DNA Methylation Regulates p27Kip1 Expression in Rodent Pituitary Cell Lines

We previously reported loss of expression of p27^Kip1 (p27) protein in rat GH₃ and mouse GHRH-CL1 pituitary tumor cells compared with normal pituitary (NP). The molecular basis for the loss of expression of p27 protein in GH₃ and GHRH-CL1 cells is unknown. To determine the role of p27 gene methylation in the regulation of the expression of this cell cycle protein, the methylation patterns of p27 in normal and neoplastic pituitary cells was analyzed. Inhibition of DNA methyltransferase (DNA-MTase) with 5-aza-2′-deoxycytidine (AZAdC) induced expression of both p27 protein and mRNA when GH₃ and GHRH-CL1 cells were treated for 7 days in vitro. DNA methylation correlated inversely with the expression of p27 gene products in NP and pituitary tumor cell lines. Bisulfite genomic sequencing analysis showed that the normally unmethylated cytosines in exon 1 in NP and AtT20 cells were extensively methylated in GH₃ and GHRH-CL1 cells. After treatment of GH₃ and GHRH-CL1 cells with 10 μmol/L AZAdC, there were decreased numbers of methylated cytosines (by 60% to 90%) with variable methylation patterns observed by bisulfite genomic sequencing. Analysis of genomic DNA with methylation-sensitive enzymes showed that all SmaI, HhaI, and AvaI enzyme sites of the p27 gene in exon 1 were methylated in GH₃ cells but not in NP, confirming the bisulfite genomic sequencing results. AtT20 cells and a human pituitary null cell adenoma cell line (HP75), which expressed abundant p27, had a methylation pattern similar to the NP. DNA-MTase activity was elevated fourfold in GH₃ cells and twofold in GHRH-CL1 cells compared with DNA-MTase activity in NP and AtT20 cells. These results suggest that increased DNA methylation is another mechanism of silencing of the p27 gene in some pituitary tumors and possibly in other types of neoplasms.     

TGF-B and Estrogen Regulate P27(Kip1) and Cyclin D(1) in Normal and Neoplastic Rat Pituitary Cells

Pituitary hyperplasia and tumor growth are regulated by various hormones and growth factors. Estrogen (E₂) stimulates pituitary cell proliferation and prolactin (PRL) production. Estrogen also regulates transforming growth factor-β (TGF-β) effects in the pituitary. TGF-β in turn regulates various cell cycle proteins including p15 and p27^Kip1 (p27). To better understand the regulatory role of growth factors and hormones in the cell cycle we analyzed cyclin D₁, cyclin E, and p27 expression in normal and neoplastic rat pituitary cells. An in vitro analysis using cultured normal pituitary cells and GH₃ tumor cells and an in vivo analysis of estrogen-treated normal pituitary and implanted GH₃ cells were performed. Semiquantitative RT-PCR was used to analyze mRNA expression for cyclin D₁, cyclin E, and p27 in cultured pituitary cells and E₂-treated pituitaries in vivo. Cyclin D₁ and p27 were localized in the nuclei of normal pituitary cells by immunocytochemistry (ICC). Very weak or absent immunostaining for cyclin D₁ and p27 was present in GH₃ cells. Both normal pituitary and GH₃ cells had strong nuclear staining for cyclin E. Normal pituitary had a 20-fold greater amount of cyclin D₁ mRNA and a 3-fold greater amount of p27 mRNA compared to GH₃ cells, whereas GH₃ cells had slightly (1.5-fold) more cyclin E than normal pituitary cells. E₂ treatment in vivo stimulated cell proliferation and decreased cyclin D₁ mRNA levels in normal pituitary. GH₃ tumor cells, implanted subcutaneously in the same animal, showed increased proliferation after E₂ treatment, but there was no change in cyclin D₁ mRNA in GH₃ cells. Cyclin E and p27 mRNA levels did not change significantly in normal pituitary or in GH₃ cells after E₂ treatment in vivo. Treatment of normal pituitary cells with 10⁻⁹ M TGF-β1 for 3 d in vitro led to significant decreases in cyclin D₁ and p27 mRNAs (p < 0.05), whereas cyclin E levels were unchanged. These results indicate that cyclin D₁ and p27 mRNAs are present at significantly higher levels in normal pituitary compared to GH₃ cells, and that both E₂ and TGF-β1 can downregulate cyclin D₁ mRNA levels in normal pituitary cells, suggesting that these factors regulate G1 to S phase transition in pituitary cells. The lower levels of specific cell cycle regulators in GH₃ cells may explain the decreased regulatory control by E₂ in GH₃ tumor cells.     

