- 2 of 7 chordomas had Loss of Heterozygosity (LOH) at intron 17 of the retinoblastoma gene [9]
- 25 of 27 chordomas (85%) had LOH at 1p36.13. Putative tumor supressor genes at this locus include CASP9, EPHA2, PAX7, DAN and DVL1. 23 of 27 chordomas shared a common LOH interval centered on D1S2697 (1p36.13) and delimited by D1S436 and D1S2826. [10]
- A minimally deleted region in 1p36 . 31 – p36 . 11 was found in 57 % of chordomas . This region contains the transcription factor RUNX3 . [11]
- A tumor-suppressor gene involved in familial and sporadic chordoma maps to 1p36: LOH data relating to 6 sporadic chordomas defines a smallest region of overlapping loss of about 25 cM from D1S2845 (1p36.31) to D1S2728 (1p36.13). [12]
- 12 of 16 skull base chordoma patients showed LOH at 1p36, six of whom displayed a wide region of LOH involving all the tested markers. The remaining six patients displayed a variable region of LOH that was different in LOH extent and/or localization. Within the last group, three patients showed segmental LOH. [13]
- The lack of 1p36 LOH or the presence of TNFRSF8 expression might be associated with a better prognosis in patients with SBCs. [13]
- Of a total of 26 chordomas investigated by aCGH and FISH, 15 (58%) displayed a heterozygous deletion of the region covering the CDKN2A locus, and 3 (12%) showed a homozygous deletion. Heterogeneous imbalances of large chromosomal regions were common . [11]
- The chromosomal region on 6q27 containing the T (brachyury) gene was gained in 6 of 21 chordomas (29%). None of the 21 chordomas analyzed showed deletions that could have affected this gene. [11]
- 3 of 39 (8%) chordomas had a minor (3:1) allelic gain of brachyury. [14]
- The chromosomal regions containing TP53 and TP53BP1 were lost in 48 % and 29 % of chordomas , respectively . [Ref]
- Fluorescent in situ hybridisation analysis failed to show amplification of FGFR4 in 50 chordomas. [14]
No translocations of ETS2 and ERG were found in 27 chordomas. [14]
Cytogenetic analysis of five chordomas revealed normal karyotypes in four patients and an abnormal karyotype in only one tumor cell among 100 cells studied from the fifth patient. [15]
The most frequently observed imbalances observed in six sacral chordomas by CGH were gains of chromosomal areas 1q23~q24 (3/6), 7p21~p22 (3/6), 7q (4/6), and 19p13 (3/6), as well as loss of chromosomal segment 9p22~p23 (3/6). IP-FISH confirmed the CGH findings and showed that chromosome 7 was polysomic in four of the tumors. Six of six sacral chordomas had trisomic and tetrasomic clones for chromosome 7, and two of them had pentasomic clones as well. [16].
27 of 64 (42%) skull base chordomas had clonal chromosome aberrations and 37 (58%) had a normal karyotype. [17]
A normal karyotype was found in three cases of chordoma of the skull base, whereas a recurrent vertebral chordoma showed 46,XY,t(6;11)(q12;q23). [18]
117 chromosomal aberrations (median, six per tumor) were found in 16 chordoma samples. On average, 3.2 losses and 4.2 gains were detected per tumor. [19]
No statistically significant difference in the number of chromosomal imbalances was detected by CGH between 10 sacrococcygeal chordomas and 5 sphenooccipital chordomas or between primary chordomas and recurrences. FISH experiments confirmed the CGH data. [19]
Losses of DNA sequences were most prevalent at 3p (8/16) and 1p (7/16). Losses of 3p were detected in five of seven primary chordomas. Loss of 13q was found in one of 10 sacrococcygeal tumors, in the single spinal tumor, and in 3 of 5 sphenoocciptial tumors. [19]
DNA sequence copy number gains occurred most frequently at 7q (11/16), 20 (8/16), 5q (6/16), and 12q (6/16). Gains of chromosome 17 and the majority of gains of 5q, 12q, and 20 were found in recurrences. [19]
- Clonal chromosome losses were identified in six of 11 recurrent tumors. In four of these, losses included the deletion or loss of chromosomes 3, 4, 13, and 18, followed by the loss of chromosome 10 in three of the four tumors. [20]
- The most common chromosome gains included all or part of chromosome 1q and chromosome 7. Three of 11 recurrent tumors shared chromosome aberrations of isochromosome 1q, -3,-4,+7,-10,del13/-13, -18. [20]
- 6 of 21 (27%) chordomas had EGFR amplification (4 or more copies) [21]
- 0 of 35 chordomas had amplification of EGFR [22]
COX2 was detected by immunohistochemical analysis in 19 (90%) of 21 chordomas. [28]
Androgen Receptor (AR) was detected in 19 of 21 chordomas, with the number of AR+ nuclei varying widely from case to case (mean ± SD, 72% ± 36% of nuclei; range, 5%-100 %). [28]
No specific nuclear staining was observed for ER-α or PR among 21 chordomas. [28]
ER-β was detected in 21 of 21 chordomas (mean ± SD, 72% ± 25% of nuclei; range, 20%-100%) [28]
Out of 35 skull-base chordomas, all were positive for brachyury expression [22]
- Brachyury protein was detected in the fetal notochord and in 53 of 53 chordomas but was not detected in the nucleus pulposus or over 300 neoplasms, including 163 chondroid tumours. [27]
- Brachyury protein was detected by IHC in 51 of 57 (89%) chordomas. [29]
All chordomas were positive for keratin but negative for GFAP [30].
