Subcutaneous panniculitis-like T-cell lymphoma: clinicopathologic, immunophenotypic, and genotypic analysis of alpha/beta and gamma/delta subtypes

Subcutaneous panniculitis-like T-cell lymphoma (SPTCL) is an uncommon cutaneous lymphoma that has been proposed as a distinct clinicopathologic entity, but studies of SPTCL are limited. We studied the clinicopathologic, immunophenotypic, and genetic features of 11 SPTCLs. All cases had a variable admixture of pleomorphic small, medium, or large lymphocytes and histiocytes infiltrating the subcutis in a lobular panniculitis-like pattern. A granulomatous reaction was seen in three cases and erythrophagocytosis in four. Karyorrhexis and fat necrosis were present in all cases. Angioinvasion was seen in seven SPTCLs; four had areas of coagulation necrosis. All cases expressed T-cell-associated antigens (CD3epsilon, CD45RO, or CD43) and T-cell receptors (TCR); nine expressed alphabeta TCRs and two expressed gammadelta TCRs. T-cell receptor-gamma, TCRbeta, or TCRdelta genes were clonally rearranged in 8 of 10 cases studied. Both gammadelta SPTCLs expressed Vdelta2+ TCRs and were CD4-, CD8- and CD56+. CD56 was negative in seven of nine alphabeta SPTCLs and inconclusive in the other two. Six of nine alphabeta SPTCLs were CD8+; the CD4/CD8 phenotypes were indeterminate in the other three. Cytolytic granule-associated proteins were expressed by all SPTCLs (11 of 11 were TIA-1+, 4 of 4 were perforin+). In situ hybridization for Epstein-Barr virus-encoded RNA (EBER-1) was negative in all cases. Most patients responded to systemic chemotherapy or local radiation therapy. Seven patients are alive: four without disease (19-73 months) and three with disease (32-72 months); four died: three of disease (3-25 months) and one without disease (42 months). We conclude that SPTCLs are clonal, EBV-, cytotoxic T-cell lymphomas derived from alphabeta T-cells or gammadelta T-cells. The gammadelta SPTCLs appear to be preferentially derived from the Vdelta2+ subset. Subcutaneous panniculitis-like T-cell lymphoma may be rapidly fatal or indolent; local therapy may be appropriate for some patients.

T-cell subset analysis of peripheral T-cell lymphomas by paraffin section immunohistology and correlation of CD4/CD8 results with flow cytometry

Peripheral T-cell lymphomas (PTCLs) are often diagnosed after demonstration of T-lineage-related antigen expression on neoplastic lymphocytes by paraffin immunoperoxidase (PIP). However, complete T-cell subset analysis for helper, suppressor/cytotoxic, alphabeta, and gammadelta phenotypes has not been examined by PIP. Therefore, PIP was performed for CD4, CD8, T-cell intracellular antigen (TIA)-1, and betaF1 expression in 31 PTCLs previously studied for CD4 and CD8 by flow cytometry. The CD4 and CD8 results from both methods were compared. All betaF1- PTCLs were studied for T-cell receptor (TCR)gammadelta by PIP. PIP showed 71% correlation with the 21 PTCLs that had distinct CD4+ CD8- or CD4- CD8+ phenotypes by flow cytometry, with 64% and 90% sensitivity for CD4 and CD8 expression, respectively. Tumor cells in four of six PTCLs that had no clear CD4 or CD8 predominance or coexpression of these antigens by flow cytometry were shown to be CD4+ CD8- or CD4- CD8+ by PIP. Twelve (39%) PTCLs demonstrated a cytotoxic (TIA-1+) phenotype by PIP, including eight CD4- CD8+, one CD4+ CD8- and three CD4- CD8- cases. Of 30 immunoreactive PTCLs, 26 (87%) were alphabeta (betaF1+) by PIP. Both large cell cases among four betaF1- PTCLs were TCRgammadelta+ by PIP, including one gammadelta+ case confirmed by flow cytometry. We conclude that CD4 and CD8 T-cell subsets can be assigned for most PTCLs by PIP, with CD4 showing moderate and CD8 showing strong correlation with flow cytometric results. PIP can also define CD4 or CD8 expression on tumor cells in the PTCLs in which flow cytometry produces inconclusive results. Cytotoxic PTCLs can be identified easily with TIA-1, which can also distinguish cytotoxic from "suppressor" CD8+ PTCLs. Most PTCLs are derived from alphabeta T-cells, however some large cell gammadelta PTCLs may be identified by PIP.

