Search

Dobson L, Conway C, Hanley A, Johnson A, Costello S, O’Grady A, Connolly Y, Magee H, O’Shea D, Jeffers M & Kay E (2010) Histopathology 57, 27–38 Image analysis as an adjunct to manual HER-2 immunohistochemical review: a diagnostic tool to standardize interpretation Aims: Accurate determination of HER-2 status is critical to identify patients for whom trastuzumab treatment will be of benefit. Although the recommended primary method of evaluation is immunohistochemistry, numerous reports of variability in interpretation have raised uncertainty about the reliability of results. Recent guidelines have suggested that image analysis could be an effective tool for achieving consistent interpretation, and this study aimed to assess whether this technology has potential as a diagnostic support tool. Methods and results: Across a cohort of 275 cases, image analysis could accurately classify HER-2 status, with 91% agreement between computer-aided classification and the pathology review. Assessment of the continuity of membranous immunoreactivity in addition to intensity of reactivity was critical to distinguish between negative and equivocal cases and enabled image analysis to report a lower referral rate of cases for confirmatory fluorescence in situ hybridization (FISH) testing. An excellent concordance rate of 95% was observed between FISH and the automated review across 136 informative cases. Conclusions: This study has validated that image analysis can robustly and accurately evaluate HER-2 status in immunohistochemically stained tissue. Based on these findings, image analysis has great potential as a diagnostic support tool for pathologists and biomedical scientists, and may significantly improve the standardization of HER-2 testing by providing a quantitative reference method for interpretation.

Background and Aims: Trastuzumab provides clinical benefit for advanced and early breast cancer patients whose tumours over-express or have gene amplification of the HER2 oncogene. The UK National External Quality Assessment Scheme (NEQAS) for immunohistochemical testing was established to assess and improve the quality of HER2 immunohistochemical testing. However, until recently, no provision was available for HER2 fluorescence in situ hybridisation (FISH) testing. A pilot scheme was set up to review the performance of FISH testing in clinical diagnostic laboratories. Methods: FISH was performed in 6 reference and 31 participating laboratories using a cell line panel with known HER2 status. Results: Using results from reference laboratories as a criterion for acceptable performance, 60% of all results returned by participants were appropriate and 78% either appropriate or acceptable. However, 22.4% of results returned were deemed inappropriate, including 13 cases (4.2%) where a misdiagnosis would have been made had these been clinical specimens. Conclusions: The results of three consecutive runs show that both reference laboratories and a proportion of routine clinical diagnostic (about 25%) centres can consistently achieve acceptable quality control of HER2 testing. Data from a significant proportion of participating laboratories show that further steps are required, including those taken via review of performance under schemes such as NEQAS, to improve quality of HER2 testing by FISH in the “real world”.

For a variety of technical reasons it is rarely possible to study cytogenetic abnormalities in ductal carcinoma in situ (DCIS) using traditional techniques. However, by combining molecular biology and computerized image analysis it is possible to carry out cytogenetic analyses on formalin-fixed, paraffin-embedded tissue, using comparative genomic hybridization (CGH). The purpose of this study was to identify the prevalence of chromosomal amplifications and deletions in high-grade DCIS and to look specifically for unique or consistent abnormalities in this pre-invasive cancer. Twenty-three cases of asymptomatic, non-palpable, screen-detected, high-grade DCIS were examined using CGH on tumour cells obtained from histology slides. All cases showed chromosomal abnormalities. A wide variety of amplifications and deletions were spread across the genome. The most frequent changes were gains of chromosomes 17 (13 of 23), 16p (13 of 23), and 20q (9 of 23) and amplifications of 11q13 (22 of 23), 12q 24.1–24.2 (12 of 23), 6p21.3 (11 of 23), and 1q31-qter (6 of 23). The most frequent deletions were on 13q 21.3–q33 (7 of 23), 9p21 (4 of 23), and 6q16.1 (4 of 23). These findings indicate that high-grade DCIS is, from a cytogenetic viewpoint, an advanced lesion. There was no absolutely consistent finding in every case, but amplification of 11q13 was found in 22 of the 23 cases. The precise significance of this is unknown at present. This region of chromosome 11q harbours a number of known oncogenes, including cyclinD1 and INT2. It is likely that many of these findings are the result of accumulated chromosomal abnormalities, reflecting an unstable genome in established malignancy. Copyright © 1999 John Wiley & Sons, Ltd.