Chordomas frequently expressed alpha 2,6-ST and alpha 2,6-linked sialoglycoconjugates [31].
Cytokeratins and an epithelial membrane-specific oligosaccharide sequence are found in chordomas but not in chondrosarcomas or normal cartilage [32].
The cartilage-typical large aggregating proteoglycan aggrecan was present throughout all chordomas and, thus, a very characteristic gene product and marker of these neoplasms [33] .
TrkA and NGF expression was significantly higher in chordoma cells compared to notochordal cells [34]
CK18 was detected in 34 of 34 benign notochordal cell tumors and 3 of 3 chordomas but 0 of 27 notochordal vestiges [35]
HMW-MAA was detected in 13 of 21 chordomas (62%) [26]
We conclude that c-MET expression as frequent as that observed in the notochordal tissue, chordomas, articular cartilage, and cartilaginous tumors is related to the development of both normal tissue and chondroid tumors [36] .
Most chordomas displayed strong expression of EGFR and c Met , whereas a variable level of expression of HER2 / neu was seen [37] .
Most chordomas had strong expression of both the hepatocyte growth factor/scatter factor receptor and EGFR [37].
All studied chordomas expressed both PR and ER - alpha [38] .
47 of 50 ( 94%) chordomas showed immunoreactivity for at least one of the fibroblast growth factor receptors. FGFR1, FGFR2, FGFR3, and FGFR4 were detected in 56%, 80%, 78%, and 70% of cases,respectively. In the majority of cases, more than 50% of the tumour cells showed immunoreactivity. [14]
Expression of the cell adhesion molecules (CAMs) E-cadherin, α-catenin, β-catenin, γ-catenin and neuralcell adhesion molecule was decreased in most of 15 chordoma specimens examined. Immunohistochemistryfor CAMs can be used as a diagnostic tool for differentiation between chordoma and chondrosarcoma [39]
Immunohistochemical analysis of 51 chordomas revealed extensive expression (>50% tissue stained) of E-cadherin, N-cadherin, b-Catenin, g-Catenin, a-Catenin was observed in 47, 36, 45, 37, and 15 percent of tumors, respectively. In the majority of chordomas, E-cadherin and N-cadherin expression was inversely correlated, whereas b-catenin and g-Catenin expression was directly correlated . [40]
Telomerase activity was observed in 1 of 2 chordoma tumors, but only at 1% of the activity of HeLa cells. [15]
Intraosseous benign notochordal cell tumors (BNCT) were found in 7.3% (6/82) sacrectomy/coccygectomy surgical specimens. In 4 cases the BNCT was immediately adjacent to the chordoma, whereas in 2 cases it was in a bone uninvolved by the chordoma. An immunohistochemical stain for keratin performed on 2 of the BNCTs showed that both lesions were diffusely and strongly positive for keratin. [41]
E-cadherin, N-cadherin, b-Catenin, g-Catenin, a-Catenin proteins were detected in 24, 82, 71, 60, and 53 percent of chordomas, respectively, by quantitative immunoblotting. [40]
E-cadherin expression was reduced by twelve-fold and N-cadherin was increased by three-fold in a recurrent chordoma compared with the primary tumor. [40]
12 of 16 chordomas stained positive for Galectin-3. [42]
- Immunodetection of S-100 protein and vimentin showed strong staining in neoplastic cells throughout 22 of 22 chordomas. [43]
- Cytokeratin 19 and EMA were positive in 15 of 15 classic chordomas. Cytokeratin 19 and EMA were positive in chordoid areas and in part of the chondrocyte-like cells in 7 of 7 chondroid chordomas. The chondrocyte-like cells in the centers of the chondroid areas were negative for these epithelial cell markers. [43]
- The chondroid areas of 7 chondroid chordomas expressed aggrecan and glycosaminoglycans. In situ hybridization showed high mRNA expression levels for aggrecan in neoplastic cells of classic and chondroid chordoma; thus, this aggrecan-rich matrix was is not produced by surrounding nonneoplastic stromal cells. [43]