Natural killer-like T-cell lymphoma in the small intestine of a child without evidence of enteropathy

A previously healthy 6-year-old boy developed symptoms of small intestinal obstruction and was found to have a large intraabdominal mass. At laparotomy the mass involved the jejunum and adjacent mesenteric lymph nodes, requiring resection. Microscopic and immunohistochemical studies demonstrated a T-cell non-Hodgkin's lymphoma, confirmed by finding clonal T-cell receptor-beta and -gamma gene rearrangements by Southern blot analysis. The immunophenotype of this lymphoma-CD3+CD4-CD8-CD56+TIA-1+ beta F1(-)-suggests that the tumor cells are cytotoxic natural killer (NK)-like T cells, probably of CD3+CD4-CD8- intraepithelial cell origin. Examination of the adjacent and distal small intestinal mucosa failed to show any significant pathologic change. This case was unusual because intestinal lymphomas in children are usually of B-cell origin and most commonly have small noncleaved cell morphology. Childhood intestinal T-cell lymphomas have not been the focus of specific study but appear to be rare. In adults, intestinal T-cell lymphomas often arise in the background of gluten-sensitive enteropathy (celiac disease). In contrast, this child had peripheral T-cell lymphoma, with NK-like T-cell features, in the small intestine with no clinical or histologic evidence of enteropathy.

TIA-1 expression in lymphoid neoplasms. Identification of subsets with cytotoxic T lymphocyte or natural killer cell differentiation

TIA-I is a 15-kd cytotoxic granule-associated protein expressed in natural killer (NK) cells and cytotoxic T lymphocytes. TIA-1 expression was evaluated by paraffin immunohistochemistry in 115 T- or NK-cell neoplasms, 45 B-cell neoplasms, and 45 Hodgkin's lymphomas. TIA-1-positive granules were identified within the cytoplasm of neoplastic cells in 6/6 large granular lymphocytic leukemias, 11/11 hepatosplenic T-cell lymphomas, 15/15 intestinal T-cell lymphomas, 6/6 NK-like T-cell lymphomas of no special type, 2/2 NK-cell lymphomas, 8/9 nasal T/NK-cell lymphomas, 7/8 subcutaneous T-cell lymphomas, 4/5 pulmonary T- or NK-cell angiocentric lymphomas (lymphomatoid granulomatosis), 12/19 T-cell anaplastic large-cell lymphomas, 2/12 nodal peripheral T-cell lymphomas, 1/3 CD8+ cutaneous T-cell lymphomas, and 5/38 classical Hodgkin's disease. All B-cell neoplasms, nodular lymphocyte-predominant Hodgkin's disease (7 cases), CD4+ cutaneous T-cell lymphomas (6 cases), adult T-cell leukemia/lymphomas (3 cases), T-cell chronic or prolymphocytic leukemias (3 cases), and T-cell lymphoblastic leukemia/lymphomas (7-cases) were TIA-1 negative. These findings indicate that most large granular lymphocytic leukemias, hepatosplenic T-cell lymphomas, intestinal T-cell lymphomas, NK-like T-cell lymphomas, NK-cell lymphomas, nasal T/NK-cell lymphomas, subcutaneous T-cell lymphomas, pulmonary angiocentric lymphomas of T or NK phenotype, and anaplastic large-cell lymphomas are cytotoxic T-or NK-cell neoplasms.