Aneuploidy is an important prognostic factor in many cancers. Chromosome 1 abnormalities are present in most breast carcinomas. These may be part of a non-specific increase in DNA (aneuploid status) or represent a restricted chromosomal abnormality. In 16 breast carcinomas we compared chromosome 1 aneusomy with ploidy status. Patients were selected from a mammographically screened population and interphase tumour nuclei were studied by in situ hybridization using a chromosome 1 pericentromeric probe. Ploidy status was assessed by image cytometry on disaggregated cells from paraffin blocks. Of eight cases showing chromosome 1 aneusomy, six (75%) were aneuploid and two diploid. Six (75%) of the eight eusomic cases were aneuploid. This study demonstrates that chromosome 1 aneusomy does not always reflect a gross aneuploid status but, in some tumours, is part of a more restricted chromosomal abnormality. Interphase cytogenetics, possibly using a small panel of pericentromeric probes, may be more sensitive than DNA cytometry for detecting abnormal nuclear DNA content.

Benign and malignant breast lesions in cytologic preparations can be difficult to distinguish. The authors applied the silver nucleolar organizer region (AgNOR) technique to cytologic preparations obtained from surgical specimens to evaluate its diagnostic usefulness in making this distinction. Sixty-two benign and 36 malignant lesions were examined. AgNORs were counted and evaluated using a subjective scoring technique to examine AgNOR size, shape, and clustering. The benign lesions had a mean count of 4.44 AgNORs/nucleus (95% CI, 2.4–6.5) and the malignant cases, 9.52 AgNORs/nucleus (95% confidence interval, 7.4–11.7; P < .0005). The median score for benign cases was 7 and for malignant cases, 13 ( P < .0001). With a few exceptions, cases with high counts had high scores. The diagnostic accuracy of combined counting and pattern assessment was 90%. The likelihood ratio for correct diagnosis using AgNOR counting was 14:1 and for AgNOR scoring, 13:1.

The authors studied the retention/loss of immunoreactivity in frozen tissue over a period of 8 weeks using a variety of storage strategies. Ki67 immunoreactivity is com monly used to assess the growth fraction of tumors in frozen histopathology sections. Many antigens are lost with prolonged storage, even in frozen specimens. To evaluate the loss of Ki67 in frozen tissue, 30 tumors were studied over a storage period of 8 weeks. Cryostat sections, acetone-fixed and stored at — 20°C for 8 weeks, showed identical immunoreactivity to tissue stained immediately on arrival in the laboratory. The other methods gave less satisfactory results. The authors conclude that routine acetone-fixed frozen sections provide excellent results for up to 8 weeks when stored at - 20°C. Int J Surg Pathol 1(2):135-138, 1993

Dobson L, Conway C, Hanley A, Johnson A, Costello S, O’Grady A, Connolly Y, Magee H, O’Shea D, Jeffers M & Kay E (2010) Histopathology57, 27–38 Image analysis as an adjunct to manual HER-2 immunohistochemical review: a diagnostic tool to standardize interpretation Aims: Accurate determination of HER-2 status is critical to identify patients for whom trastuzumab treatment will be of benefit. Although the recommended primary method of evaluation is immunohistochemistry, numerous reports of variability in interpretation have raised uncertainty about the reliability of results. Recent guidelines have suggested that image analysis could be an effective tool for achieving consistent interpretation, and this study aimed to assess whether this technology has potential as a diagnostic support tool. Methods and results: Across a cohort of 275 cases, image analysis could accurately classify HER-2 status, with 91% agreement between computer-aided classification and the pathology review. Assessment of the continuity of membranous immunoreactivity in addition to intensity of reactivity was critical to distinguish between negative and equivocal cases and enabled image analysis to report a lower referral rate of cases for confirmatory fluorescence in situ hybridization (FISH) testing. An excellent concordance rate of 95% was observed between FISH and the automated review across 136 informative cases. Conclusions: This study has validated that image analysis can robustly and accurately evaluate HER-2 status in immunohistochemically stained tissue. Based on these findings, image analysis has great potential as a diagnostic support tool for pathologists and biomedical scientists, and may significantly improve the standardization of HER-2 testing by providing a quantitative reference method for interpretation.