Consequences of p16 tumor suppressor gene inactivation in mycosis fungoides and Sézary syndrome and role of the bmi-1 and ras oncogenes in disease progression

In examining the expression of oncogenes and tumor suppressor genes in mycosis fungoides and Sézary syndrome, we found the cell cycle-regulating protein p16 to be absent in T cells. Immunohistochemical staining with p16-specific antibodies showed that the number of p16-expressing cells in cutaneous lesions decreases in late stages. The repression of p16 was not attributable to deletion or methylation of this gene; however, the Bmi-1 oncogene, a known suppressor of p16, was present in mycosis fungoides and Sézary syndrome cell lines and skin lesions. The absence of p16 correlated with the phosphorylation of the retinoblastoma protein on cyclin D/CDK4- or cyclin D/CDK6-specific sites. Ki-ras, which stimulates phosphorylation of retinoblastoma via cyclin-dependent kinases, was found in all tested cutaneous T-cell lymphoma samples; and its expression generally was stronger in advanced stages. Thus, cutaneous T-cell lymphoma cells show changes in oncogene and tumor suppressor gene expression that increase proliferation.     

Downregulation of E-cadherin and its undercoat proteins in pituitary growth hormone cell adenomas with prominent fibrous bodies

The cadherin-catenin complex regulates cellular adhesion and motility, and genetic alterations in these molecules play a critical role in multistage tumorigenesis. In this study, the expression of three major type I classic cadherins—E-, N-, and P-cadherin—and their undercoat proteins—α-, β-, and γ-catenin, and pp120–was investigated in 127 pituitary adenomas and 10 normal adenohypophyseal glands using an immunohistochemical technique with highly specific monoclonal antibodies. In normal pituitary glands, E-cadherin, catenins, and pp 120 were strongly expressed on almost all hormone-producing cell-cell boundaries, N-cadherin was weakly immunoreactive on a few cell-cell boundaries, and P-cadherin was negative. In pituitary adenomas, a correlation was not identified among expression of E-cadherin, catenins, or pp120 with patient age, sex, hormone level, tumor size, and/or invasiveness, respectively. Expression of E-cadherin, catenins, and pp 120 was significantly reduced in 24 growth hormone (GH) cell adenomas with prominent fibrous bodies compared with the other subtypes of pituitary adenomas and normal pituitary glands (p<0.0001, respectively). Methylation-specific polymerase chain reaction analysis revealed that the E-cadherin gene promoter region was methylated in 6 of 16 (37.5%) GH cell adenomas with prominent fibrous bodies examined, 2 of which displayed total methylation, but not in 10 GH cell adenomas without fibrous bodies. No mutation of exon 3 of the β-catenin gene was found in 16 GH cell adenomas with prominent fibrous bodies or in 10 other subtypes of pituitary adenomas that showed unremarkable intracellular presence of β-catenin protein. In conclusion, the decreased expression of the E-cadherin-catenin complex and methylation of the E-cadherin gene promoter region only in GH cell adenomas with prominent fibrous bodies may be an event associated with the formation of fibrous bodies.     

p16 and p27 are functionally correlated during the progress of hepatocarcinogenesis

The molecular mechanism of the cell-cycle machinery in hepatocellular carcinoma (HCC) has not yet been fully elucidated. Among the various types of cell-cycle regulators, p16 and p27 are now considered to be potent tumor suppressors. p16 is a G1-specific cell-cycle inhibitor that prevents the association of cyclin-dependent kinase (CDK) 4 and CDK6 with cyclin D₁. Many studies have reported that p16 is inactivated not only in aggressive types of HCC but also in preneoplastic liver cirrhosis. In many cases of HCC, p16 is mainly inactivated by extensive CpG methylation, suggesting that epigenetic changes in the p16 gene may be important events during hepatocarcinogenesis. p27, an inhibitor of CDK2, is presently regarded as a potent adverse prognostic factor in many aggressive cancers. It should be noted that some cases of HCC show increased cell proliferation despite the expression of considerable amounts of p27. In these cases, p27 is inactivated by sequestration into cyclin D₁–CDK4-containing complexes. Although the reason for the compositional changes in the p27-containing complexes is unclear, our experimental results indicate that loss of p16 following DNA methylation is closely related to the functional inactivation of p27 in HCC. We suggest that assessment of the p16 status may be useful for a precise prognostic prediction for individuals with HCCs expressing high levels of p27.     