Latent membrane protein antibody reacts with normal hematopoietic precursor cells and leukemic blasts in tissues lacking Epstein-Barr virus genome by polymerase chain reaction

Epstein-Barr virus (EBV)-encoded latent membrane protein (LMP) has been detected in various reactive and neoplastic lymphoproliferations and in some epithelial malignancies. However, data are lacking on LMP immunoreactivity in normal and neoplastic hematopoietic cells. Therefore, the authors studied LMP staining in 29 paraffin-embedded tissues containing these cells and correlated the findings with polymerase chain reaction (PCR) analysis for a EBV-BAM HI W genome sequence in 15 of these cases. Latent membrane protein immunostains showed strong uniform marking of normal early myeloid and erythroid precursors, whereas neutrophils, bands and late normoblasts were negative. Leukemic myeloblasts and lymphoblasts were also strongly positive for LMP. Polymerase chain reaction analysis showed no evidence of EBV-BAM HI W genome in 13 cases with amplifiable DNA. This study indicates normal hematopoietic precursor cells and leukemic blasts mark strongly with monoclonal LMP antibody. Furthermore, the absence of EBV genome in these tissues suggests a lack of specificity of monoclonal LMP for EBV-infected cells in the marrow.

Natural killer-like T-cell lymphomas: aggressive lymphomas of T-large granular lymphocytes

Natural killer (NK)-like T cells are major histocompatibility complex-unrestricted cytotoxic T cells that are surface CD3-positive, express NK-cell antigens, and rearrange their T-cell receptor. Most neoplasms arising from this T-cell subpopulation have been a chronic lymphoproliferative disease referred to as T-large granular lymphocyte (LGL) leukemia. Only 10 NK-like T-cell lymphomas have been described in detail previously; this study presents the clinicopathologic features of six others and distinguishes these lymphomas from T-LGL leukemia. All patients presented with B-symptoms and often had marked hepatosplenomegaly without significant peripheral lymphadenopathy. Four of the six patients were immunosuppressed. All had CD3, CD8, CD56-positive tumors, presumably of hepatosplenic (n = 3), intestinal (n = 1), pulmonary (n = 1), or nodal (n = 1) origin. Three patients had lymphomatous bone marrow infiltrates, and four had peripheral blood involvement by neoplastic large lymphocytes, some of which had a blastic appearance or resembled virocytes. Azurophilic granules, ultrastructurally corresponding to cytoplasmic dense core and/or double density granules, were seen in all cases. T-cell clonality was shown in five tumors by Southern blot analysis, and three had abnormal karyotypes. Two untreated patients died 20 days after presentation, and three patients who received combination chemotherapy died within 5 months of presentation. One patient remains in complete remission 22 months after treatment. These findings suggest NK-like T-cell lymphomas are aggressive, are clinicopathologically distinct from T-LGL leukemia, and should be in the differential diagnosis of extranodal T-cell lymphoproliferations, including those in immunosuppressed patients. Furthermore, the LGL morphology, phenotype, and tissue distribution of some NK-like T-cell lymphomas suggest they arise from thymic-independent T cells of the hepatic sinusoids and intestinal mucosa.

Ringed sideroblasts: a frequent observation after bone marrow transplantation

Dyserythropoiesis after bone marrow transplantation is common, but the presence of ringed sideroblasts (RS) has not been evaluated fully. To examine for RS, a combined silver and Perls' Prussian blue stain, shown previously to be more sensitive for detecting RS than Perls' Prussian blue stain alone, was used on post-transplant marrow aspirate sections from 39 patients who received marrow transplants (allogeneic, 28; autologous, 11) for a variety of disorders. Marrow aspirate sections were available for comparison from 11 of these patients before any treatment as well as from five patients with normal marrows and normal peripheral blood cell counts. Aspirates were not performed on donor marrows. By the modified silver stain, RS were present in 34 (87%) patients whose marrows were sampled 0.5 to 39 (median, 1.5) months post-transplant including 10 of 11 patients with autologous transplants (no graft versus host disease prophylaxis). In contrast, seven of 36 (19%) of these marrows contained RS when stained with Perls' reaction alone. Only one of 11 pretransplant marrows and none of five normal marrows contained RS when stained by either method. These results demonstrate that RS are present in most post-transplant marrows even beyond the usual period of reconstitution (28 days), and this finding can be included among the features of dyserythropoiesis seen after transplantation. RS apparently are not related to pretransplant pathology or post-transplant therapy. This study also confirms previous observations that modified silver stains are more sensitive for detecting RS than Perls' Prussian blue stain.

