Improving practice in PD-L1 testing of non-small cell lung cancer in the UK: current problems and potential solutions.

AIMS: Programmed cell death ligand 1 (PD-L1) expression, used universally to predict response of non-small cell lung cancer (NSCLC) to immune-modulating drugs, is a fragile biomarker due to biological heterogeneity and challenges in interpretation. The aim of this study was to assess current PD-L1 testing practices in the UK, which may help to define strategies to improve its reliability and consistency. METHODS: A questionnaire covering NSCLC PD-L1 testing practice was devised and members of the Association of Pulmonary Pathologists were invited to complete this online. RESULTS: Of 44 pathologists identified as involved in PD-L1 testing, 32 (73%) responded. There was good consistency in practice and approach, but there was wide variability in the distribution of PD-L1 scoring. Although the proportions of scores falling into the three groups (negative, low and high) defined by the 1% and 50% 'cut-offs' (38%, 33% and 27%, respectively) reflect the general experience, the range within each group was wide at 23-70%, 10-60% and 15-36%, respectively. CONCLUSIONS: There is inconsistency in the crucial endpoint of PD-L1 testing of NSCLC, the expression score that guides management. Addressing this requires formal networking of individuals and laboratories to devise a strategy for its reduction.

Machine-learning-based image analysis algorithms improve interpathologist concordance when scoring PD-L1 expression in non-small-cell lung cancer.

Programmed death ligand 1 (PD-L1) expression on tumour cells is the only predictive biomarker of response to immuno-modulatory therapy for patients with non-small-cell lung cancer (NSCLC). Accuracy of this biomarker is hampered by its challenging interpretation. Here we explore if the use of machine-learning derived image analysis tools can improve interpathologist concordance of assessing PD-L1 expression in NSCLC.Five pathologists who routinely score PD-L1 at a major regional referral hospital for thoracic surgery participated. 13 NSCLC small diagnostic biopsies were stained for PD-L1 (SP263 clone) and digitally scanned. Each pathologist independently scored each case with and without the Roche uPath PD-L1 (SP263) image analysis NSCLC algorithm with a wash-out interim period of 6 weeks.A consistent improvement in interpathologist concordance was seen when using the image analysis tool compared with scoring without: (Fleiss' kappa 0.886 vs 0.613 (p<0.0001) and intraclass coefficient correlation 0.954 vs 0.837 (p<0.001)). Five cases (38%) were classified into clinically relevant different categories (negative/weak/strong) by multiple pathologists when not using the image analysis algorithm, whereas only two cases (15%) were classified differently when using the image analysis algorithm.The use of the image analysis algorithm improved the concordance of assessing PD-L1 expression between pathologists. Critically, there was a marked improvement in the placement of cases into more consistent clinical groupings. This small study is evidence that the use of image analysis tools may improve consistency in assessing tumours for PD-L1 expression and may therefore result in more consistent prediction to targeted treatment options.

Landscape of cancer biomarker testing in England following genomic services reconfiguration: insights from a nationwide pathologist survey.

AIMS: Cancer diagnostics have been evolving rapidly. In England, the new National Health Service Genomic Medicine Service (GMS) provides centralised access to genomic testing via seven regional Genomic Laboratory Hubs. The PATHways survey aimed to capture pathologists' experience with current diagnostic pathways and opportunities for optimisation to ensure equitable and timely access to biomarker testing. METHODS: A nationwide survey was conducted with consultant pathologists from regional laboratories, via direct interviews based on a structured questionnaire. Descriptive analysis of responses was undertaken using quantitative and qualitative methods. RESULTS: Fifteen regional centres completed the survey covering a median population size of 2.5 (1.9-3.6) million (each for n=12). The median estimated turnaround time (calendar days) for standard molecular markers in melanoma, breast and lung cancers ranged from 2 to 3 days by immunohistochemistry (excluding NTRKfus in breast and lung cancers, and PD-L1 in melanoma) and 6-15 days by real-time-PCR (excluding KIT for melanoma), to 17.5-24.5 days by next-generation sequencing (excluding PIK3CA for breast cancer). Tests were mainly initiated by pathologists and oncologists. All respondents discussed the results at multidisciplinary team (MDT) meetings. The GMS roll-out was perceived to have high impact on services by 53% of respondents, citing logistical and technical issues. Enhanced education on new pathways, tissue requirements, report interpretation, providing patient information and best practice sharing was suggested for pathologists and other MDT members. CONCLUSION: Our survey highlighted the role of regional pathology within the evolving diagnostic landscape in England. Notable recommendations included improved communication and education, active stakeholder engagement, and tackling informatics barriers.