Oncogene Activation in Pituitary Tumors

Pituitary tumors constitute 10% of intracranial neoplasms and are mostly benign, monoclonal adenomas derived from single mutant cells. Pituitary oncogenes have been intensively studied and three of them, gsp, ccnd1, and PTTG are abundant in significant numbers of cases. gsp is present in approximately 40% of Caucasian patients with GH-secreting tumors and results from a mutated, constitutively active alpha subunit of Gs protein. Persistent activation of the cAMP-PKA-CREB pathway may lead to uncontrolled cell proliferation and GH secretion. ccnd1 is overexpressed cyclin D1, and cyclin D1 gene is amplified in some pituitary tumors. PTTG is expressed in most pituitary tumors. PTTG is localized to both the nucleus and cytoplasm and interacts with several protein partners. At least three tumorigenesis mechanisms are proposed for human PTTG. 1) PTTG and FGF form a positive feedback loop and stimulate tumor vascularity. 2) PTTG transactivates c-myc or other pro-proliferation genes. 3) PTTG overexpression causes aneuploidy. PTTG expression activates p53 and causes p53-dependent and -independent apoptosis. Due to lack of functional human pituitary cell cultures and appropriate animal models for pituitary tumors, many of the results reviewed here are obtained from heterologous systems.     

Frequent loss of the CDKN2C (p18INK4c) gene product in pituitary adenomas

Genomic alterations of cyclin-dependent kinase inhibitors have been demonstrated in a variety of tumor types including brain tumors. Among them, the cyclin-dependent kinase inhibitor 2A (CDKN2A or p16(INK4a)) gene has been shown to be frequently deleted or inactivated in astrocytic tumors. The CDKN2C (p18(INK4c)) gene is functionally related to CDKN2A. Moreover, mice with targeted disruption of CDKN2C alone or combined CDKN2C and cyclin-dependent kinase inhibitor 1B (CDKN1B or p27(Kip1)), or CDKN2C and TP53 gene disruption develop pituitary adenomas (PA) at high frequencies. The purpose of our study was to investigate genetic alterations of the CDKN2C gene by analysis of loss of heterozygosity (LOH), screening for mutations, analysis of promoter methylation, and protein expression in 38 PAs. In addition, genomic alterations and protein expression of the cell cycle genes CDKN2A and its alternatively spliced form, p14(ARF), as well as the retinoblastoma RB1 gene were investigated. LOH at the CDKN2C gene locus was detected in 25% of pituitary adenomas, whereas the RB1 and CDKN2A loci were altered in only 10%. No mutations were detected within the coding regions of the CDKN2C gene. However, 39.5% of adenomas displayed CDKN2C promoter methylation. The absence of CDKN2C protein was correlated with LOH of the CDKN2C locus on chromosome 1 and with methylation of the CDKN2C promoter. This is the first report to describe that the tumor suppressor gene CDKN2C is frequently targeted by genomic alterations in pituitary adenoma. The most common genetic alteration was promoter methylation suggesting that inactivation of CDKN2C by this mechanism may play an important role in pituitary adenoma development. Additional Supporting Information may be found in the online version of this article.     

Expression of cyclin Ds in relation to p53 status in human breast carcinomas

 Cyclin D1 has been reported to be overexpressed in many tumours, including breast carcinomas. Cyclin D1 was first identified as a protooncogene (BCL1/PRAD1), and its overexpression was related to tumour proliferation. The product has also recently been identified as important in mediating cell cycle growth arrest via the p53 pathway in murin fibroblast cell lines. Ninety breast carcinomas previously analysed for p53 status were analysed for amplification of cyclin D1, D2 and D3 genes by Southern blot analysis and for protein expression by immunhistochemistry. In 10 samples gene amplification was detected at the cyclin D1 locus. No gene amplification was detected at the cyclin D2 and D3 loci. Immunoreactivity for cyclin D1 was detected in 38 (42.2%) tumour tissue samples. Fifty samples were immunostained for cyclin D2 and D3. Only 2 samples (4%) showed immunoreactivty for cyclin D2, and 9 samples (18%) for cyclin D3. Cyclin D1 protein overexpression was significantly more often found in tumours with wild type p53 and in tumours with higher grades of differentiation expressing ER. No association was seen between gene amplification of the cyclin D1 gene and p53 status. We conclude there is a relationship between wild type p53 and cyclin D1 protein overexpression in clinical material, indicating that cyclin D1 may be another downstream effector of p53. Received: 25 November 1997 / Accepted: 30 March 1998     

