Strategies for clinical implementation of TNM-Immunoscore in resected nonsmall-cell lung cancer.

Immunoscore is a prognostic tool defined to quantify in situ immune cell infiltrates and appears highly promising as a supplement to the tumor-node-metastasis (TNM) classification of various tumors. In colorectal cancer, an international task force has initiated prospective multicenter studies aiming to implement TNM-Immunoscore (TNM-I) in a routine clinical setting. In breast cancer, recommendations for the evaluation of tumor-infiltrating lymphocytes (TILs) have been proposed by an international working group. Regardless of promising results, there are potential obstacles related to implementing TNM-I into the clinic. Diverse methods may be needed for different malignancies and even within each cancer entity. Nevertheless, a uniform approach across malignancies would be advantageous. In nonsmall-cell lung cancer (NSCLC), there are several previous reports indicating an apparent prognostic importance of TILs, but studies on TILs in a TNM-I setting are sparse and no general recommendations are made. However, recently published data is promising, evoking a realistic hope of a clinical useful NSCLC TNM-I. This review will focus on the TNM-I potential in NSCLC and propose strategies for clinical implementation of a TNM-I in resected NSCLC.

Vascular phenotypes in primary non-small cell lung carcinomas and matched brain metastases.

BACKGROUND: Anti-angiogenic therapy with bevacizumab (an anti-vascular endothelial growth factor (VEGF) antibody) predominantly targets immature blood vessels. Bevacizumab has shown a survival benefit in non-small cell lung carcinoma (NSCLC) and has recently been demonstrated to be safe in patients with brain metastases. However, it is not known whether bevacizumab is effective against brain metastases or whether metastases are representative of their primary in terms of VEGF expression, hypoxia, proliferation and vascular phenotype. The aim of this study was to evaluate these factors in a series of matched primary NSCLCs and brain metastases. METHODS AND RESULTS: Immunohistochemistry showed strong correlation of carbonic anhydrase 9 expression (a marker of hypoxia) in primary and secondary cancers (P=0.0002). However, the proliferation index, VEGF expression, microvessel density and the proportion of mature vessels were discordant between primary and secondary cancers. The mean proportion of mature vessels was 63.2% higher in the brain metastases than the primary tumours (P=0.004). Moreover, the vascular pattern of the primary tumour was not representative of the metastasis. CONCLUSIONS: Brain metastases have a significantly higher proportion of mature vasculature, suggesting that they may be refractory to anti-VEGF therapy. These findings may have implications for clinical trials and biomarker studies evaluating anti-angiogenic agents in brain metastases.

Expression of delta-like ligand 4 (Dll4) and markers of hypoxia in colon cancer.

BACKGROUND: Delta-like ligand 4 (Dll4) is a Notch ligand that is upregulated by hypoxia and vascular endothelial growth factor-A (VEGF-A) and is reported to have a role in tumor angiogenesis. Evidence from xenograft studies suggests that inhibiting Dll4-Notch signalling may overcome resistance to anti-VEGF therapy. The aim of this study was to characterise the expression of Dll4 in colon cancer and to assess whether it is associated with markers of hypoxia and prognosis. METHOD: In all, 177 colon cancers were represented in tissue microarrays. Immunohistochemistry was performed using validated antibodies against Dll4, VEGF, hypoxia-inducible factor (HIF)-1alpha, HIF-2alpha, prolyl hydroxylase (PHD)1, PHD2, PHD3 and carbonic anhydrase 9 (CA9). RESULTS: The expression of Dll4 was observed preferentially in the endothelium of 71% (125 out of 175) of colon cancers, but not in the endothelium adjacent to normal mucosa (none out of 107, P<0.0001). The expression of VEGF was significantly associated with HIF-2alpha (P<0.0001) and Dll4 (P=0.010). Only HIF-2alpha had a significant multivariate prognostic effect (hazard ratio 1.61, 95% confidence interval 1.01-2.57). Delta-like ligand 4 was also expressed by neoplastic cells, particularly neoplastic goblet cells. CONCLUSION: Endothelial expression of Dll4 is not a prognostic factor, but is significantly associated with VEGF. Assessing endothelial Dll4 expression may be critical in predicting response to anti-VEGF therapies.

The key hypoxia regulated gene CAIX is upregulated in basal-like breast tumours and is associated with resistance to chemotherapy.