Biomarker Testing for People With Advanced Lung Cancer in England.

INTRODUCTION: Optimal management of people with advanced NSCLC depends on accurate identification of predictive markers. Yet, real-world data in this setting are limited. We describe the impact, timeliness, and outcomes of molecular testing for patients with advanced NSCLC and good performance status in England. METHODS: In collaboration with Public Health England, patients with stages IIIB to IV NSCLC, with an Eastern Cooperative Oncology Group performance status of 0 to 2, in England, between June 2017 and December 2017, were identified. All English hospitals were invited to record information. RESULTS: A total of 60 of 142 invited hospitals in England participated in this study and submitted data on 1157 patients. During the study period, 83% of patients with advanced adenocarcinoma underwent molecular testing for three recommended predictive biomarkers (EGFR, ALK, and programmed death-ligand 1). A total of 80% of patients with nonsquamous carcinomas on whom biomarker testing was performed had adequate tissue for analysis on initial sampling. First-line treatment with a tyrosine kinase inhibitor was received by 71% of patients with adenocarcinoma and a sensitizing EGFR mutation and by 59% of those with an ALK translocation. Of patients with no driver mutation and a programmed death-ligand 1 expression of greater than or equal to 50%, 47% received immunotherapy. CONCLUSIONS: We present a comprehensive data set for molecular testing in England. Although molecular testing is well established in England, timeliness and uptake of targeted therapies should be improved.

Cellular pathology in the COVID-19 era: a European perspective on maintaining quality and safety.

COVID-19 is a zoonotic viral infection that originated in Wuhan, China, in late 2019. WHO classified the resulting pandemic as a 'global health emergency' due to its virulence and propensity to cause acute respiratory distress syndrome. The COVID-19 pandemic has had a major impact on diagnostic laboratories, particularly those handling cell and tissue specimens. This development carries serious implications for laboratory practice in that safety of personnel has to be balanced against high-quality analysis and timely reporting of results. The aim of this article is to present some recommendations for the handling of such specimens in the preanalytical, analytical and postanalytical phases of laboratory testing and analysis in an era of high COVID-19 prevalence, such as that seen, for example, in the UK, Spain, Italy and France.

Common variable immunodeficiency with granulomatous-lymphocytic interstitial lung disease and preceding neurological involvement: a case-report.

BACKGROUND: Common variable immunodeficiency (CVID) is a group of heterogeneous primary immunodeficiencies characterised by a dysregulated and impaired immune response. In addition to an increased susceptibility to infection, it is also associated with noninfectious autoimmune and lymphoproliferative complications. CVID is rarely associated with neurological complications. Pulmonary involvement is more common, and patients can develop an interstitial lung disease known as granulomatous-lymphocytic interstitial lung disease (GLILD). CASE PRESENTATION: A 50-year-old Caucasian female with a history of Evans syndrome (idiopathic thrombocytopaenic purpura and autoimmune haemolytic anaemia) and hypogammaglobulinaemia initially presented to the neurology clinic with marked cerebellar ataxia and headaches. Following extensive investigation (which included brain biopsy), she was diagnosed with neuro-sarcoidosis and her symptoms resolved following treatment with immunosuppressive therapy. Over the following 10 years, she was extensively investigated for recurrent pulmonary infections and abnormal radiological findings, which included pulmonary nodules, infiltrates and splenomegaly. Subsequently, she was referred to an immunology clinic, where immunoglobulin replacement treatment was started for what was ultimately considered to be CVID. Shortly afterwards, evaluation of her clinical, radiological and histological findings at a specialist interstitial lung disease clinic led to a diagnosis of GLILD. CONCLUSION: CVID is a condition which should be suspected in patients with immunodeficiency and recurrent infections. Concomitant autoimmune disorders such as haemolytic anaemia and immune thrombocytopenia may further support the diagnosis. As illustrated in this case, there is a rare association between CVID and inflammatory involvement of the neurological system. Respiratory physicians should also suspect CVID with associated GLILD in patients with apparent pulmonary granulomatous disease and recurrent infections. In addition, this case also highlights the challenge of diagnosing CVID and its associated features, and how the definitive exclusion of other pathologies such as malignancy, mycobacterial infection and lymphoma is required as part of this diagnostic process.