Inactivation of the p16 gene in human pituitary nonfunctioning tumors by hypermethylation is more common in null cell adenomas

Recent studies have shown that methylation of the CpG island within the p16/CDKN2A/MTS1 (p16) gene is associated with loss of expression of p16 protein in pituitary tumors. We analyzed a series of 21 pituitary adenomas and three normal pituitaries along with a human pituitary cell line (HP75) for methylation of exon 1 by methylation-specific PCR, immunohistochemistry, and Western blotting. PCR analysis showed that 5/7 (71%) of null cell adenomas, but only 2/7 (29%) gonadotroph tumors were hypermethylated. In addition, 1 of 2 ACTH tumors but no GH (n=4) or PRL (n=1) adenoma examined were hypermethylated. Immunostaining and Western blot analysis of protein expression supported the methylation-specific PCR analyses. These results show that p16 gene silencing by hypermethylation is more common in null cell adenomas compared to other nonfunctioning adenomas such as gonadotroph tumors and that the role of p16 in the pathogenesis of pituitary adenomas is restricted to specific tumor subtypes. Supported in part by NIH CA90249 and by a grant from the Jarislowsky Foundation.     

Solid pseudopapillary neoplasms of the pancreas show an interruption of the Wnt-signaling pathway and express gene products of 11q.

Solid pseudopapillary neoplasms of the pancreas almost consistently show a beta-catenin mutation activating the Wnt-signaling pathway, resulting in overexpression of cyclin D1, but not in overt malignancy of this tumor. Besides cyclin D1, a set of markers (ie FLI-1, CD56 and progesterone receptor), whose genes map to chromosome 11q, are frequently expressed in solid pseudopapillary neoplasms. Chromosome 11q is a region that is also often affected in pancreatic neuroendocrine tumors. This immunohistochemical study was undertaken to gain insights into the downstream regulation of the Wnt-signaling pathway and the significance of overexpressed gene products belonging to chromosome 11q for the tumorigenesis in solid pseudopapillary neoplasms. Fourteen solid pseudopapillary neoplasms were analyzed for the expression of cyclin-dependent kinase inhibitors p21, p27, p16 and hyperphosphorylated retinoblastoma (pRb) proteins. In an extended series of 93 solid pseudopapillary neoplasms, beta-catenin, cyclin D1, FLI-1 and CD56 expression was examined and compared with that in 22 pancreatic neuroendocrine tumors. Solid pseudopapillary neoplasms (98%) showed aberrant expression of beta-catenin with a concomitant cyclin D1 expression in 69% of the cases, but no expression of pRb (0%) was found. p27 and p21 were expressed in 100% (14/14) and 86% (12/14) of the cases, but only 2/14 (14%) were positive for p16. FLI-1 was expressed in 63% of solid pseudopapillary neoplasms, but only in 1/22 pancreatic neuroendocrine tumors (5%), cyclin D1 expression was present in 14% of the latter. We conclude that in solid pseudopapillary neoplasms the activated Wnt-signaling pathway is disrupted, and that p21 and p27 are contributing to this fact by blocking of the hyperphosphorylation of the Rb protein, thus causing the very low proliferation rate characterizing the solid pseudopapillary neoplasms. The accumulation of high expression of proteins whose genes are located on chromosome 11q is characteristic of solid pseudopapillary neoplasms, but not of pancreatic neuroendocrine tumors.     

Survey of neuropeptide gene expression in tumor cell lines.