Basal-like tumours account for 15% of invasive breast carcinomas and are associated with a poorer prognosis and resistance to therapy. We hypothesised that this aggressive phenotype is because of an intrinsically elevated hypoxic response. Microarrayed tumours from 188 patients were stained for hypoxia-inducible factor (HIF)-1alpha, prolyl hydroxylase (PHD)1, PHD2, PHD3 and factor inhibiting HIF (FIH)-1, and carbonic anhydrase (CA) IX stained in 456 breast tumours. Tumour subtypes were correlated with standard clincopathological parameters as well as hypoxic markers. Out of 456 tumours 62 (14%) tumours were basal-like. These tumours were positively correlated with high tumour grade (P<0.001) and were associated with a significantly worse disease-free survival compared with luminal tumours (P<0.001). Fifty percent of basal-like tumours expressed HIF-1alpha, and more than half expressed at least one of the PHD enzymes and FIH-1. Basal-like tumours were nine times more likely to be associated with CAIX expression (P<0.001) in a multivariate analysis. Carbonic anhydrase IX expression was positively correlated with tumour size (P=0.005), tumour grade (P<0.001) and oestrogen receptor (ER) negativity (P<0.001). Patients with any CAIX-positive breast tumour phenotype and in the basal tumour group had a significantly worse prognosis than CAIX-negative tumours when treated with chemotherapy (P<0.001 and P=0.03, respectively). The association between basal phenotype and CAIX suggests that the more aggressive behaviour of these tumours is partly due to an enhanced hypoxic response. Further, the association with chemoresistance in CAIX-positive breast tumours and basal-like tumours in particular raises the possibility that targeted therapy against HIF pathway or downstream genes such as CAs may be an approach to investigate for these patients.

Standard operating procedure for the collection of fresh frozen tissue samples.

Studies using fresh-frozen tissue samples originating from different centres, as is often the case in EORTC related translational research, can show conflicting research results due to heterogeneity in the quality of samples and associated data from each centre. The development of infrastructure for the European Human Frozen Tumour Tissue Bank (TuBaFrost) anticipated this problem and Standard Operating Procedures (SOPs) have been developed to ensure samples collected are of consistent high quality and variation in research results is minimised. The SOPs drew on the best practice standard workflows and operating procedures employed by members of the TuBaFrost Consortium and key tissue bank initiatives worldwide. It was essential to provide workable solutions that reflect the variety in infrastructure and resources at the potential collecting centres and also the fact that it is not necessary to standardise every step of the collection and storage process in order to collect high quality tissue. Hence, the TuBaFrost SOPs detail the compulsory measures that must be implemented in order to become a TuBaFrost collecting centre and also make advisory recommendations regarding the less critical factors. Accordingly, the TuBaFrost SOPs are very flexible and to illustrate this the complete SOP for collecting, freezing and storing tissue at the Erasmus MC Tissue Bank is included. These TuBaFrost SOPs could equally be applicable to centres collecting samples for EORTC related translational research studies in order to standardise sample quality and produce reliable and reproducible research results.

TuBaFrost 6: virtual microscopy in virtual tumour banking.

Many systems have already been designed and successfully used for sharing histology images over large distances, without transfer of the original glass slides. Rapid evolution was seen when digital images could be transferred over the Internet. Nowadays, sophisticated Virtual Microscope systems can be acquired, with the capability to quickly scan large batches of glass slides at high magnification and compress and store the large images on disc, which subsequently can be consulted through the Internet. The images are stored on an image server, which can give simple, easy to transfer pictures to the user specifying a certain magnification on any position in the scan. This offers new opportunities in histology review, overcoming the necessity of the dynamic telepathology systems to have compatible software systems and microscopes and in addition, an adequate connection of sufficient bandwidth. Consulting the images now only requires an Internet connection and a computer with a high quality monitor. A system of complete pathology review supporting bio-repositories is described, based on the implementation of this technique in the European Human Frozen Tumor Tissue Bank (TuBaFrost).

TuBaFrost 1: Uniting local frozen tumour banks into a European network: an overview.

