Due to the small number of studies, reviews that explained the role of p57kip2 in EOC have also been included following agreement between the authors.
Studies that did not involve EOC, performed only on cell lines, western blotting and gene profiling studies (non IHC studies) have not been reviewed.
Information from the studies was extracted by 2 researchers (TKM & BH) using the data extraction form. Both researchers were in agreement regarding the studies and the data collected included first author’s name, country study was performed in, year of publication, no of cases, IHC findings and clinical outcome if recorded.
Showing the 4 studies with IHC performed
No of EOC cases
Correlation with clinical outcome
Showing histological subtypes of the 4 studies
No of EOC cases
Papillary serous 34
Low vs high Expressors
Poorly differentiated 10
Clear cell 3
Low vs high Expressors
13 Grade 2
13 Grade 3 incl 2 undifferentiated
4 levels of scoring
Clear cell 8
Mixed NOS 1
Mixed without clear cell 4
Mixed with clear cell 4
Unclassifiable adenoca 6
Positive vs negative
Clear cell 1
Non epithelial 8
Immature teratoma 1
Granulosa Cell 3
Metastatic cancer 1
(site not defined)
Patient Demographics recorded in the studies is tabulated below
No of EOC cases
Treatment status at surgery
FIGO Stage of EOC
Other mentioned criteria
2 & 3 only
only G3 mentioned
16-77 Median 49
G1 vs G2/3
Nodal status, ascites, residual disease
21-70 Median 54
G1 vs G2/3
Ascites, residual disease and response to chemotheraphy
18-66 Median 42
Showing results of the 4 studies
p57 kip2 Positivity in EOC
No relation to survival status
21/33 Normal ovarian tissue expression was positive (data not included)
Significant correlation between p57kip2 expression and tumour grades/clinical stages noted. Low p57kip2 expression associated with poor survival. Important prognostic implications exist
<10%Positivity in 130/171 & >10% in 41/171
No correlation exists with clinical and prognostic outcomes
0/10 positivity in normal Ovarian tissue
12/15 Normal ovarian tissue expression was positive in 16/17
Not described as clinical correlation not assessed
In 2007, Khouja et al. published their experience with p57kip2 and other proteins in 171 treatment naïve patients with Stage 3 EOC (Stage 3a- 13, 3b −20 & Stage 3c −138) . Serous adenocarcinoma was the predominant subtype (127/171) and the other subtypes are shown in Table 2. 156/171 consisted of grade 2/3 tumours and the remaining cases (15/171) were Grade 1. Distinct nuclear staining was considered positive. Semiquantitative scoring classes were adopted depending on the number of positive stained tumour cells with 4 scoring groups; no staining, less than 10% positive, between 10% and 50% positive and more than 50%. The IHC results for p57kip2 was negative in 19% of the cases and more than 50% staining was obtained in only 2% (4 cases). This study concluded that there was no correlation between p57kip2 and clinical and prognostic outcomes, Table 4.
The final study by Guo et al. in 2011 evaluated 103 ovarian tissues . Of these, 17 were normal ovaries, 15 benign, 8 borderline, 55 EOC and 8 non-epithelial tumours as shown in Table 2. Positive staining was defined as brown-yellow granules distributed in either the nucleus or cytoplasm with higher intensity of stain than the nonspecific background. Greater than 20% scoring over 5 fields was a positive result.
This study principally focussed on EZH2 (enhancer of zeste homolog 2) and compared p57kip2 IHC expression since p57kip2 is a direct target of EZH2 as shown in other studies. No assessment of correlation p57kip2 and clinical outcome was looked into. The study concluded that EZH2 depletion increased p57kip2 expression indicating that EZH2 acts as an EOC oncogene by targeting p57kip2 Table 4.
The above 4 studies demonstrated varying results. There are several factors that might explain the discrepancies seen in our meta-analysis. The first criterion might involve the technique used – particularly with respect to the antibody used. While Rosenberg and Sui et al. used polyclonal p57kip2 antibodies at dilutions of 1:25 and 1:500 respectively, Khouja et al. used a monoclonal p57 kip2 antibody at 1:500 dilutions. Whilst it is common knowledge that high specificity of monoclonal antibodies decreases cross-reactivity and background noise providing reproducible results and might explain the results by Khouja et al., it is unclear whether technique alone may explain the differences in the results of Rosenberg and Sui et al. Other variations in technique depending on the timing, type of autostaining technology employed (e.g. Dako/Ventana) may also account for the differences in results. This may also explain the differences in the p57kip2 in normal ovarian tissue which was observed by Sui & Guo et al. and none of the 10 cases in Khouja et al. series showed the expression. Only Sui & Guo et al. have assessed for expression in normal ovarian tissue and along the spectrum of ovarian tumours ranging from benign to borderline to malignant. While this gives an indication of p57kip2 expression across the tumour gradient, meaningful data analysis remains difficult from such small numbers.