The presence of 3 different neuropeptide mRNAs with a strict cell-specific expression in vivo was investigated in 13 tumor cell lines from neuroendocrine and in 23 tumor cell lines from non-neuroendocrine origin. Northern blots showed no expression of mRNA for vasopressin (VP) in the 36 tested cell lines. Very low oxytocin (OT) mRNA hybridization signals were detected in the rat pituitary tumor cell line GH4C2 and the rat pancreas tumor cell line RIN5. Both the rat pituitary tumor cell line AtT-20 and the human myeloid leukemia cell line K562, contained proopiomelanocortin (POMC) mRNA. The low incidence of VP, OT and POMC gene expression in the tested tumor cell lines was not influenced by treatments inducing differentiation. In contrast, the cholecystokinin (CCK) gene which is widely present in nervous and endocrine systems was abundantly expressed in the human primitive neuroepithelioma cell line SK-N-MC and its clonal derivative SK-N-MC-IX-C. The results indicate that the expression of neuropeptide genes is very rare in tumor cell lines. The lack of expression in undifferentiated cells agrees with the appearance of expression after day 13 of the embryogenesis when maturation of neurons begins.     

Increased cyclin D3 expression significantly correlates with p27 nuclear positivity in gastrointestinal stromal tumors

Gastrointestinal stromal tumors, the most common mesenchymal tumors of the gastrointestinal tract, are characterized by strong expression of c-Kit protein. Recently, it has been shown that gastrointestinal stromal tumors may also contain alterations of genes involved in the regulation of cell cycle. In this study, we evaluate the prevalence and clinical significance of cyclin D1 and D3, Ki-67, p27, and retinoblastoma protein expression in a group of 50 human gastrointestinal stromal tumors selected from the files of the Moffitt Cancer Center. Tissue sections from each case were subjected to immunostaining using the avidin-biotin complex method. Cyclin D1 nuclear positivity was detected in 21 of 50 (42%) and cyclin D3 in 24 of 50 (48%) cases. p27 high immunoreactivity and negative or decreased retinoblastoma protein expression were identified in 33 of 50 (66%) gastrointestinal stromal tumors. In 19 of 50 (38%) tumors, Ki-67 had high labeling index. Direct correlation was observed between cyclin D3 and p27 expression (P < .0001), and between cyclin D1 and retinoblastoma protein (P = .03). Coexpression of cyclin D3 and p27 was demonstrated by immunofluorescence. The p27 protein expression inversely correlated with tumor size (P = .004), but was not correlated with tumor grade (P = .12). Ki-67 directly correlated with both tumor size (P = .03) and tumor grade (P = .008). We report a direct correlation between cyclin D3 and p27 expression in gastrointestinal stromal tumors. Additional alterations in cyclin D1, Ki-67, and retinoblastoma protein expression indicate a disregulated cell cycle in these tumors.     

Critical roles for immunoglobulin translocations and cyclin D dysregulation in multiple myeloma

Summary: Multiple myeloma (MM) is a tumor of long‐lived bone marrow plasma cells (PCs). Nearly 40% of MM tumors have immunoglobulin H (IgH) translocations involving four recurrent chromosomal loci (oncogenes): 11q13 (cyclin D1), 6p21 (cyclin D3), 4p16 (MMSET and FGFR3), and 16q23 (c‐maf). Other MM tumors have Ig translocations involving different loci, none of which is involved in more than 1% of tumors. At least 25% of MM tumors have no Ig translocation. Unlike normal PCs, MM tumors usually express one of the three cyclin D genes at a high level. Translocations involving 4p16 and 16q23 do not directly target a cyclin D gene, but they are associated with a high level of cyclin D2 expression. Although cyclin D1 is not expressed in normal hematopoietic cells, one‐third of MM tumors ectopically express cyclin D1 in the absence of t(11;14). Despite a low proliferation index in MM, dysregulation of a cyclin D gene seems to be a unifying oncogenic event. Analysis of 34 MM cell lines indicates that tumors having an IgH translocation are significantly over‐represented, whereas tumors that ectopically express cyclin D1 are not represented. We speculate that ectopic cyclin D1 expression without t(11;14) is dependent on tumor‐specific interaction with bone marrow stromal cells.     