TuBaFrost is the consortium responsible for the creation of a virtual European human frozen tumour tissue bank: a collection of high quality frozen residual, accurately classified tumour tissue samples, which are stored in European cancer centres and universities. This virtual tissue bank, searchable on the internet, has rules for access and use, and a code of conduct to comply with the various legal and ethical regulations in European countries. The easy accessibility and the European scale of the bank will result in the availability of a large number of samples even of rarer tumour types. Standardisation of collection, storage and quality control throughout the network is achieved minimising inter-institutional variability. A website providing access to upload, search and request samples is a key tool of the tissue bank. The search engine makes use of virtual microscopy. An overview of the development of the European virtual frozen tissue bank infrastructure is described in this paper. The various key aspects are described in more detail in a series of articles to appear in this Journal.

TuBaFrost 2: Standardising tissue collection and quality control procedures for a European virtual frozen tissue bank network.

Tumour Bank Networking presents a great challenge for oncological research as in order to carry out large-scale, multi-centre studies with minimal intrinsic bias, each tumour bank in the network must have some fundamental similarities and be using the same standardised and validated procedures. The European Human Frozen Tumour Tissue Bank (TuBaFrost) has responded to this need by the promotion of an integrated platform of tumour banks in Europe. The operational framework for TuBaFrost has drawn upon the best practice of standard workflows and operating procedures employed by members of the TuBaFrost project and key initiatives worldwide.

TuBaFrost 4: access rules and incentives for a European tumour bank.

When designing infrastructure for a networked virtual tumour bank (samples remain at the collector institutes and sample data are collected in a searchable central database), it is apparent that this can only function properly after developing an adequate set of rules for use and access. These rules must include sufficient incentives for the tissue sample collectors to remain active within the network and maintain sufficient sample levels in the local bank. These requirements resulted in a key TuBaFrost rule, stating that the custodianship of the samples remains under the authority of the local collector. As a consequence, the samples and the decision to issue the samples to a requestor are not transferred to a large organisation but instead remain with the collector, thus allowing autonomous negotiation between collector and requestor, potential co-authorship in publications or compensation for collection and processing costs. Furthermore, it realises a streamlined cost effective network, ensuring tissue visibility and accessibility thereby improving the availability of large amounts of samples of highly specific or rare tumour types as well as providing contact opportunities for collaboration between scientists with cutting edge technology and tissue collectors. With this general purpose in mind, the rules and responsibilities for collectors, requestors and central office were generated.

TuBaFrost 5: multifunctional central database application for a European tumor bank.

Developing a tissue bank database has become more than just logically arranging data in tables combined with a search engine. Current demand for high quality samples and data, and the ever-changing legal and ethical regulations mean that the application must reflect TuBaFrost rules and protocols for the collection, exchange and use of tissue. To ensure continuation and extension of the TuBaFrost European tissue bank, the custodianship of the samples, and hence the decision over whether to issue samples to requestors, remains with the local collecting centre. The database application described in this article has been developed to facilitate this open structure virtual tissue bank model serving a large group. It encompasses many key tasks, without the requirement for personnel, hence minimising operational costs. The Internet-accessible database application enables search, selection and request submission for requestors, whereas collectors can upload and edit their collection. Communication between requestor and involved collectors is started with automatically generated e-mails.

TuBaFrost: European virtual tumor tissue banking.

TuBaFrost is a consortium responsible for the task to create a virtual European human frozen tumor tissue bank, composed of high quality frozen tumor tissue collections with corresponding accurate diagnosis stored in European cancer centers and universities, searchable on the Internet, providing rules for access and use and a code of conduct to comply with the various legal and ethical regulations in European countries. Such infrastructure would enlarge tissue availability and accessibility in large amounts of specified or even rare tumor samples. Design of an infrastructure for European residual tissue banking with the described characteristics, clear focus points emerge that can be broken down in dedicated subjects: (1) standardization and quality assurance (QA) to avoid inter-institute quality variation; (2) law and ethics enabling exchange of tissue samples possible between institutes in the different European countries, where law and ethics are characterized by a strong variability; (3) rules for access, with sufficient incentives for collectors; (4) central database application containing innovations on search and selection procedures; (5) support when needed with histology images; and (6) Internet access to search and upload, with in addition a solid website giving proper information on the procedures, intentions and activities not only to the scientific community, but also to the general public. One consortium decision, part of the incentives for collectors, had major impact on the infrastructure; custodianship over the tissues as well as the tissues stay with the collector institute. Resulting in specimens that are not given to an organization, taking decisions on participation of requests, but instead the local collected tissues stay very easy to access by the collector and allows autonomous negotiation between collector and requestor on cooperation, coauthorship in publication or compensation in costs. Thereby, improving availability of large amounts of high quality samples of a highly specified or rare tumor types and contact opportunities for cooperation with other institutes.