- In 15 classic chordomas, the extracellular tumor matrix was mostly collagen-poor as shown by histochemical analysis. Type VI collagen could be demonstrated via IHC. Type I collagen and Type III collagen were found only focally. Type II collagen was absent in most tumor areas. Focal expression and deposition of type II collagen was seen in classic chordomas without histologically evident chondroid matrix formation. Histochemically, these areas had a relatively collagen-rich tumor matrix. [43]
- In 7 chondroid chordomas, high type II collagen mRNA and protein levels were found in the chondroid tumor compartments. Type VI collagen was mainly concentrated in the pericellular matrix compartment. Type I collagen and Type III collagen were focally present. [43]
- The mean MIB-1 labeling index in 22 chordomas was 0.5 (range, 0 to 3.8). [44]
- Mean cyclin D1 LI in 22 chordomas was 35.6 (range, 0 to 82.4). [44]
- No p53 immunostaining was observed in 15 of 22 tumors, 1+ staining in 2 of 22 tumors, 2+ staining in 3 of 22 tumors, and 3+ staining in 2 of 22 tumors. [44]
- No bcl-2 immunostaining was observed in 19 of 22 tumors, 1+ staining was noted in 2 of 22 tumors, and 3+ staining was noted in 1 of 22 tumors. [44]
- A fate-mapping experiment found that the majority of mouse notochord cells ended up within the nucleus pulposus, however a small number of cells were found to reside, and persist throughout life, in the vertebral column, primarily in the middle of each vertebrae along the ventral surface. These notochordal remnants were found along the entire length of the vertebral column in all twelve animals examined. [3]
- 6 of 6 chordomas showed moderate to strong staining for galectin-3. Subcellularly, galectin-3 was localized mostly in the cytoplasm, while a subset of tumor cells also showed nuclear distribution. Differences in staining patterns of chordoma cells could not be correlated to any histological features of these tumors. [45]
- Cytokeratin and S-100 were expressed in five conventional chordomas and among a few cells with physaliphorous appearance in two chondroid chordoma cases. [46]
- No Ep-CAM expression was noted in chordomas, but E-cadherin was detected in most chordoma cells irrespective of histological subtypes. E-cadherin expression can be used to distinguish between chondroid chordomas and chordosarcomas. [46]
- The MIB-1 proliferative index was significantly higher in grade 2 or 3 lesions than in grade 1 lesions (P = .014), and the MIB-1 index tended to be higher in solid type than in trabecular type tumors (P = .088). [47]
- Higher nuclear atypia (grade 2 or 3) occurred in 5 of 9 solid type chordoma specimens and 0 of 8 trabecular type specimens. [47]
- p53 overexpression was detected in two specimens of solid type, and the MIB-1 LI in these two specimens was significantly higher (P = .037) than that in the specimens without p53 overexpression. [47]
- In 15 chordomas, positive immunoreactivity for E-cadherin, alpha-catenin, beta-catenin, gamma-catenin, and NCAM was seen in 11, 7, 13, 10, and 14 specimens, respectively. Negative immunoreactivity was seen in the remaining chordomas. [39]
- [39]
Immunohistochemical analysis of chordoma tumors revealed p53 overexpression in 30.4% of SBCs and 20.8% of NSBCs; MDM2 overexpression in 8.7% of SBCs and in 16.7% NSBCs; and cyclin D1 overexpression in 26.1% of SBCs and in 50.0% of NSBCs. Possible pRb overexpression was detected in 37% of SBCs and in 62.5% of NSBCs. Few chordomas expressed p27Kip1 or p16INK4a. [48]
Alterations in cyclin D1 and pRb were seen more often in recurrent SBCs than in primary SBCs. [48]
A mdm2 amplification was detected in 8 of 52 samples(15.4%) tumors. MDM2 overexpression was not associated with mdm2 amplification. [48]
Alterations of p53, MDM2, cyclin D1, and pRb proteins were found to have cooperative effects on both higher proliferative ability (MIB-1 labeling index) and increased nuclear pleomorphism, a previously described prognostic indicator for patients with chordoma. In recurrent lesions, only p53 overexpression and MDM2 overexpression were associated with higher MIB-1 LI. No correlation was detected between cell cycle althigher MIB-1 LI. No correlation was detected between cell cycle alterations and nuclear pleomorphism in patients with recurrent lesions. [48]