Cytology for PD-L1 testing: A systematic review.

Evaluation of tumoral programmed cell death ligand-1 (PD-L1) expression is standard practice for patients with advanced non-small-cell lung cancer (NSCLC) who may be candidates for treatment targeting the programmed cell death-1 (PD-1)/PD-L1 pathway. Currently, all of the commercially available immunohistochemistry assays have been validated for use with histology specimens although, in routine clinical practice, approximately 30-40 % of patients with advanced NSCLC have only cytology specimens available for diagnosis, staging, and biomarker analysis. This systematic review evaluated the success rate, concordance, and clinical utility of using cytology specimens to assess tumor PD-L1 expression levels compared with histology specimens from patients with advanced NSCLC. EMBASE and PubMed database searches identified 142 unique, relevant publications, of which 15 met the inclusion criteria for at least one analysis. In 709 specimens, across seven publications, the proportion of cytology specimens evaluable for PD-L1 testing was 92.0 %. Among nine studies eligible for concordance analysis between cytology and histology specimens at a PD-L1 tumor cell expression cutoff of ≥50 %, overall percentage agreement was 89.7 % (n = 428), 72.0 % for positive percentage agreement (n = 218), and 95.0 % for negative percentage agreement (n = 258); results using a tumor PD-L1 expression cutoff of ≥1 % were similar. Our analyses suggest that using cytology specimens to assess PD-L1 expression is feasible, with good levels of concordance between cytology and histology specimens using PD-L1 tumor cell expression cutoffs of ≥1 % and ≥50 %. In conclusion, there is no convincing evidence that cytology specimens are inadequate or inferior to histology specimens for assessing PD-L1 expression in patients with NSCLC.

Accelerated instability testing reveals quantitative mass spectrometry overcomes specimen storage limitations associated with PD-L1 immunohistochemistry.

Immunohistochemistry (IHC) using formalin-fixed, paraffin embedded (FFPE) tissue is limited by epitope masking, posttranslational modification and immunoreactivity loss that occurs in stored tissue by poorly characterized mechanisms. Conformational epitopes recognized by many programmed-death-ligand-1 (PD-L1) IHC assays are particularly susceptible to degradation and provide an ideal model for understanding signal loss in stored FFPE tissue. Here we assessed 1206 tissue sections to evaluate environmental factors impacting immunoreactivity loss. PD-L1 IHC using four antibodies (22C3, 28-8, E1L3N, and SP142), raised against intracellular and extracellular epitopes, was assessed in stored FFPE tissue alongside quantitative mass spectrometry (MS). Global proteome analyses were used to assess proteome-wide oxidation across an inventory of 3041 protein groups (24,737 distinct peptides). PD-L1 quantitation correlated well with IHC expression on unaged sections (R2 = 0.744; P < 0.001), with MS demonstrating no loss of PD-L1 protein, even in sections with significant signal loss by IHC impacting diagnostic category. Clones 22C3 and 28-8 were most susceptible to signal loss, with E1L3N demonstrating the most robust signal (56%, 58%, and 33% reduction respectively; p < 0.05). Increased humidity and temperature resulted in significant acceleration of immunoreactivity loss, which was mitigated by storage with desiccant. MS demonstrated only modest oxidation of 274 methionine-containing peptides and aligned with IHC results suggesting peptide oxidation is not a major factor. These data imply immunoreactivity loss driven by humidity and temperature results in structural distortion of epitopes rendering them unsuitable for antibody binding following epitope retrieval. Limitations of IHC biomarker analysis from stored tissue sections may be mitigated by cost-effective use of desiccant when appropriate. In some scenarios, complementary MS is a preferred approach for retrospective analyses of archival FFPE tissue collections.

Programmed death ligand 1 expression in EBUS aspirates of non-small cell lung cancer: Is interpretation affected by type of fixation?