Rosenberg et al. included only stage 2 & 3 EOC cases in their study. Since EOC tends to present as an advanced stage disease (>Stage 3) in more than 80% of cases excluding cases of stage 4 disease which represents a big cohort of the cases is unlikely to shed meaningful results especially when extensive drug development research is focused on advanced stage EOC in an attempt to improve survival. Due to more aggressive treatment (surgery & chemotherapy) survival in Stage 4 disease has definitely improved from the 13% 2 year survival mentioned in Rosenberg et al. to around 30-40% 5 year survival.
On the contrary, involving the early stages may shed light on the tumour development and biology which might improve our understanding of this silent killer disease and possibly assist in moving a step closer towards the ultimate dream of an ovarian cancer vaccine.
All 4 studies have clarified the grade of tumours as shown in Table 4. 3 out of the 4 studies focused on the adenocarcinomas and clearly describe the number of cases in each histological subtype though one wonders whether focusing on one subtype or histological sub-type might yield clarity of data especially in studies with small number of cases.
The above studies interestingly have shown great variation in IHC expression of p57kip2 in EOC. This has obviously then led to differing conclusions being drawn as to the usefulness of this cell cycle regulator as an IHC marker for EOC. Conflicting results from the above studies may be explained by differences in technique, type of antibodies, case selection and different methodologies as described above. However, while the above studies incorporated the different histological subtypes of EOC, work on individual histological subtypes might provide more data. Multicentre collaborative studies might help to increase the number of cases especially for rarer tumours (e.g. clear cell). Combined IHC coupled with serum and RNA work to assess the full range of p57kip2 expressed in a wide variety of ovarian cancers along with a complete clinical dataset might help clarify the discrepancy from the above studies.
p57kip2 as potential prognostic and therapeutic marker
Originally discovered as a CDKI, and considered as having only tumour suppressor function, it is becoming obvious that the role of p57kip2 has expanded beyond cell regulation. The above studies show that low p57kip2 function is associated with high grade, advanced clinical stage and overall poor prognosis . This association may become crucial in target oriented cancer therapy by modulating p57kip2 levels. Several clinical trials including some Phase I & II clinical trials in haematological malignancies and some solid tumours are currently looking at reactivation of p57kip2 in cancer cells as a therapeutic strategy. This is based on the principle that downregulation of p57kip2 in the majority of cancers makes it an obvious tumour suppressor candidate . As shown in various studies, suppression of the protein expression is due to the loss of p57kip2activity in cancer. Reactivation of p57kip2 expression can be a useful mechanism and some therapies (HDAC inhibitors and demethylating agents) that cause reactivation are already being used successfully.
The involvement of p57kip2 is certainly more than just inhibition of cell growth as was previously thought. It is now known that its function includes promotion of cell differentiation and apoptosis and metastasis inhibition. In trying to unravel the mystery of p57kip2, its role as a diagnostic marker in the differential diagnosis of early hydatidiform mole has emerged. The four studies discussed above have suggested a role for p57kip2 as a prognostic marker at various levels and needs further clarification. With aggressive surgery being advocated for EOC including secondary debulking surgery, serial IHC work in conjunction with cell line and serum work from women at varying stages of treatment for EOC (both chemo-naïve and post treatment) may help to elucidate its role better. It may also help clarify its role in the development of drug resistance in women following first line platinum chemotherapy, both to assess potential response to treatment and to further explore if upregulation helps prevent development of resistance. Larger number of cases from each individual histological subtype needs to be investigated for meaningful data to emerge which might require collaborations. As a protein that appears to regulate various functions in the “cancer hallmarks cycle” resolving the enigma of its function might provide interesting results in terms of prognosis and therapy for EOC.
TKM’s research is funded by GRACE, a gynaecological cancer charity based in the UK.
- Sankaranarayanan R, Ferlay J: Worldwide burden of gynaecological cancer: the size of the problem. Best Pract Res Clin Obstet Gynaecol 2006,20(2):207–225. 10.1016/j.bpobgyn.2005.10.007PubMedView Article
- Madhuri TK, Tailor A, Butler-manuel SA: Laparoscopic Interval Surgery for Stage 4 Primary Fallopian tube carcinoma. Gynecol Surg 2009,6(4):71–373.View Article
- Hanahan D, Weinberg RA: The hallmarks of Cancer. Cell 2000, 100: 57–70. 10.1016/S0092-8674(00)81683-9PubMedView Article
- Hanahan D, Weinberg RA: Hallmarks of cancer: the next generation. Cell 2011,144(5):646–674. 10.1016/j.cell.2011.02.013PubMedView Article
- Coley HM, Arnida MSN, Chivers P, Papacharalbous E, et al.: The Cyclin-Dependent Kinase Inhibitor p57Kip2 is epigenetically regulated in carboplatin-resistance and results in collateral sensitivity to the CDK inhibitor seliciclib in ovarian cancer. Br J Cancer 2012 Jan 31,106(3):482–489. Epub 2012 Jan 10 10.1038/bjc.2011.566PubMed CentralPubMedView Article
- Nam EJ, Kim YT: Alteration of cell-cycle regulation in epithelial ovarian cancer. Int J Gynecol Cancer 2008,18(6):1169–1182. 10.1111/j.1525-1438.2008.01191.xPubMedView Article
- Kavanagh E, Joseph B: The hallmarks of CDKN1C (p57, KIP2) in cancer. Biophys Acta 2011,1816(1):50–56.
