Selective thrombosis of tumor blood vessels in mammary adenocarcinoma implants in rats.

Adenocarcinomas in rats and humans frequently contain perivascular, degranulating mast cells that release heparin. Protamine is a low-molecular weight, cationic polypeptide that binds avidly to heparin and neutralizes its anticoagulant properties. We hypothesized that mast-cell heparin functions as a localized anticoagulant that modulates hemostasis and blood perfusion in tumors. Consequently, systemically administered protamine should be able to neutralize the endogenous heparin within tumors, thereby inducing selective thrombosis of blood vessels within tumors. Here we demonstrate with magnetic resonance imaging (MRI) that an intravenous dose of protamine labeled with gadolinium accumulated within the parenchyma of subcutaneous implants of a mammary adenocarcinoma in Fischer 344 rats. Moreover, we show with dynamic contrast enhanced MRI that sequential intravenous doses of protamine in 12 tumor-bearing rats resulted in significantly decreased signal enhancement kinetics (blood perfusion) of the tumor. This functional impairment of MRI signal enhancement was accompanied by histological evidence of thrombosis in the blood vessels within the tumor. There was no histological evidence of thrombosis within normal liver, kidney, lung, spleen, or adjacent muscle of tumor-bearing animals that received protamine treatment or in the tumors of animals that had not been pretreated with protamine. On the basis of these results, we conclude that protamine accumulates within adenocarcinoma implants and induces selective thrombosis of blood vessels within the tumor, probably by neutralizing the endogenous heparin within tumors.

Eosinophils are a major source of nitric oxide-derived oxidants in severe asthma: characterization of pathways available to eosinophils for generating reactive nitrogen species.

Eosinophil recruitment and enhanced production of NO are characteristic features of asthma. However, neither the ability of eosinophils to generate NO-derived oxidants nor their role in nitration of targets during asthma is established. Using gas chromatography-mass spectrometry we demonstrate a 10-fold increase in 3-nitrotyrosine (NO(2)Y) content, a global marker of protein modification by reactive nitrogen species, in proteins recovered from bronchoalveolar lavage of severe asthmatic patients (480 +/- 198 micromol/mol tyrosine; n = 11) compared with nonasthmatic subjects (52.5 +/- 40.7 micromol/mol tyrosine; n = 12). Parallel gas chromatography-mass spectrometry analyses of bronchoalveolar lavage proteins for 3-bromotyrosine (BrY) and 3-chlorotyrosine (ClY), selective markers of eosinophil peroxidase (EPO)- and myeloperoxidase-catalyzed oxidation, respectively, demonstrated a dramatic preferential formation of BrY in asthmatic (1093 +/- 457 micromol BrY/mol tyrosine; 161 +/- 88 micromol ClY/mol tyrosine; n = 11 each) compared with nonasthmatic subjects (13 +/- 14.5 micromol BrY/mol tyrosine; 65 +/- 69 micromol ClY/mol tyrosine; n = 12 each). Bronchial tissue from individuals who died of asthma demonstrated the most intense anti-NO(2)Y immunostaining in epitopes that colocalized with eosinophils. Although eosinophils from normal subjects failed to generate detectable levels of NO, NO(2-), NO(3-), or NO(2)Y, tyrosine nitration was promoted by eosinophils activated either in the presence of physiological levels of NO(2-) or an exogenous NO source. At low, but not high (e.g., >2 microM/min), rates of NO flux, EPO inhibitors and catalase markedly attenuated aromatic nitration. These results identify eosinophils as a major source of oxidants during asthma. They also demonstrate that eosinophils use distinct mechanisms for generating NO-derived oxidants and identify EPO as an enzymatic source of nitrating intermediates in eosinophils.

Eosinophils generate brominating oxidants in allergen-induced asthma.