BACKGROUND: Much of the reluctance about using cytology specimens rather than histology specimens to assess programmed death ligand 1 (PD-L1) expression for guiding the use of immune modulating drugs in the management of non-small cell lung cancer (NSCLC) is based on the belief that the alcohol-based fixatives favored by cytopathologists might reduce the antigenicity of PD-L1 and lead to artifactually low expression levels and false-negative reporting. Therefore, this study was performed to determine whether there is any difference in PD-L1 expression between endobronchial ultrasound (EBUS)-guided aspirates of NSCLC fixed in alcohol-based fixatives and those fixed in neutral buffered formalin (NBF), the standard laboratory fixative for histology specimens. METHODS: The expression of PD-L1 was compared in 50 paired EBUS aspirates of NSCLC taken from the same lymph node during the same procedure. One aspirate of each pair was fixed in an alcohol-based fixative, and the other was fixed in NBF. RESULTS: In none of the 50 pairs was there any significant difference, qualitative or quantitative, in the strength, pattern, or extent of PD-L1 expression. In the great majority, the expression was identical, regardless of fixation. CONCLUSIONS: There is no evidence from this study showing that the use of alcohol-based fixatives has any effect on the expression of PD-L1 or its interpretation. Notwithstanding the general challenges in accurately assessing such expression in cytology specimens, pathologists should feel able to interpret them with confidence, and clinicians should feel able to rely on the results.

Heterogeneity of PD-L1 expression in non-small cell lung cancer: Implications for specimen sampling in predicting treatment response.

OBJECTIVES: PD-L1 expression on tumour cells can guide the use of anti-PD-1/PD-L1 immune modulators to treat patients with non-small cell lung cancer (NSCLC). Heterogeneity of PD-L1 expression both within and between tumour sites is a well-documented phenomenon that compromises its predictive power. Our aim was to better characterise the pattern and extent of PD-L1 heterogeneity with a view to optimising tumour sampling and improve its accuracy as a biomarker. MATERIALS AND METHODS: Expression of PD-L1 was assessed by immunochemistry using the SP263 clone in 107 resected primary NSCLCs and their nodal metastases. Intra-tumoural heterogeneity, defined as 'small-scale' (mm²), 'medium-scale' (cm²) and 'large-scale' (between tumour blocks), was assessed by digital imaging using a novel 'squares method'. Inter-tumoural heterogeneity between the primary tumours and their nodal metastases and between N1 and N2 nodal stages was also assessed. RESULTS: The majority of tumours demonstrated intra-tumoural heterogeneity (small-scale 78%, medium-scale 50%, large-scale 46%). Inter-tumoural heterogeneity between the primary and nodal metastases was present in 53% of cases and, in 17%, between N1 and N2 disease. These differences were occasionally sufficient to lead to discrepancy across the ≥1%, ≥25% and ≥50% cut-offs used to guide therapy. CONCLUSION: Heterogeneity of PD-L1 expression is common, variable in scale and extent, and carries significant implications for its accuracy as a predictive biomarker. Extensive sampling reduces, but cannot eliminate, this inaccuracy.

ALK IHC and FISH discordant results in patients with NSCLC and treatment response: for discussion of the question-to treat or not to treat?

INTRODUCTION: Lung cancer is the most common cancer worldwide. Latest guidelines from the College of American Pathologist and the European society of medical oncologists indicate anaplastic lymphoma kinase (ALK) rearrangement testing is standard practice. Historically, diagnostics for ALK used fluorescence in situ hybridisation (FISH); however, immunohistochemical (IHC) assays are becoming common practice. Unfortunately, recent assessment of current practice indicated that not all patients who should be tested for ALK translocation are undergoing ALK testing. METHODS: From a series of European and Israeli labs, we collected patients with discordant IHC and FISH testing, which were subsequently treated with ALK-targeted therapy, for discussion of the question, to treat or not to treat? RESULTS: Our study may support ALK IHC testing as a better predictor of response to targeted therapy provided that the labs implement controlled preanalytical procedures, use correct clone, run protocols on automated staining platforms and validate using external quality assessments.

Programmed Death-Ligand 1 Immunohistochemistry Testing: A Review of Analytical Assays and Clinical Implementation in Non-Small-Cell Lung Cancer.