p53 overexpression was correlated significantly with the overexpression of MDM2 and cyclin D1. [48]
For forty patients who had primary lesions, the 5-year survival rate was 38.9% for patients with p53 overexpression and 79.4% for patients without p53 overexpression. [48]
Compared with primary lesions, MIB-1 LI was significantly increased in recurrent SBCs (mean, 7.2), but not in recurrent NSBCs, leading to differences in MIB-1 LI between recurrent SBCs and NSBCs (P = 0.006). [7]
The mean percentage of p75 expression was very similar in 10 chordoma samples and notochordal cells (P=0.394). [34]
The mean percentages of TrkA and NGF expression were significantly higher in chordoma cells than in notochordal cells (P=0.002). [34]
The mean apoptosis index of chordoma cells was significantly lower compared with that of notochordal cells (P=0.03). [34]
The mean proliferation potential index of chordoma cells was significantly higher than in notochordal cells (P=0.01). [34]
- In NSBCs, intralesional fibrous septum and lobular growth patterns were associated with nuclear pleomorphism, mitosis, and the MIB-1 labeling index. [5]
- Among 23 chordomas hepatocyte growth factor (HGF) was expressed in less than 10% of tumor cells in 5 tumors, and not at all in the remaining 18 tumors. Among 34 chordomas five did not express c-MET, 9 expressed c-MET in 10-50% of cells and 20 expressed c-MET in greater than 50% of cells. [49]
- There was a significant correlation between the expression of c-MET and CAM5.2. [49]
- Lesions with higher c-MET expression had higher expression of of proteinases, including MMP-1, MMP-2, uPA and cathepsin B, than those with lower c-MET expression, and the differences were statistically significant for MMP-1 (P = 0.013) and MMP-2 (P = 0.022). [49]
- There were no correlations between HGF or c-MET expression and patient age, gender, nuclear pleomorphism, mitosis, apoptosis, necrosis or bleeding. [49]
- The HGF score was 1 (faint) in only a single primary skull base chordoma and was scored as 0 (negative) in the remaining 45 primary and 25 recurrent lesions. [50]
- 32 of 46 primary chordomas and 22 of 25 recurrent lesions exhibited c-MET expression. [50]
- There were no correlations noted between c-MET expression and sex, subsequent disease recurrence(s), nuclear pleomorphism, mitosis, apoptosis, necrosis, bleeding, proliferative ability, MIB-1 labeling index, pancytokeratin, vimentin, or S-100 protein expression. [50]
- There was a significant correlation noted between the expression of c-MET and CAM5.2 in both primary and recurrent lesions. [50]
- Recurrent lesions with higher c-MET expression presented with higher average scores of MMP-1, MMP-2, TIMP-1, uPA, and CatB, and the differences were found to be statistically significant or nearly significant for MMP-1, MMP-2, TIMP-1, and uPA. In primary lesions, only uPA was found to be correlated with c-MET expression. [50]
- Positivity for cytokeratin and Brachyury had a sensitivity of 98% and specificity of 100% for detecting chordoma. [29]
- Positivity for both epithelial membrane antigen and AE1/AE3 had a sensitivity of 90% and a specificity of 100% for detecting chordoma. [29]
- SOX-9 is common to both notochordal and cartilaginous differentiation, and is not useful in the chordoma-chondrosarcoma differential diagnosis. [29]
- Significant correlation was observed between RECK and MMP-9 expression (p=0.036). No correlation was observed between MMP-2 and MMP-9 (p=0.78), or between RECK and MMP-2 (p=0.193). [51]
- Chordomas were not found to express high levels of the hypertrophic gene collagen X, platelet-derived growth factor alpha, or reticulocalbin 3. [27]
- Immunohistochemical staining with HMW-MAA-specific mAb resulted in the staining of 62% of 21 chordomas tested. [26]