- Guo H, Tian T, Nan K, Wang W: p57: A multifunctional protein in cancer. Int J Oncol 2010,36(6):1321–1329.PubMed
- Pateras IS, Apostolopoulou K, Niforou K, Kotsinas A, et al.: p57KIP2: “Kip”ing the cell under control, Mol. Cancer Res 2009, 7: 1902–1919. 10.1158/1541-7786.MCR-09-0317
- Besson A, Dowdy SF, Roberts JM: CDK inhibitors: cell cycle regulators and beyond. Dev Cell 2008, 14: 159–169. 10.1016/j.devcel.2008.01.013PubMedView Article
- Vlachos P, Nyman U, Hajji N, Joseph B: The cell cycle inhibitor p57(Kip2) promotes cell death via the mitochondrial apoptotic pathway. Cell Death Differ 2007, 14: 1497–1507. 10.1038/sj.cdd.4402158PubMedView Article
- Besson A, Assoian RK, Roberts JM: Regulation of the cytoskeleton: an oncogenic function for CDK inhibitors? Nat Rev Cancer 2004, 4: 948–955. 10.1038/nrc1501PubMedView Article
- Nan KJ, Guo H, Ruan ZP, Jing Z, et al.: Expression of p57(kip2) and its relationship with clinicopathology, PCNA and p53 in primary hepatocellular carcinoma. World J Gastroenterol 2005, 11: 1237–1240.PubMed CentralPubMedView Article
- Matsuura T, Takahashi K, Nakayama K, Kobayashi T, et al.: Increased expression of vascular endothelial growth factor in placentas of p57(Kip2) null embryos. FEBS Lett 2002, 532: 283–288. 10.1016/S0014-5793(02)03681-5PubMedView Article
- McCluggage WG: Recent advances in immunohistochemistry in gynaecological pathology. Histopathology 2002,40(4):309–326. 10.1046/j.1365-2559.2002.01384.xPubMedView Article
- Marjoniemi VM: Immunohistochemistry in gynaecological pathology: a review. Pathology 2004,36(2):109–119. 10.1080/00313020410001671948PubMedView Article
- Wilson GE, McVey RJ: An overview of recent advances in gynaecological pathology. CPD Bull Cell Pathol 2006,6(1):3–9.
- Silva E: Update in Gyn Pathology II. Pathologica 2005,97(1):34–36.
- Rosenberg E, Demopoulos RI, ZeleniuchJacquotte A, Yee H, et al.: Expression of cell cycle regulators p57KIP2, cyclin D1, and cyclin E in epithelial ovarian tumors and survival. Hum Pathol 2001,32(8):808–813. 10.1053/hupa.2001.26462PubMedView Article
- Sui L, Dong Y, Ohno M, Watanabe Y, et al.: Expression of p57kip2 and its clinical relevance in epithelial ovarian tumors. Anticancer Res 2002,22(6A):3191–3196.PubMed
- Khouja MH, Baekelandt M, Nesland JM, Holm R: The clinical importance of Ki-67, p16, p14, and p57 expression in patients with advanced ovarian carcinoma. Int J Gynecol Pathol 2007,26(4):418–425. 10.1097/pgp.0b013e31804216a0PubMedView Article
- Guo J, Cai J, Yu L, Tang H, et al.: EZH2 regulates expression of p57 and contributes to progression of ovarian cancer in vitro and in vivo. Cancer Sci 2011,102(3):530–539. 10.1111/j.1349-7006.2010.01836.xPubMedView Article
- B’Andrilli G, Kumar C, Scambia G, Giordano A: Cell cycle genes in ovarian cancer: Steps toward earlier diagnosis and novel therapies. Clin Can Res 2004,10(24):8132–8141. 10.1158/1078-0432.CCR-04-0886View Article
This article is published under license to BioMed Central Ltd. This is an Open Access article distributed under the terms of the Creative Commons Attribution License (http://creativecommons.org/licenses/by/2.0), which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.