Purpose Three programmed death-1/programmed death-ligand 1 (PD-L1) inhibitors are currently approved for treatment of non-small-cell lung cancer (NSCLC). Treatment with pembrolizumab in NSCLC requires PD-L1 immunohistochemistry (IHC) testing. Nivolumab and atezolizumab are approved without PD-L1 testing, though US Food and Drug Administration-cleared complementary PD-L1 tests are available for both. PD-L1 IHC assays used to assess PD-L1 expression in patients treated with programmed death-1/PD-L1 inhibitors in clinical trials include PD-L1 IHC 28-8 pharmDx (28-8), PD-L1 IHC 22C3 pharmDx (22C3), Ventana PD-L1 SP142 (SP142), and Ventana PD-L1 SP263 (SP263). Differences in antibodies and IHC platforms have raised questions about comparability among these assays and their diagnostic use. This review provides practical information to help physicians and pathologists understand analytical features and comparability of various PD-L1 IHC assays and their diagnostic use. Methods We reviewed and summarized published or otherwise reported studies (January 2016 to January 2017) on clinical trial and laboratory-developed PD-L1 IHC assays (LDAs). Studies assessing the effect of diagnostic methods on PD-L1 expression levels were analyzed to address practical issues related to tissue samples used for testing. Results High concordance and interobserver reproducibility were observed with the 28-8, 22C3, and SP263 clinical trial assays for PD-L1 expression on tumor cell membranes, whereas lower PD-L1 expression was detected with SP142. Immune-cell PD-L1 expression was variable and interobserver concordance was poor. Inter- and intratumoral heterogeneity had variable effects on PD-L1 expression. Concordance among LDAs was variable. Conclusion High concordance among 28-8, 22C3, and SP263 when assessing PD-L1 expression on tumor cell membranes suggests possible interchangeability of their clinical use for NSCLC but not for assessment of PD-L1 expression on immune cells. Development of LDAs requires stringent standardization before their recommendation for routine clinical use.

COL1A1, PRPF40A, and UCP2 correlate with hypoxia markers in non-small cell lung cancer.

PURPOSE: Collagen 1A1 (COL1A1), RNA-binding and pre-mRNA Processing Factor (PRPF40A), and Uncoupling Protein 2 (UCP2) were identified as downstream effectors of cytoglobin (CYGB), which was shown implicated in tumour biology. Although these three genes have been previously associated with cancer, little is known about their status in lung malignancies. METHODS: Hereby, we investigated the expression and promoter methylation of COL1A1, PRPF40A, and UCP2 in 156 non-small cell lung cancer (NSCLC) and adjacent normal tissues. RESULTS: We demonstrate that COL1A1 and PRPF40A mRNAs are significantly overexpressed in NSCLC (p < 1 × 10-4), while UCP2 exhibits a trend of upregulation (p = 0.066). Only COL1A1 promoter revealed hypermethylation in NSCLCs (36%), which was particularly evident in squamous cell carcinomas (p = 0.024) and in the tumours with moderate-to-good differentiation (p = 0.01). Transcript level of COL1A1, as well as PRPF40A and UCP2, exhibited striking association (p ≤ 0.001) with the expression of hypoxia markers. In addition, we demonstrate in lung cancer cell lines exposed to hypoxia or oxidative stress that COL1A1 transcription significantly responds to oxygen depletion, while other genes showed the modest upregulation in stress conditions. CONCLUSION: In conclusion, our data revealed that COL1A1, UCP2, and PRPF40A are novel players implicated in the complex network of hypoxia response in NSCLC.

Aurora B expression modulates paclitaxel response in non-small cell lung cancer.

BACKGROUND: Taxanes are mitotic poisons widely used in the treatment of non-small cell lung cancer (NSCLC), however, little is known about potential molecular modulators of response to these compounds. Aurora B (AURKB) is a critical regulator of the mitotic spindle assembly, previously shown overexpressed in NSCLC. Here we investigated the hypothesis that AURKB expression modulates the efficacy of taxanes in NSCLC cells. METHODS: AURKB mRNA expression was determined by qPCR in 132 frozen NSCLC tissues and nine NSCLC cell lines. Aurora B expression was knocked down in cell lines using multiple shRNA constructs. Barasertib was used to specifically inhibit AURKB activity, determined by the level of H3S10 phosphorylation. RESULTS: Frequent AURKB mRNA upregulation was observed in NSCLC tissues (P<0.0001), being more prominent in squamous carcinomas (P<0.0001). Aurora B expression in cell lines strongly correlated with sensitivity to both docetaxel (P=0.004) and paclitaxel (P=0.007). Aurora B knockdown derivatives consistently showed a dose-dependent association between low-AURKB expression and resistance to paclitaxel. Specific chemical inhibition of Aurora B activity also demonstrated a strong dose-dependent efficiency in triggering paclitaxel resistance. CONCLUSIONS: Aurora B activity is an important modulator of taxane response in NSCLC cells. This may lead to further insights into taxane sensitivity of NSCLC tumours.