- Most of 34 notochordal lesions found in autopsy cases and 3 classic chordomas demonstrated positive immunostaining for vimentin, S-100 protein, epithelial membrane antigen, CAM 5.2, AE1/AE3, and CK18. 27 notochordal vestiges in the fetal intervertebral discs failed to demonstrate any positive reaction for CK18, although the other antigens were positive. [35]
- EGFR was detected by immunohistochemistry in 21 of 26 (81%) chordomas. [21]
- 10 of 21 chordoma cases showed strong expression of EGFR. 6 of these 10 presented with metastatic disease. 4 of 11 chordoma with low EGFR expression presented with metastatic disease. [21]
- 26 of 26 lesions demonstrated positive immunostaining with vimentin, S100 protein, EMA, AE1⁄AE3 and CAM5.2. [52]
Western blot analysis showed phosphorylation of FRS2a and ERK1/2 in 6 out of 6 chordomas that were also reactive by immunohistochemistry for at least one of the FGFRs : FGFR1, FGFR2, FGFR3, and FGFR4. [14]
Among 5 chordomas, most demonstrated activation of the Akt/mTOR cascade including expression of p-Akt (4/5), mTOR (5/5), p-mTOR (5/5), p-S6K(5/5) and 4E-BP1 (4/5). [53]
Cells in a tissue microarray of 70 chordomas samples showed nuclear staining for phosphorylated signal transducers and activators of transcription (Stat3). The Stat3 pathway is constitutively activated in chordomas. [54]
MTT assay showed that the growth of 3 out of 3 chordoma cell lines (UCH1, CH 8 and GB 60) was inhibited by SD-1029, an inhibitor of Stat3 activation. Treatment with SD-1029 inhibited expression of Stat3 signaling cascade (Western blot), phosphorylation of Stat3 in chordoma cells in vitro, proliferation in three-dimensional culture (immunofluorescence), and antiapoptotic proteins Bcl-xL and MCL-1. [54]
IHC analysis of 21 cases of chordoma for expression of proteins involved in signal transduction from RTKs indicated platelet-derived growth factor receptor-b (PDGFR-b), epidermal growth factor receptor (EGFR), KIT and HER2 were detected in 100%, 67%, 33% and 0% of cases, respectively. [55]
Phosphorylated isoforms of p44/42 mitogen-activated protein kinase, Akt and STAT3, indicative of tyrosine kinase activity, were detected in 86%, 76% and 67% of cases, respectively. [55]
Correlation between EGFR expression and p-EGFR was not statistically significant. Positive staining for p-EGFR correlated with p-STAT3 (McNemar’s test; P = 0.0235), but not with p-MAPK or p-Akt. [55]
mTORC1 signaling in chordoma-derived cell lines is deregulated in response to growth factor deprivation, but remains sensitive to amino acid availability. mTORC1 signaling is hyperactivated in sporadic sacral chordomas. [2]
Serum-starved U-CH1 cells exhibited high levels of pAkt, pTSC2, and pPRAS40. Akt signaling in U-CH1 cells is also constitutively activated in a growth factor-independent manner. ERK phosphorylation was also high in serum-starved U-CH1 cells. [2]
PTEN expression was not observed in U-CH1 cells and was significantly reduced in Ch1 cells. These results suggest that constitutively high Akt activity, due to PTEN loss, may be responsible for hyperactivation of mTORC1 signaling through inactivation of TSC2 and PRAS40 in U-CH1 cells. Partial or complete deficiency of PTEN could be responsible for hyperactivation of mTORC1 signaling in at least a subset of chordomas. [2]
Phosphorylated PDGFRA, PDGFRB and KIT were detected 12/12, 18/18 and 12/14 chordomas, respectively. PDGFRB was detected in all of these samples at levels higher than synovial sarcoma specimen used as a positive control, while PDGFRA and KIT were less highly expressed than the GIST sample used as a positive control. [56]
No gain-of-function mutations were found in PDGFRA, PDGFRB or KIT. [56]
PDGFA, PDGFB, and SCF mRNA were detected in 31 of 31 chordomas, suggesting that activation of their corresponding receptors is via the the autocrine/paracrine loop. [56]
- No correlations between sex steroid hormone receptor expression and disease-free survival were observed [28]
- E-cadherin expression correlated with disease-free survival, tumor recurrence, and low survival rate; MIB-1 labeling index correlated with tumor recurrence and low survival rate; CD44 expression did not correlate with recurrence or survival rate [57]