Eosinophils promote tissue injury and contribute to the pathogenesis of allergen-triggered diseases like asthma, but the chemical basis of damage to eosinophil targets is unknown. We now demonstrate that eosinophil activation in vivo results in oxidative damage of proteins through bromination of tyrosine residues, a heretofore unrecognized pathway for covalent modification of biologic targets in human tissues. Mass spectrometric studies demonstrated that 3-bromotyrosine serves as a specific "molecular fingerprint" for proteins modified through the eosinophil peroxidase-H(2)O(2) system in the presence of plasma levels of halides. We applied a localized allergen challenge to model the effects of eosinophils and brominating oxidants in human lung injury. Endobronchial biopsy specimens from allergen-challenged lung segments of asthmatic, but not healthy control, subjects demonstrated significant enrichments in eosinophils and eosinophil peroxidase. Baseline levels of 3-bromotyrosine in bronchoalveolar lavage (BAL) proteins from mildly allergic asthmatic individuals were modestly but not statistically significantly elevated over those in control subjects. After exposure to segmental allergen challenge, lung segments of asthmatics, but not healthy control subjects, exhibited a >10-fold increase in BAL 3-bromotyrosine content, but only two- to threefold increases in 3-chlorotyrosine, a specific oxidation product formed by neutrophil- and monocyte-derived myeloperoxidase. These results identify reactive brominating species produced by eosinophils as a distinct class of oxidants formed in vivo. They also reveal eosinophil peroxidase as a potential therapeutic target for allergen-triggered inflammatory tissue injury in humans.

An immunofluorescent assay for acute promyelocytic leukemia cells.

Sequential treatment with all-trans retinoic acid followed by chemotherapy significantly improves the long-term survival of patients who have acute promyelocytic leukemia (APL). Consequently, a simple and accurate test is needed to establish the diagnosis of APL and to identify those patients having a relapse of the disease. We describe an accurate, 2-hour indirect immunofluorescent assay for identifying APL cells in bone marrow specimens. The assay uses the PML (PG-M3) murine monoclonal antibody that is directed against the amino-terminal portion of the PML gene product. We observed a distinctive, finely speckled pattern of fluorescence in the NB4 cell line (a positive control), as well as in 15 clinical specimens that were confirmed to have APL by cytogenetic, cytochemical, and immunophenotypic studies, including four cases of microgranular variant of APL. By contrast, a coarse globular pattern of fluorescence was observed in 53 other clinical specimens that did not contain APL. When we performed dilution studies using artificial mixtures of APL cells with normal bone marrow cells, we detected as few as 5% APL cells in the mixture. Finally, there was complete concordance between the immunofluorescent assay and a polymerase chain reaction-based assay for the PML-retinoic acid receptor alpha chimeric gene in 12 other clinical specimens. We conclude that the immunofluorescent assay for PML protein is a rapid, sensitive, and accurate method for determining the presence of APL cells in clinical specimens. This assay therefore should be considered as a cost-effective alternative to other diagnostic tests, such as karyotyping or polymerase chain reaction, for the diagnostic evaluation of APL.

Lack of prognostic value of cathepsin D levels for predicting short term outcomes of breast cancer patients.

The value of cathepsin D determinations done on tumor cytosols in evaluating the prognosis of breast cancer patients has been debated in the literature. Our previous work suggested that cathepsin D determinations were not of prognostic value, but in that study we used immunoblotting and immunohistochemical methods rather than the more widely used double antibody immunoradiometric (IRMA) assay for measuring cathepsin levels. Here we report our results determining cathepsin D using components of a commercially available IRMA system on a large patient sample (n = 1984). Reagents from a commercially available IRMA kit were used to analyze cathepsin D levels in the cytosols of 1984 patients with breast cancer. All patients had invasive breast cancer with known tumor size and with some axillary nodes pathologically examined. Only patients with T1 and T2 tumor sizes were included. Median follow-up was 37 months. The hypothesis that high cathepsin D levels correlated with poorer outcome (poorer DFS or OS) was not confirmed, either in all patients, or in node-positive or node-negative subsets. Only in patients treated with adjuvant therapy were higher cathepsin D levels correlated with negative outcome (worsened OS, but not DFS), although given the large number of subsets analyzed this correlation may be spurious. Multivariate analyses using interaction terms did not support the concept that high cathepsin D levels correlate with resistance to adjuvant therapy. In this study evaluating the value of cathepsin D using components from a kit widely used for measuring cathepsin D levels, we conclude that cathepsin D is of doubtful value in predicting risk of early relapse or death for patients with newly diagnosed invasive breast cancer.

Association between flow cytometric S-phase fraction and apoptotic rate in breast cancer.

