Immunocytochemical progesterone receptor (PR) assays employing two different monoclonal antireceptor antibodies (mPRi, JZB39) were compared with each other and with the dextran-coated charcoal assay (DCC) in human breast carcinoma. The immunocytochemical analyses were semiquantitatively scored (histoscore) based on relative staining intensity and the percentage of positively stained carcinoma cell nuclei. The immunocytochemically determined PR status agreed with that obtained by the DCC assay in 76% of the cases when the monoclonal antibody mPRi was used. The agreement was slightly better (82%) with the immunocytochemical assay using the JZB39 antibody. The semiquantitative histoscores correlated significantly with log-transformed cytosol PR concentrations in the DCC assay (mPRi, R = 0.543; JZB39, R = 0.618; P less than 0.0001 for both). The histoscores obtained with the mPRi and JZB39 antibodies were highly correlated with each other in adjacent frozen sections (R = 0.84, P less than 0.0001) the latter antibody giving somewhat higher histoscores. These results further validate the use of immunocytochemical PR assays for routine diagnostic purposes, especially when sufficient material is not available for DCC assays.
In our recent cancer registry-based study, the incidence of gastric carcinoma was increased up to 5-fold in male relatives of early-onset prostate cancer (PCA) patients. This association may reflect the influence of genetic factors predisposing individuals to both tumor types. Germ-line mutations of the CDH1 gene at 16q have recently been associated with familial gastric cancer. Furthermore, two genome-wide linkage studies of PCA recently reported positivity at 16q. We therefore identified families and individual patients with both gastric and PCA and investigated whether the CDH1 gene mutations were involved in cancer predisposition in these cases. Fifteen of the 180 Finnish hereditary PCA families (8.3%) had one or more gastric cancer cases. No truncating or splice site CDH1 mutations were identified by PCR single-strand conformational polymorphism in these families or in eight individual patients who had both prostate and gastric cancer. However, a novel S270A missense mutation in exon 6 of the CDH1 gene was seen in a single family with four prostate and two gastric cancers. A large-scale population-based survey indicated a higher prevalence of S270A among both familial PCA cases (3.3%; n = 120; P = 0.01) and unselected PCA patients (1.5%; n = 472; P = 0.12) as compared with blood donors serving as population controls (0.5%; n = 923). We conclude that individual rare mutations and polymorphisms in the CDH1 gene, such as S270A, may contribute to the onset of PCA and warrant further investigations in other populations. However, the CDH1 gene does not appear to explain the link between prostate and gastric cancer.
Loss of heterozygosity (LOH) on chromosome 10 is the most common genetic changes in glioblastomas of World Health Organization (WHO) Grade IV.1, 2, 3, 4, 5, 6, 7 In most cases, the allelic losses involve loci along the entire chromosome, a finding in line with monosomy 10. LOH on chromosome 10 may also be detected in anaplastic astrocytomas (WHO Grade III) and diffuse astrocytomas (WHO Grade II), but is less common than in glioblastomas and more frequently restricted to partial losses.7 In astrocytomas with partial LOH of chromosome 10, 3 regions, i.e., 10p14–pter, 10q23–q24 and 10q25–qter, have been identified.8, 9 Whereas the PTEN/MMAC1 tumor suppressor gene (TSG) at 10q23 is mutated in approximately one-third of glioblastomas,10 target genes for 10p14–pter and 10q25–qter remain undiscovered. At 10p15, the presence of a TSG has been suggested by a cell fusion study,11 and introduction of 10p15 chromosomal fragment into glioblastoma cell line T98G suppressed colony formation in soft agar.12 A candidate TSG, KLF6 (Krüppel-like factor 6) on chromosome 10p15, undergoes alterations in up to 55% of prostate carcinomas by deletion, point mutations and loss of expression.13, 14 KLF6 is a transcription factor involved in the regulation of cell proliferation and differentiation. Moreover, functional studies showed that, whereas wild-type KLF6 upregulates p21 (WAF1/CIP1) in a p53-independent manner and leads to reduced cell proliferation, tumor-derived KLF6 mutants do not.13 In a recent study,15 mutations of KLF6 were found in Grade II–IV astrocytic brain tumors. We therefore investigated the possible involvement of KLF6 in astrocytoma formation by detecting its mutations in an independent tumor series. No mutation was detected in 52 astrocytomas of various WHO grades and 30 glioma cell lines, questioning the role of KLF6 in brain tumor development. We obtained formalin-fixed, paraffin-embedded tumor specimens from 52 patients who were