- The 5-year survival rate among 40 patients was 38.9% for patients with p53 overexpression and 79.4% for patients without p53 overexpression. Anatomic site, age, gender, aberrations in other proteins (MDM2, cyclin D1, pRb, and p27Kip1) and mdm2 amplification did not correlate to clinical outcome [48]
- Proliferative ability of chordoma was found to be related to increased age, clinical status and nuclear pleomorphism in patients with both skull based chordomas (SBC) and nonskull based chordomas (NSBC). The prognosis for NSBC patients with nuclear pleomorphism was found to be significantly poorer than for those without . [7]
Extensive expression (>50% tissue stained) of N-cadherin correlates with a diminished recurrence-free survival rate, and a 3.28 fold increase in probability of recurrence. Minimal or sparse expression (<50% tissue stained) of E-cadherin correlated with increased probabilities of death, and a 10.98 fold increase in probability of death. These results suggest that changes in the relative expression of the cadherin–catenin complex reflect chordoma aggressiveness; and that decreased expression of E-cadherin and increased expression of N-cadherin may underlie the transition from a less to a more aggressive tumor phenotype. [40]
High (>75% of tumor nuclei ) pStat3 immunohistochemical staining was associated with significantly worse prognosis than low pStat3 (<10% of tumor nuclei ) staining . [54]
Recurrent tumors expressed greater level of pStat3 than primary tumors, and pStat3 expression intensity was higher in tumors with metastasis compared with the primary tumors without metastasis. [54]
The staining patterns of PDGFR-B, EGFR, p-EGFR, KIT, HER2, p-MAPK, p-Akt, and p-stat3 did not correlate with disease-free survival or site of origin . [55]
The odds ratio for recurrence in lesions with an abnormal versus a normal karyotype was 12. [17]
Aberrations in chromosomes 3, 4, 12, 13, and 14 were associated with frequent recurrence and decreased survival time. [17]
In de novo cases, 31 of 42 lesions (74%) had a normal karyotype, while among recurrent cases only 6 of 22 (27%) had a normal karyotype. There were significantly more cases of chromosomal aberrations in recurrent chordomas than in de novo lesions.[17]
7 of 37 chordomas with a normal karyotype (1 de novo and 6 recurrent lesions) and 20 of 27 lesions with an abnormal karyotype (5 de novo and 15 recurrent cases) had progression or recurrence noted at follow-up. The 5-year recurrence-free survival rate was 81% in 37 patients with chordomas of normal karyotypes, and 34% in 27 patients with chordomas of an abnormal karyotype. [17]
95% of cases with progression involved chromosome 3 and/or 13 abberations. The median survival time was 4 months when both of these chromosomes had aberrations. [17]
In patients with NSBC, prognosis was found to be significantly poorer in cases with nuclear pleomorphism than in those without (P = 0.027). [7]
The 5-year survival rate was 79.9% in patients with c-MET expression, and was 44.4% in those without c-MET expression; there was a significant difference noted with regard to their survival rate. [50]
- Higher MMP-9 expression was correlated with poorer outcome. [51]
- Cases consisted of 6 clival chordomas, 1 lumbosacral chordoma, and 1 chordoma arising from the sphenoid bone. In 8 pediatric chordomas, there was a correlation between higher MIB-1 LI and tumor recurrence (p = 0.007) and MIB-1 LI and low survival rate (p = 0.007). [57]
- The expression of E-cadherin correlated with disease-free survival (P = 0.009), tumor recurrence (P > 0.0007), and low survival rate (P > 0.0007). [57]
- There was no correlation between percentage of expression of CD44 and recurrence (p = 0.056) or survival rate (p = 0.056). [57]
The world's first wiki where authorship really matters (Nature Genetics, 2008). Due credit and reputation for authors. Imagine a global collaborative knowledge base for original thoughts. Search thousands of articles and collaborate with scientists around the globe.