Tumor heterogeneity may adversely affect the flow cytometric measurement of S-phase fraction (SPF) in breast cancer specimens, and 10-20% of breast cancer specimens are not evaluable by flow cytometry due to technical factors such as debris, high coefficients of variation, poor specimen quality, or small sample size. Therefore, we performed this study on 207 specimens of breast cancer in order to determine if the apoptotic rate (AR) could serve as a useful adjunct to flow cytometric SPF measurements in breast cancers. The average AR in each specimen was determined by microscopic examination of tumor tissue that was specifically stained for apoptotic bodies by a commercially available TUNEL (Tdt-mediated dUTP digoxigenin nick end labelling) assay kit. The mean AR (4.5 +/- 3.0, n = 37) in the high SPF (> 10%) group was significantly (P < 0.01) higher than the mean AR (1.3 +/- 1.2, n = 72) in the low SPF (< 6%) group. Although the distributions of AR values in the two groups had substantial overlap, AR values greater than 5.5 per high power field (h.p.f.) were not observed in the low SPF cases but were present in 13 out of 37 cases with a high SPF. Simple linear regression analyses relating SPF to the mean AR in 57 DNA diploid cases and 41 DNA aneuploid cases yielded a minimal correlation (r2 = 0.21) between the two parameters only in the DNA aneuploid group. We conclude that an elevated AR has an association with high SPF in breast cancers, but the association is too weak to permit the general use of AR as a predictor of SPF. Our study also identified a subset of breast cancers with both a high SPF (> 10%) and a high AR (> 5.5/h.p.f.) that may warrant further investigation to determine its clinical significance.

Pharmacokinetic changes induced by vasomodulators in kidneys, livers, muscles, and implanted tumors in rats as measured by dynamic Gd-DTPA-enhanced MRI.

The effects of three physiologically different vasomodulators, angiotensin II (a vasoconstrictor), hydralazine (a vasodilator), and histamine (a permeability modulator), on the pharmaco-kinetics of entry of small molecules (measured by Gd-DTPA concentration) into normal and abnormal tissue were studied in rats implanted with R3230 AC tumors. Sequential dynamic Gd-DTPA-enhanced MRI studies, one before and one after vasomodulator administration, were performed, and the signal intensities of various tissues analyzed. Angiotensin II (6 micrograms/kg) reduced blood flow in tumors, but increased it in muscles. Hydralazine (5 mg/kg) reduced blood flow in tumors, kidneys, and livers, and slowed Gd-DTPA clearance from tumors, livers, and muscles. Histamine (25 micrograms/kg) increased renal blood flow, hastening Gd-DTPA clearance causing reduced measurable blood flow in tumors and muscles. By simultaneously monitoring the effects in various tissues, the pharmacokinetic effect of each drug in the entire body could be obtained.

Occult deposition of eosinophil peroxidase in a subset of human breast carcinomas.

Degranulation of eosinophils has been observed in a variety of human tumors and in other diseases but has not been previously described in breast cancer. To determine whether eosinophil degranulation also occurs in breast carcinomas, we performed immunohistological studies on cryostat sections obtained from 26 breast cancer biopsies and from 2 benign breast tissues using a monoclonal antibody specific for human eosinophil peroxidase (EPO). For control purposes, the tissues were also immunostained with a mouse IgG1 negative control antibody and with monoclonal mouse anti-human myeloperoxidase. Of the 26 breast cancer specimens, 14 (53%) had extensive, unsuspected deposition of EPO that was located primarily in the connective tissue stroma around and within the tumor. Only 3 of the breast cancer cases had no immunohistochemical evidence of EPO. Thus, 23 of 26 cases of breast cancer (88%) had EPO deposits detectable within or around the tumor. By contrast, none of the benign breast tissues had similar deposits of EPO, and substantial extracellular myeloperoxidase deposition was detectable in only 3 cases of breast cancer. From these studies we conclude that there is eosinophil degranulation and extensive occult deposition of EPO in a major subset of human breast cancers.

Eradication of interleukin 5-transfected J558L plasmacytomas in mice by hydrogen peroxide-generating Stealth liposomes.