The effect of time changes in diagnosing lung cancer type on its recorded distribution, with particular reference to adenocarcinoma.

Among lung cancers, a substantial shift over time has occurred in the recorded frequency of adenocarcinoma (AdC) relative to that of squamous cell carcinoma (SqCC). This is evident in many countries, and also in those who have never smoked. We attempted to address the extent to which this increase is real, or an artefact of changing diagnostic practices. We reviewed studies re-evaluating diagnoses using more up-to-date criteria, and studies applying standard criteria to cases collected over a long period. We also describe changes to classifications, and factors affecting diagnostic accuracy and consistency. While the four main types have long remained essentially unchanged, successive WHO classifications differ in how tumours are ascribed to these types. Despite refinement of classifications and technological advances, the decision is ultimately the pathologist's. In 11 studies, 189/1212(15.6%) originally diagnosed AdCs were reclassified as non-AdC on review, whereas 541/1564(34.6%) of non-AdCs were reclassified as AdC, increasing AdCs by 30%. Studies examining trends in the proportion of AdC were conflicting; three showing a declining trend, seven no trend, and six some increase. Some studies find lepidic (bronchioloalveolar) carcinoma, but not other AdC sub-types, increased. The rising AdC/SqCC ratio results at least partly from diagnostic changes.

Left ventricular apical masses: distinguishing benign tumours from apical thrombi.

Differential diagnoses for cardiac left ventricular apical masses presenting following acute myocardial infarction include thrombi and cardiac tumours. We present two such cases and the multidisciplinary assessment that is required to assist with diagnosis.

Increased expression of Id1 and Id3 promotes tumorigenicity by enhancing angiogenesis and suppressing apoptosis in small cell lung cancer.

Constant deregulation of Id1 and Id3 has been implicated in a wide range of carcinomas. However, underlying molecular evidence for the joint role of Id1 and Id3 in the tumorigenicity of small cell lung cancer (SCLC) is sparse. Investigating the biological significance of elevated expression in SCLC cells, we found that Id1 and Id3 co-suppression resulted in significant reduction of proliferation rate, invasiveness and anchorage-independent growth. Suppressing both Id1 and Id3 expression also greatly reduced the average size of tumors produced by transfectant cells when inoculated subcutaneously into nude mice. Further investigation revealed that suppressed expression of Id1 and Id3 was accompanied by decreased angiogenesis and increased apoptosis. Therefore, the SCLC tumorigenicity suppression effect of double knockdown of Id1 and Id3 may be regulated through pathways of apoptosis and angiogenesis.

BAI3, CDX2 and VIL1: a panel of three antibodies to distinguish small cell from large cell neuroendocrine lung carcinomas.

AIMS: Discriminating small-cell lung carcinoma (SCLC) from large-cell neuroendocrine carcinoma (LCNEC) rests on morphological criteria, and reproducibility has been shown to be poor. We aimed to identify immunohistochemical markers to assist this diagnosis. METHODS AND RESULTS: Gene expression profiling on laser captured frozen tumour samples from eight SCLC and eight LCNEC tumours identified a total of 888 differentially expressed genes (DEGs), 23 of which were validated by qRT-PCR. Antibodies to four selected gene products were then evaluated as immunohistochemical markers on a cohort of 173 formalin-fixed paraffin-embedded (FFPE) SCLC/LCNEC tumour samples, including 26 indeterminate tumours without a consensus diagnosis. Three markers, CDX2, VIL1 and BAI3, gave significantly different results in the two tumour types (P < 0.0001): CDX2 and VIL1 in combination (either marker positive) showed sensitivity and specificity of 81% for LCNEC while BAI3 showed 89% sensitivity and 75% specificity for SCLC. Of the 26 indeterminate tumours 15 (58%) showed an immunophenotype suggesting either SCLC or LCNEC, eight (31%) showed staining of both tumour types, and three (11%) were negative for all markers. CONCLUSION: A panel of three markers, BAI3, CDX2 and VIL1, is a useful adjunct in the diagnosis of these tumour types.