HER2 and TOP2A genes, located on 17q, can be coamplified in cancer. Overexpression of both genes has been reported in high-grade, androgen-resistant prostate cancer. Both genes have not been compared in a single prostate cancer study and the frequency of TOP2A amplifications in prostate cancer is unknown. Using tissue microarrays, we did immunohistochemistry and fluorescence in situ hybridization for HER2 and TOP2A in 100 prostate cancers (41 localized and 59 advanced) and 42 cases of benign prostatic hyperplasia (BPH). Amplification was defined as a target/centromere signal ratio of ≥1.5. HER2 immunohistochemistry was scored from 0 to 3+. Percentage nuclei staining for topoisomerase IIα (topoIIα) was recorded; overexpression was defined as ≥5% cells staining. Eighteen (31%) advanced prostate cancers showed topoIIα overexpression; 12 (26%) showed TOP2A low-level amplification; 9 (16%) expressed HER2; and 6 (13%) showed HER2 low-level amplification. No high-level amplification of either gene (target/centromere signal ratio of ≥3.0) was detected. TOP2A coexpression and coamplification were seen in 75% and 66% of HER2-positive cases, respectively. Localized prostate cancer or BPH showed no gene amplification or topoIIα overexpression. Gene amplification or overexpression correlated with high stage and Gleason score. The presence of TOP2A amplification in advanced cancer was associated with androgen resistance and decreased survival by multivariate analysis. This is the first study to document low-level TOP2A amplification in prostate cancer and an association with reduced survival. TOP2A amplification may occur with or without HER2 duplication and is often associated with topoIIα expression. Therapies directed against topoIIα (and HER2) in such patients may improve survival. [Cancer Res 2007;67(6):2893–8]

Background: The majority of cases of breast cancer scoring HER2 weak positive (2+) on immunohistochemistry (IHC) using the HercepTest are not associated with amplification of the HER2/neu gene. Aim: To examine the reproducibility of IHC in cases scoring 2+ subsequently shown to have gene amplification by fluorescence in-situ hybridisation (FISH). Methods: A retrospective analysis of 153 cases referred for FISH confirmation of a weak positive HercepTest (2+) result was performed. Repeat IHC was undertaken in cases with weak positive (2+) referral IHC and amplification of the HER2 gene by FISH. Results: Amplification of the HER2 gene was confirmed in 29/153 cases (19%) scoring 2+ on IHC. Repeat IHC was carried out on 25 IHC 2+ cases: 7 (28%) scored 2+ on repeat IHC, 18 (72%) scored 3+ and were reclassified as strong positive. A heterogeneous expression pattern was present in 3/17 cases scoring 3+. Conclusions: The majority of HercepTest 2+ results are not accompanied by gene amplification and represent “false positive” IHC in terms of prognostic or therapeutic relevance. A small proportion of HercepTest 2+ scores represent true 2+ IHC positive cases accompanied by gene amplification: a category probably biologically related to 3+ IHC cases. The remainder of cases of HercepTest™ 2+ accompanied by gene amplification represent a category of referral IHC 2+ weak positive, FISH amplified, repeat IHC 3+ strong positive, best described as “false negative 2+ IHC”. This has implications for selection of cases for FISH analysis where weak positive (2+) IHC score is used as a triage for FISH testing, and for testing strategies in referral laboratories undertaking FISH analysis.