Certain human tumors are extensively infiltrated by eosinophils and contain extracellular deposits of eosinophil peroxidase, which uses hydrogen peroxide as a substrate to produce highly toxic hypohalous acids. We hypothesized that J558L HI, an interleukin 5-transfected murine plasmacytoma that is infiltrated by numerous degranulating eosinophils, would be especially sensitive to killing by hydrogen peroxide generated by glucose oxidase (beta-D-glucose:oxygen-oxido reductase; EC 1.13.4). Here we report that 4 i.v. injections of 0.5 ml of hydrogen peroxide-generating, anionic Stealth liposomes containing 50 micrograms of glucose oxidase eradicated s.c. implants of 10(6) J558L HI plasmacytoma cells in 6 of 13 mice. By contrast, the J558L HI tumors grew rapidly in 13 of 13 untreated mice and in 10 of 10 mice treated with daily i.v. injections of 50 micrograms of unencapsulated (free) glucose oxidase (P = 0.002 by log-rank test of survival curves constructed using the Kaplan-Meier method). Antisense transfected J558L tumors that did not contain eosinophils were not eradicated by the peroxide-generating liposomes in any of the 10 mice that were tested. Treatment with the liposomes was well tolerated for the first three doses (given on days 3, 4, and 5 after tumor inoculation). The fourth dose given on day 10 produced significant allergic toxicity and was, therefore, omitted in a second trial with only minimal reduction in the therapeutic response. We conclude that peroxide-generating, anionic Stealth liposomes can eradicate plasmacytomas infiltrated by eosinophils in mice. Our results, therefore, suggest that peroxide-generating compounds may be a useful experimental approach for treating those human tumors that are naturally infiltrated by eosinophils but resistant to conventional therapies.

Effects of sonicated eosinophils on the in vitro sensitivity of human lymphoma cells to glucose oxidase.

We report here that cultured human lymphoma cells in the absence of sonicated eosinophils are sensitive to killing by glucose oxidase (beta-D-glucose:oxygen-oxido reductase; EC 1.1.3.4) at concentrations as low as 0.025 microgram/ml, a level that can be rapidly attained in s.c. tumor implants in mice that receive a single nonlethal injection of enzyme. Multiple clonogenic assays were used to measure the survival of human lymphoma cell lines (H9 and ARH-77) cultured for 14 days in complete RPMI 1640 supplemented with exogenous glucose oxidase (0.025-2.5 micrograms/ml) or an immunoconjugate containing glucose oxidase (0.25-25 micrograms/ml) in the presence or absence of catalase (10 micrograms/ml) or an equal number of sonicated human eosinophils with or without supplemental 100 microM Br-, I-, or SCN-. In addition, we used an immunoassay to measure the concentration of glucose oxidase in s.c. implants of the Sp 2/0 myeloma tumor at 0-30 min after an i.v. injection of 50 micrograms of enzyme into 21 BALB/c mice. Doses of glucose oxidase as small as 0.025 microgram/ml killed more than 3 logs of tumor cells. Catalase completely inhibited, and sonicated human eosinophils partially inhibited, the killing by glucose oxidase or immunoconjugate, whereas supplemental halides had no effect. Glucose oxidase i.v. produced levels > 0.04 microgram/g of tumor for 30 min after injection with a peak concentration of 0.079 microgram/g of tumor within 5 min of injection. These results are important because certain human lymphomas contain extensive extracellular deposits of eosinophil peroxidase, thereby making these tumors potentially less susceptible to killing by otherwise therapeutic doses of glucose oxidase.

Radioimmunodetection of Hodgkin's disease and non-Hodgkin's lymphomas with monoclonal antibody to eosinophil peroxidase.

The purpose of this study was to determine if a radiolabeled murine monoclonal antibody (EOS) directed against eosinophil peroxidase would localize specifically to tumor sites in patients with lymphomas infiltrated by eosinophils. Ten patients with Hodgkin's disease and eosinophilia, three patients with non-Hodgkin's lymphomas and eosinophilia and five control patients received an intravenous injection of 3-10 mg of EOS antibody radiolabeled with 74-155 MBq (2.0-4.2 mCi) of 111In. At intervals of 24, 48 and 72 hr after injection, gamma camera images were obtained along with blood and urine specimens and the imaging results were correlated with the results of other staging modalities. As early as 24 hr after antibody injection, there was clear visualization of identifiable sites of lymphoma with eosinophilia greater than 1 cm in size, including the spleen, bone marrow and lymph nodes. Although EOS also localized nonspecifically to the liver and, in some patients, to the nasopharynx, there was no appreciable uptake in normal bone marrow, spleen, uninvolved lymph nodes, lymphomas without eosinophilia or various other pathologic conditions without eosinophilia. Except for transient pain at tumor sites in three patients, no adverse reactions were noted. We conclude that a radiolabeled monoclonal antibody directed against eosinophil peroxidase localizes to lymphoma sites infiltrated by eosinophils.