Virtual microscopy in virtual tumor banking.

Many systems have already been designed and successfully used for sharing histology images over large distances, without transfer of the original glass slides. Rapid evolution was seen when digital images could be transferred over the Internet. Nowadays, sophisticated virtual microscope systems can be acquired, with the capability to quickly scan large batches of glass slides at high magnification and compress and store the large images on disc, which subsequently can be consulted through the Internet. The images are stored on an image server, which can give simple, easy to transfer pictures to the user specifying a certain magnification on any position in the scan. This offers new opportunities in histology review, overcoming the necessity of the dynamic telepathology systems to have compatible software systems and microscopes and in addition, an adequate connection of sufficient bandwidth. Consulting the images now only requires an Internet connection and a computer with a high quality monitor. A system of complete pathology review supporting biorepositories is described, based on the implementation of this technique in the European Human Frozen Tumor Tissue Bank (TuBaFrost).

The distribution of the deleted in colon cancer (DCC) protein in human tissues.

A gene called deleted in colon cancer (DCC) has been identified on a region of chromosome 18, which is deleted in 70% of colorectal cancers. The DCC gene encodes a protein belonging to the immunoglobulin superfamily with similarity to the N-CAM transmembrane proteins and is a putative tumor-suppressor gene. Alternative splicing of transcripts of transmembrane proteins, including N-CAM, is known to occur, resulting in different isoforms of the protein. Using five antibodies against the DCC gene product (three monoclonal antibodies raised in our laboratory, one commercially available antibody, and a rabbit polyclonal antibody), we have demonstrated by immunostaining a DCC protein isoform in reticuloendothelial cells in human thymus, tonsil, and lymph node. This can be distinguished from another isoform described in normal colonic epithelium, because this latter is not demonstrable with the antibodies we have used. It could not be detected in normal colonic epithelium, polyps or colorectal carcinomas. This restrictive distribution suggests that not all DCC gene products are important in colonic cancer.

Loss of FHIT function in lung cancer and preinvasive bronchial lesions.

We previously cloned and characterized the tumor suppressor gene FHIT (fragile histidine triad) at chromosome 3p14.2 and found that this gene is altered by deletions in human tumors, including lung cancer. To assess the frequency and specificity of inactivation and its relevance in a clinical setting, we have produced antibodies against the Fhit protein and studied its expression in a series of non-small cell lung cancers and normal bronchial mucosa and a spectrum of preinvasive lesions by immunohistochemistry. The data indicate that the loss of Fhit protein is the most frequent alteration in non-small cell lung cancer (73%) and precancerous lesions (93%), is significantly higher in the tumors of smokers (75%) than in those of nonsmokers (39%; P < 0.0005), and is an independent and more frequent event than p53 overexpression in tumors and precancerous lesions (73 versus 46%). The percentage of cases lacking Fhit expression was higher in the squamous type compared to adenocarcinoma (87 versus 57%; P < 0.00001), whereas other histotypes (large cell, mucoepidermal) showed an intermediate value (69%). Loss of Fhit expression in a very high percentage of primary lung carcinomas and precancerous lesions supports the notion that FHIT alterations play an important role in the growth control of bronchial cells. FHIT inactivation is particularly important in squamous cell carcinomas that are often associated with precursor dysplastic lesions. The overall high frequency and precocity of Fhit loss in lung carcinogenesis and the development of the presently described immunohistochemical approach suggest a potential use of this gene in the early detection of lung cancer and in chemopreventive studies as an intermediate biomarker.

Absence of Fhit protein in primary lung tumors and cell lines with FHIT gene abnormalities.

Genomic alterations and abnormal expression of the FHIT gene at 3p14.2 have been observed in cell lines and primary tumors of the lung. To correlate FHIT locus DNA and RNA lesions with effects on Fhit protein expression, we have analyzed 11 lung cancer cell lines, 15 small cell lung carcinomas, and 38 pairs of non-small cell primary tumors and bronchial mucosa specimens by molecular genetic and immunocytochemical methods. Using specific antibodies against the Fhit protein, we observed concordance between RNA abnormalities and lack of Fhit protein expression in lung tumors and cell lines. In addition, absence of Fhit protein in some precancerous dysplastic lesions suggested that FHIT inactivation may occur at an early phase of lung carcinogenesis.