diagnosed and operated for astrocytomas at the Tampere University Hospital between 1991–2001. These included 10 pilocytic astrocytomas (WHO Grade I), 9 WHO Grade II astrocytomas, 11 anaplastic astrocytomas (WHO Grade III), and 22 glioblastomas (WHO Grade IV). Snap-frozen tumor samples were also available for 30 of the samples (WHO Grade I, 10 tumors; WHO Grade II, 3 tumors; WHO Grade III, 1 tumor; WHO Grade IV, 16 tumors). A neuropathologist (H.H.) classified and graded the tumors according to the WHO classification system.16 The research protocol was approved by the Ethical Committee of the Tampere University Hospital. Thirty glioblastoma cell lines were used in our study. These included LN-series (LN-18, -71, -215,-229, -235, -Z308, -340, -343, -401, -405, -427, -428, -432, -443, -444, -462, -464, -702, -767, -827, -963 and -992), U87MG, U118MG, U138MG, U251MG, U343MG, A172, SF767 and D247MG. Their source, culture conditions and genetic status were described previously.17 DNA was isolated using QIAamp DNA Mini Kit (Qiagen, Hilden, Germany). Formalin-fixed tissues were deparaffinized with xylene before DNA isolation. Mutation detection was done using single-strand conformation polymorphism (SSCP) as described previously.18 In the first round of SSCP analysis using DNA from formalin-fixed and paraffin-embedded tissues, some shifted bands were detected. None of the shifts was reproduced in the repeated SSCP analysis and DNA sequencing, however, suggesting that the sequence alterations resulted from PCR artifacts. To further clarify if mutations occurred in astrocytoma, we analyzed DNA from fresh tissues for cases that showed band shifts, and found that none of the shifted bands could be detected in fresh tissues. Only the 201G>A (Arg201Arg) germ-line polymorphism, which is present in 3.8% of healthy subjects (unpublished data), was detected in 2 tumors. We also analyzed 30 cell lines using the denaturing high performance liquid chromatography (DHPLC) method,14 and again, no mutations were found in any of them. Oligonucleotide primers used are available upon request. We thereafter conclude that mutations of KLF6 are rather infrequent in astrocytoma. Previously, Jeng and Hsu15 examined KLF6 for mutations in 155 formalin-fixed astrocytomas, and detected mutations in 5/79 (6%) of Grade II and III astrocytomas as well as in 9/76 (12%) of glioblastomas. In our present study, we could not detect any KLF6 mutations in 30 WHO Grade II–III astrocytomas, 22 glioblastomas and 30 malignant glioma cell lines. A similar discrepancy in the frequency of KLF6 mutation has also been observed in prostate cancer, as the high frequency of KLF6 mutation initially reported by Narla et al.13 has not been confirmed in 2 following studies, which showed KLF6 mutations in 0–15% of tumors and none of the prostate cancer cell lines.14, 19 The difference in mutation frequencies between studies could be due to sample selection, cell isolation or analytic techniques used.20 In addition, the problem of non-specific "mutations" in formalin-fixed tumors, which has been acknowledged earlier in the study of TP53 mutations,21 merits attention in future investigations of KLF6 mutations. We conclude that somatic mutations of KLF6 do not play a major role in astrocytoma formation. We are grateful to S. Kekkonen, R. Randen, S.N. Devi and Z. Zhang for assistance with samples and laboratory analyses. Our work was supported by grants from the Nona and Kullervo Väre Foundation, Medical Research Fund of Tampere University Hospital (to E.-R.H. and P.A.K.), the Finnish Cancer Institute (to P.A.K.), and the United States of America National Institutes of Health grant CA86335 (to E.G.V.M.). Yours sincerely, Pasi A. Koivisto, Xiaohui Zhang, Satu-Leena Sallinen, Pauli Sallinen, Heikki J. Helin, Jin-Tang Dong, Erwin G. Van Meir, Hannu Haapasalo, Eija-Riitta Hyytinen
Monoclonal antibodies to human estrogen receptor (ER), and rabbit progesterone receptor (PR), also recognizing human PR, were used to detect the receptors by peroxidase immunocytochemistry in frozen sections of 124 primary breast carcinomas. Both ER and PR were almost exclusively located in carcinoma cell nuclei, with heterogenous distribution and intensity. The staining results were evaluated semiquantitatively (histoscore), based on the percentage of positively stained carcinoma cells and nuclear staining intensity. The receptor status thus determined was as follows: ER+PR+ in 50 patients, ER+PR− in 23, ER−PR− in 26, and ER−PR+ in 3 patients. There was a 79% (ER) or 70% (PR) agreement in the positivity/negativity between the immunocytochemical and steroid-binding assay (in 102 patients) with a highly significant correlation. The his-toscore values increased significantly with cytosol receptor levels (ER, r = 0.623, P < 0.001; PR, r = 0.366, P < 0.01).