Because of the relative rarity of ductal carcinoma in situ (DCIS) in the premammographic screening era, the unavailability of adequate fresh tissue for culture, and the lack of cytogenetic expertise in most pathology departments, there is little information on karyotypic abnormalities in DCIS. The purpose of this study was to investigate the frequency of chromosome 1 aneusomy in DCIS, using interphase cytogenetic techniques, and to correlate the findings with nuclear grade. Twenty-one cases of DCIS, identified in a mammographically screened population, were studied by in situ hybridization. Chromosome 1 numbers were identified in interphase nuclei in conventional histology sections, using a specific centromeric probe (pUC 1.77). In each case, 100 tumour nuclei were compared with 100 normal nuclei. Eighteen of 21 (86 per cent) cases were aneusomic for chromosome 2. This included 15 of 16 (94 per cent) pure comedo or predominantly comedo DCIS. Fifteen of 16 (94 per cent) DCIS with grade 3 nuclei and 3 of 5 (60 per cent) cases with grade 2 nuclei were aneusomic. One case with grade 3 nuclei (a comedo carcinoma) was negative. We conclude that chromosome 1 aneusomy precedes invasion and is a relatively consistent occurrence in those DCIS with high nuclear grade.

Summary Aims: To examine the pattern of reactivity of Ki-67 in neuroblastoma and correlate this with a) clinical prognostic criteria and b) cell cycle statistics (using flow cytometry). Methods: Four micron sections of paraffin embedded (PE) tissue from 55 patients (25 pre chemotherapy and 30 post) were placed on to aminosialinised slides, de-waxed and rehydrated. Slides were pretreated in a microwave oven, endogenous per-oxidase activity blocked using 3% hydrogen peroxide and Ki-67 reactivity investigated using a streptavidin/biotin/peroxidase technique. DNA ploidy was also performed from an immediately adjacent section on the same block using a FACScan and Cellfit software. Results: Ki-67 reactivity was well defined and highly reproducible. Eighteen out of 30 post chemotherapy samples were totally negative, despite evidence of proliferation on flow cytometry and all subsequently died of disease. As interpretation post chemotherapy was therefore deemed unreliable, this group was excluded from analysis. Reactivity in pretreatment samples ranged from 0% to 67%; staining was restricted to the nucleus with a distinct pattern noted in the nucleolus. Ki-67 positivity was lower in aneuploid compared with diploid tumours (mean 26% vs 36%, NS). Among diploid tumours, a lower percentage positivity was noted in those patients with better clinical prognostic parameters. Correlation however between Ki-67 and SG 2 M phases of cell cycle was poor (R S = 0.39, NS). Conclusion: Assessment of proliferation in neuroblastoma by Ki-67 reactivity in paraffin embedded tissue is reliable in pretreatment samples and can be incorporated into routine immunohistochemical evaluation. Larger multicentre studies are required to further evaluate Ki-67 reactivity as a prognostic indicator.

Pigmented solar keratosis (PSK) is sometimes clinically indistinguishable from lentigo maligna, a form of malignant melanoma in situ. Occasionally histologic diagnosis is also difficult. Accurate diagnosis is essential, as the treatment and prognosis for each condition differs considerably. To determine whether there was a significant overlap in the number of melanocytes in these sun-damaged skin lesions, or whether immunohistochemistry might be helpful in the differential diagnosis, the authors examined skin biopsy specimens from 26 patients with obvious lentigo maligna and 15 patients with PSK using 3 monoclonal antibodies (HMB-45, NK1C3, and vimentin) and 1 polyclonal antibody (S-100 protein). Formalin-fixed paraffin sections were immunostained with each of the above antibodies, and immunopositive cells per mm2 of epidermis were counted. The difference between lentigo maligna and PSK counts was statistically significant at a level of P < .0001; furthermore, there was almost no overlap between the two groups. The sensitivity for the diagnosis of lentigo maligna was high with all antibodies. However, HMB-45 had the highest sensitivity and the lowest false-positive rate and was visually most pleasing. Using a cut-off count of 60 cells per mm2 of epidermis, HMB-45 had a sensitivity of 96% and a 0% false-positive rate. In this study, lentigo maligna was easily differentiated from PSK. The real value of immunohistochemistry in the differential diagnosis of these pigmented lesions should be tested in a prospective study using cases that are difficult to diagnose by routine light microscopy.