Immunocytochemical markers in stage I lung cancer: relevance to prognosis.

PURPOSE: This study investigated the frequency of the expression and prognostic significance of a panel of immunocytochemical markers in resected non-small-cell lung cancer (NSCLC). PATIENTS AND METHODS: A total of 515 cases of pathologic stage I NSCLC were analyzed. The median follow-up time of surviving patients was 102 months. The following immunocytochemical markers were tested: blood group A and precursors of blood antigens; laminin receptor; c-erbB1/epidermal growth factor receptor (EGFR) and c-erbB2/Neu; BCl2; p53; and angiogenesis. Kaplan-Meier estimates of survival and time to recurrence were calculated for clinical variables and biologic markers using the Cox model for multivariate analysis. RESULTS: The pathologic tumor extension (pT) represented the most powerful prognostic factor for survival (P = .0008) and time to recurrence (P = .0007). None of the immunocytochemical markers emerged as an independent predictive factor for survival. Bcl2-positive tumors showed a better time to recurrence (P = .03), but the difference lost statistical significance in the multivariate analysis. Of interest, in the group of 137 patients classified as pT1N0, both EGFR expression and nonangiogenic type of vascular pattern were associated with a poorer survival (P = .02). However, data derived from subset analysis must be interpreted cautiously. CONCLUSION: Our findings do not support a relevant prognostic role of immunocytochemical markers in NSCLC. The evidence is not sufficient to alter clinical practice or even to restrict clinical trials of adjuvant treatments to predefined biologic subsets of patients.

Vascular endothelial growth factor, wild-type p53, and angiogenesis in early operable non-small cell lung cancer.

Vascular endothelial growth factor (VEGF) is a cytokine that is involved in tumor angiogenesis. Wild-type p53 (wt-p53) protein has been shown in cell lines to suppress angiogenesis through thrombospondin regulation. In this study, we immunohistochemically examined the expression of VEGF, nuclear and wild-type cytoplasmic p53, bcl-2, epidermal growth factor receptor, and c-erbB-2 oncoprotein; vascular grade; proliferation index; and extent of necrosis in non-small cell lung cancer (NSCLC). We analyzed 120 cases of early-stage NSCLCs (81 squamous cell carcinomas and 39 adenocarcinomas) treated with surgery alone (median follow-up, 63 months; range, 45-74 months). VEGF expression showed a positive association with high vascular grade (microvessel score of >75 per x250 field; P = 0.008), although about half of the LVG cases also expressed VEGF. None of the p53 antibodies examined correlated with angiogenesis. However, wt-p53 expression was inversely associated with VEGF expression, suggesting that wt-p53 is involved in the suppression of the VEGF gene. Combined analysis of VEGF, wt-p53, and microvessel counting showed that, although wt-p53 loss associates with VEGF switch-on, p53 protein may not be involved in the regulation of the angiogenic events downstream of VEGF expression. Moreover, no significant association of bcl-2 and c-erbB-2 oncoprotein expression with VEGF expression was observed. T/N stage, grade, Ki67 proliferation index, and extent of necrosis were not correlated with VEGF expression. Survival analysis showed that VEGF correlated with poor survival (P = 0.04) and was significant in node-negative cases (P = 0.03). We conclude that VEGF is an important angiogenic factor in NSCLC, its expression being dependent on wt-p53 loss.

Predictive and prognostic markers in a series of patients with head and neck squamous cell invasive carcinoma treated with concurrent chemoradiation therapy.