Cyclins D1 and D3 and topoisomerase II alpha in inactive pituitary adenomas

The oncogenes cyclin D1 and D3 are overexpressed in many tumors. Topoisomerase IIα is found in proliferating cells. The immunohistological expression of cyclin D1, cyclin D3, and Topoisomerase IIα was studied in a collection of 60 clinically inactive surgically removed pituitary adenomas of the follicle-stimulating hormone/luteinizing hormone (FSH/LH) cell complex (20 null cell adenomas, 20 oncocytomas, and 20 FSH/LH cell adenomas) for correlation with other proliferation markers (Ki-67, PCNA) and with clinical data. Whereas cyclin D1 was positive only in one invasive null cell adenoma (1.7%) with some p53-positive nuclei, cyclin D3 was overexpressed in the nuclei of 41 tumors (68%). Topoisomerase IIα was demonstrated in the nuclei of 42 adenomas (70%) with no significant differences discernible between the three adenoma subtypes. There was no significant correlation to the time of development of tumor symptoms, but a correlation of Topoisomerase IIα with cyclin D3 and the proliferation marker Ki-67 (Mib1). From these data we conclude that cyclin D3 and Toposomerase IIα appear to be additional markers for proliferation which can be used for prognosis index in surgical pathology of the pituitary.     

Vitamin D and its analog EB1089 induce p27 accumulation and diminish association of p27 with Skp2 independent of PTEN in pituitary corticotroph cells

Disruption of the gene for the cyclin dependent kinase inhibitor (CDKI) p27/kip1 results in pituitary corticotroph hyperplasia while diminished expression of this protein has been described in aggressive human pituitary tumors. We have previously shown that 1,25-vitamin D3 (VD) hypophosphorylates p27 and interferes with the degradation of this CDKI in thyroid carcinoma cells. In this study we investigated whether VD/EB1089 can induce p27 accumulation and cause growth arrest of pituitary corticotroph cells. VD and EB1089 exhibited a significant reduction in AtT20 corticotroph but not PRL235 lactotroph cell growth. These changes were accompanied by selective accumulation of p27 in AtT20 but not in PRL235 cells. As p27 levels are highly dependent on protein degradation, we examined the effect of VD/EB1089 on p27 association with factors that target this CDKI to the proteasome. VD/EB1089 significantly restricted the association of p27 with Skp2 as well as with cyclin dependent kinase 2 (CDK2). As the tumor suppressor and phosphatase PTEN has been implicated in p27 regulation, we tested whether the effects of VD/EB1089 on p27 accumulation in corticotrophs could be mediated through this pathway. VD/EB1089 did not appreciably alter PTEN expression. Moreover, transfection of PTEN did not influence the effect of VD on p27 accumulation in corticotrophs. We conclude that VD/EB1089 can selectively arrest pituitary corticotroph growth and induce p27 accumulation.This effect is mediated at least partially through diminished p27 association with Skp2 and with CDK2. In contrast to other cell systems, PTEN does not participate in the regulation of corticotroph p27 and is not involved in mediating the effect of VD on p27 in these cells. Our findings highlight p27 and VD analogs as targets for manipulation and drug development respectively in the treatment of inoperable corticotroph adenomas.     

Cyclin D3 immunoreactivity in gastrointestinal stromal tumors is independent of cyclin D3 gene amplification and is associated with nuclear p27 accumulation.

An abnormal expression of cyclin D3, a key regulator of the cell cycle, has been documented in a variety of human malignancies, and the cyclin D3 gene, mapping to 6p21, may be deregulated in human tumors as a result of the t(6;14)(p21.1;q32.3) translocation or gene amplification. In the current study, we for the first time investigated by immunohistochemistry and fluorescence in situ hybridization (FISH) the prevalence of cyclin D3 abnormalities in gastrointestinal stromal tumors (GISTs), comparing the results with traditional pathological characteristics, p27 immunoreactivity (IR), and Ki-67 labeling index (LI). All the tumors showed nuclear cyclin D3 IR, with a percentage of immunostained neoplastic cells ranging from 10 to 95% (mean, 67.3 +/- 22.9%). In 4 (40%) of the 10 cases analyzed by FISH, cyclin D3 extrasignals were detected. Cohybridization with probes specific for the centromeric region and the long arm of chromosome 6 indicated trisomy in one case, whereas in the remaining three cases the pattern was highly suggestive for the occurrence of an isochromosome 6p. There was no association between the cyclin D3 gene copy number and IR for the encoded protein. Cyclin D3 IR was positively associated with p27 IR (P =.004) but not with Ki-67 LI or tumor malignant potential. On the contrary, p27 IR was inversely associated with Ki-67 LI (P =.004) and was more prevalent in tumors of low or intermediate malignant potential, though at a borderline level of statistical significance (P =.066) only. These data suggest that cyclin D3 expression in GISTs is independent of gene amplification and that this protein may be involved in the pathogenesis of GISTs by counteracting the inhibitory activities of p27.