Radioimmunodetection of degranulated human eosinophils in mice: a potential model for imaging Hodgkin's disease and other pathologic conditions.

Various human tumors such as lymphomas and carcinomas sometimes contain extensive infiltration by degranulating eosinophils. To determine if degranulated eosinophils are suitable targets for immunolocalization, we performed in vivo distribution and imaging studies in mice, using EOS (a murine monoclonal antibody directed to human eosinophil peroxidase) labeled with indium-111. Adult mice were injected intravenously with radiolabeled EOS antibody or with similarly radiolabeled normal mouse IgG before receiving an intramuscular injection into the right thigh of homogenized human eosinophils adsorbed to latex microspheres. There was striking localization in the right thigh of the radiolabeled EOS antibody detectable by gamma imaging techniques as soon as 24 hr after injection. By contrast, there was little accumulation of radiolabeled normal IgG in the right thigh. We conclude that human eosinophil peroxidase is potentially a suitable target for radioimmunodetection and therapy of neoplasms and pathologic conditions that contain degranulating eosinophils.

Evaluation of biotinylated DNA probes for the detection of gene rearrangements in clinical specimens.

To determine whether a nonisotopic procedure is suitable for analyzing clinical specimens for gene rearrangements, the authors hybridized DNA from 15 specimens of lymphoid tissue with biotinylated DNA probes directed to J beta I + J beta II (T-cell receptor beta chain gene), JH (immunoglobulin gene heavy chain J region), and J kappa (immunoglobulin gene kappa light chain J region). Five cases of benign lymphoid hyperplasia, one case of dermatopathic lymphadenopathy, and one case of small noncleaved follicular center cell lymphoma had germline hybridization patterns when digested with Bam HI, Eco RI, and Hind III restriction endonucleases. Four cases of B-cell lymphoma and three cases of T-cell lymphoma had clearly detectable rearrangements of the genes for immunoglobulin or the T-cell receptor or both. One case of dermatopathic lymphadenopathy had a faint, clonal rearrangement of the T-cell receptor after digestion with Eco RI and Bam III. The authors conclude that biotinylated DNA probes can be useful for analyzing gene rearrangements in clinical specimens.

Detection of rare cells expressing shared lymphoma idiotypes in fetal spleens.

Monoclonal antibodies have been derived against the shared and private idiotypes of immunoglobulin expressed by human B-cell lymphomas. We performed indirect immunofluorescence assays on cells from 3 fetal spleens, 3 adult spleens and 10 hyperplastic lymph nodes with a panel of 5 monoclonal antibodies directed to private lymphoma idiotypes, 2 antibodies directed to shared lymphoma idiotypes and various positive and negative control antibodies. Rare (less than 5%) cells in the fetal spleens, adult spleens and hyperplastic lymph nodes reacted with the 2 antibodies directed to shared lymphoma idiotypes. No cells in any of the specimens were reactive with the antibodies directed to private lymphoma idiotypes. We conclude that rare cells expressing shared lymphoma idiotypes are present during fetal development and in mature lymphoid tissue. This suggests that shared lymphoma idiotypes are expressed as part of a developmental process rather than in response to a specific environmental antigen.

A cytotoxic monoclonal antibody that binds to human B cells, T cells, monocytes, and a subset of lymphomas.

Monoclonal antibodies have been derived against a variety of antigens on human leukocytes. We describe a complement-fixing, IgG2a, murine monoclonal antibody that was derived against ARH-77, a human plasma cell line. In a Western blot of a lysate of ARH-77, anti-ARH-77 recognized a 155 kD protein. On normal B cells, the antibody co-capped with surface immunoglobulin. Analysis of blood and lymph nodes indicated that the determinant recognized by the antibody is expressed by most B cells, T cells, monocytes, and lymphomas with surface immunoglobulin. The determinant was not present on granulocytes, platelets, mantle zone lymphocytes, or serum immunoglobulin. We conclude that anti-ARH-77 is a cytotoxic monoclonal antibody that recognizes a determinant shared by human mononuclear leukocytes. The determinant may be an exposed part of a membrane anchor or bridging protein associated with receptors on mononuclear cells.

In-vitro sensitivity of Hodgkin's disease to hydrogen peroxide toxicity. Correlation with peroxidase activity.