It has been proposed that diverse anticancer drugs and radiation therapy may induce a mode of cell death with the characteristics of apoptosis. Since apoptosis is under the control of several oncogenes, we analyzed the expression of the protein encoded by the proto-oncogenes bcl-2 and p53. Furthermore, we studied cell proliferation [using PC-10 mAb to proliferating cell nuclear antigen (PCNA)] and vascularization [using the CD-31 mAb and by counting intratumoral microvessel density (IMD)] using immunocytochemistry. A series of 73 patients with clinical stage II-IV squamous cell invasive carcinoma of the head and neck (H&N) were treated with concurrent chemoradiation therapy (cisplatin, 80 mg/m2, versus carboplatin, 375 mg/m2, three times every 3 weeks and a total dose of radiation therapy of 64 Gy in 6-8 weeks). We correlated the expression of these markers, determined prior to treatment, with response to the therapy and prognosis. Bcl-2 protein was expressed in 37.4% of the carcinomas (25/67 evaluable), and it was not significantly associated with any other feature studied. Forty (56. 4%) of the 71 carcinomas evaluable for p53 were p53 positive; the median IMD was 38 microvessels/field at the hot spot (range, 18-80), and the median percentage of nuclei labeled by the PC-10 mAb was 50% (range, 0-95%). In the univariate analysis, regional lymph node negativity (P = 0.016), good performance status (PS) (PS >/= 90; P = 0.044), bcl-2 positivity (P = 0.070), and low vascularization (P = 0. 085) were significantly associated with a higher probability of complete remission. In the multivariate analysis (final model), only IMD (continuous variable; P = 0.045) and PS (P = 0.017) retained significance. As far as prognosis is concerned, in the univariate analysis, patients with tumors with low histological grading (grades 1-2; P = 0.006), p53 negative (P = 0.09), bcl-2 positive (P = 0.08), and high PCNA labeling (P = 0.06) had a significantly better disease-free survival. In the multivariate analysis, only grading (P = 0.003) and p53 (P = 0.04) retained significance for disease-free survival. For overall survival, in the univariate analysis, the following markers were significantly prognostic when only deaths due to progression are considered: response to therapy (P = 0.00001), PS (P = 0.04), nodal status (P = 0.028), PCNA (P = 0.04), p53 (P = 0. 08), and grading (P = 0.01). In the multivariate analysis, only patients who achieved complete response (P = 0.00002), high PCNA values (P = 0.002), and low histological grading (P = 0.01) retained a statistically significant probability of better overall survival. Our results suggest that in this series of H&N cancer patients the markers capable of predicting response to therapy are distinct from those associated with prognosis, once the remission has been achieved. This information is potentially useful to the clinician for developing a more rational therapeutic approach for H&N cancer patients eligible for concurrent chemoradiation therapy.

Schnyder corneal crystalline dystrophy: description of a new family with evidence of abnormal lipid storage in skin fibroblasts.

Schnyder corneal crystalline dystrophy (SCCD) comprises corneal opacities often associated with precocious arcus senilis and genua valga. The metabolic defect seems to be related to abnormal lipid storage in the central part of the cornea, especially the anterior stroma, consisting mainly of nonesterified cholesterol. Plasma lipid levels are not always increased suggesting that the disease may be due to abnormal lipid metabolism limited to the cornea. We observed a family with typical SCCD, in 1 case associated with mental retardation and mild cerebellar hypoplasia. Results of serum lipid analysis of all patients were normal. Ultrastructural study of a skin biopsy specimen and fibroblast pellet showed membrane-bound spherical vacuoles containing lipid material. Cultured fibroblasts stained by filipin, a fluorescent probe that specifically binds unesterified cholesterol, showed abnormal cytoplasmic fluorescent material, suggesting abnormal cholesterol metabolism. The presence of neurological impairment, associated with SCCD in 1 of our cases, may be regarded as coincidental. Evidence of storage lipids in skin and cultured fibroblasts suggests that the disorder of intracellular cholesterol metabolism is not limited to the cornea and that skin biopsy may be a useful method to confirm the diagnosis.

Expression of the bcl-2 oncogene product and chromosomal translocation t(14;18) in Hodgkin's disease.

The B-cell lymphoma/leukemia oncogene bcl-2 takes part in crucial regulatory events in B-cell maturation and differentiation. The reciprocal chromosomal translocation t(14;18), leading to overexpression of this oncogene, can be found in the majority of follicular lymphomas and much less frequently in B-cell leukemias and diffuse lymphomas. We have studied the expression of this protein in different subtypes of Hodgkin's disease using monoclonal antibodies directed against a formalin-resistant epitope of the bcl-2 protein and also have investigated these cases by polymerase chain reaction for evidence of the t(14;18) translocation. We were particularly interested to determine whether nodular paragranuloma (lymphocyte-predominant, nodular), which differs from other subtypes of Hodgkin's disease by virtue of the B-cell nature of its malignant cell population, is characterized by expression of the bcl-2 protein. Our data indicate that only a small number of nodular paragranulomas express the bcl-2 protein and that the expression is not specific for this type of Hodgkin's disease. In a smaller number of cases this expression of bcl-2 could be explained by the presence of the translocation t(14;18).