Hodgkin's disease of nodular sclerosis and mixed cellularity subtypes contains numerous eosinophils and substantial amounts of extracellular eosinophil peroxidase (EPO). To determine if the extracellular EPO retains cytotoxic activity, the authors analyzed cells from 13 cases of Hodgkin's disease and ten cases of benign lymphoid hyperplasia for their in vitro sensitivity to killing by a low concentration of hydrogen peroxide. Cells from cases of benign lymphoid hyperplasia (0.5% +/- 1% killing) and lymphocyte predominant Hodgkin's disease (4.5% +/- 6% killing) were significantly (P less than 0.05) less sensitive to killing by hydrogen peroxide than cells from nodular sclerosis Hodgkin's disease (26% +/- 13% killing) and mixed cellularity Hodgkin's disease (52% +/- 9% killing). The authors concluded that cells from Hodgkin's disease of nodular sclerosis and mixed cellularity subtypes have an increased sensitivity to killing by an otherwise nonlethal concentration of hydrogen peroxide.

Cytophilic and cytotoxic properties of human eosinophil peroxidase plus major basic protein.

The cytophilic and cytotoxic properties of an acetate-buffered solution of human eosinophil peroxidase (EPO) plus major basic protein (MBP) were studied to determine the cytotoxic potential of localized eosinophil degranulation in human tissues. When incubated with EPO + MBP for 5 minutes, viable cells of six unrelated types (Sp 2/0; HeLa; human gastric adenocarcinoma; acute lymphocytic leukemia; IM-9; benign lymphoid hyperplasia) developed varying degrees of cytochemically detectable deposits of EPO on the cell membranes. A single-step propidium iodide exclusion assay was then used to show that EPO + MBP in the absence of hydrogen peroxide is substantially cytotoxic only to the acute lymphocytic leukemia and IM-9 cells. In the presence of 0.003% hydrogen peroxide, EPO + MBP was cytotoxic to five types of cells. It is concluded that human EPO in the presence of MBP has an affinity for the membrane of diverse cell types. The toxicity of EPO + MBP is markedly enhanced by the presence of hydrogen peroxide.

Enzyme immunoassay for detection of monoclonal immunoglobulin in lymph nodes. A comparison with histologic and DNA gene rearrangement studies.

This article describes a 1-hour enzyme immunoassay (ELISA) for identifying lymph nodes in which a single immunoglobulin light chain isotype is abnormally predominant. Eighteen lymph nodes representing a variety of reactive hyperplasias and B-cell neoplasms were analyzed with the ELISA and submitted for B- and T-cell gene rearrangement studies at the CT beta, JH and J kappa (JK) loci. In all 18 specimens, there was agreement in the results of the ELISA, the histologic diagnoses, and the results of gene rearrangement studies. In three of the B-cell lymphomas identified by histologic, ELISA, and DNA studies, surface marker phenotyping by flow cytometry did not indicate the presence of a monoclonal cell population. It is concluded that the ELISA correlates well with DNA gene rearrangement studies and is a rapid, simple, and accurate method for identifying lymph nodes containing monoclonal immunoglobulin. Moreover, it is proposed that the ELISA has certain advantages over other methods for determining light chain clonality in clinical specimens, and it is therefore useful to distinguish between benign lymphoid hyperplasia and monoclonal B-cell neoplasms.

Deposition of autofluorescent eosinophil granules in pathologic bone marrow biopsies.

Eosinophil granules are intensely autofluorescent when excited by green light. To determine if eosinophils degranulate in the bone marrows of patients with a variety of diseases, we used green light epifluorescence microscopy to examine deparaffinized and dezenkerized sections of 49 bone marrow core biopsies. In 14 of the biopsies, there was striking extracellular deposition of intensely autofluorescent eosinophil granules in addition to numerous intact eosinophils. Among the 14 specimens with extracellular autofluorescence were seven cases of leukemia, four cases of non-Hodgkin's lymphoma, two cases of myelofibrosis, and one case of pancytopenia with eosinophilia. In the remaining 35 specimens, only intact eosinophils were identifiable. There was no extracellular autofluorescence in three normal marrows, four marrows from AIDS patients, or three biopsies from patients with idiopathic thrombocytopenic purpura (ITP). We conclude that green light epifluorescence microscopy identifies extracellular deposits of eosinophil granules in bone marrow biopsies of some neoplastic disorders and in diseases associated with reticulin fibrosis.