T-cell-rich B-cell lymphomas: diagnosis and response to therapy of 44 patients

PURPOSE

Clinicopathologic features of 44 patients with well-documented T-cell-rich B-cell lymphomas (TCRBCLs) were reviewed to determine if there were distinguishing clinical characteristics and to evaluate the responsiveness to therapy.

PATIENTS AND METHODS

Forty-one patients had de novo TCRBCL, while three patients had a prior diagnosis of diffuse large B-cell lymphoma. Seventeen TCRBCLs were identified from a retrospective analysis of 176 lymphomas diagnosed before 1988 as peripheral T-cell lymphoma (PTCLs). The initial pathologic diagnosis was incorrect in 36 of 44 cases (82%), usually due to the absence of adequate immunophenotypic and/or genotypic studies at the initial study.

RESULTS

The median age of patients was 53 years (range, 17 to 92), and the male-to-female ratio was 1.4:1. B symptoms were present in 22 of 41 patients (54%); splenomegaly was detected in 11 patients (25%). Clinical stage at diagnosis was as follows: I (n = 8), II (n = 6), III (n = 15), IV (n = 14), and unstaged (n = 1). Although therapy was heterogeneous, the disease-free survival (DFS) and overall survival (OS) rates at 3 years for patients with de novo TCRBCL were 29% and 46%, respectively. A complete response (CR) to combination chemotherapy for intermediate-grade lymphomas was observed in 16 of 26 patients (62%); 11 of these patients (42%) had a continuous CR, compared with one of 14 patients (7%) who received radiation therapy or therapy for low-grade lymphoma or Hodgkin's disease (HD) (P < .05). However, there was no difference in OS between patients who received chemotherapy for intermediate-grade lymphoma versus other therapies (49% v 48%) due to a high response rate to salvage therapies, including seven patients without disease after marrow transplantation.

CONCLUSION

TCRBCLs are difficult to recognize without immunoperoxidase studies. Patients with TCRBCL have clinical features similar to patients with other large B-cell lymphomas, except they may have more splenomegaly and advanced-stage disease; they should receive combination chemotherapy directed at large-cell lymphomas.

Paracortical nodular T-cell lymphoma. Identification of an unusual variant of peripheral T-cell lymphoma

Peripheral T-cell lymphomas (PTCLs) are regarded as diffuse proliferations. We describe an unusual paracortical nodular growth pattern in four nodal PTCLs that were initially interpreted as atypical lymphoid hyperplasia in three patients and small B-cell lymphoma with plasmacytic differentiation in a fourth. The nodules were vague to easily discernible and produced minimal to partial architectural distortion. Sinuses were often open, and scattered cortical lymphoid follicles with atretic to hyperplastic germinal centers were present. Clusters of tumor cells abutted some follicles in all cases, and in one case they exhibited focal T-zone expansion. Hypervascularity was not prominent, but a few nodules surrounded epithelioid venules, imparting an angiofollicular appearance. The nodules were composed primarily of small lymphocytes with irregular nuclei admixed with scattered large transformed cells, both cell types having clear cytoplasm. Paraffin immunoperoxidase showed that the nodules were composed of T cells. Dendritic cell networks were present only in follicular centers. Southern blot analysis found T-cell receptor gene rearrangements and a germline immunoglobulin gene configuration in all four nodes. These paracortical clear cell nodules of clonal T cells may be a special type of PTCL. Alternatively, they may represent early foci of lymphoma or they may be a subgroup of T-zone lymphoma. Paracortical nodular PTCL must be differentiated from atypical hyperplastic lesions and some B-cell lymphomas.

Marrow mast cell hyperplasia in hairy cell leukemia

Marrow mast cells, frequently elevated in chronic B-lymphoproliferative disorders, were counted per high magnification field (HMF) on toluidine blue stained marrow biopsies from 34 patients diagnosed with hairy cell leukemia (HCL); similar counts were performed on splenic sections from nine of these patients. Biopsies from 28 normal marrow transplant donors and 10 normal spleens served as controls. Mast cells were distributed irregularly throughout normal and HCL marrows, but tended to be more concentrated about the hairy cells in cases focally involved by HCL. HCL marrows averaged 12.7 mast cells/HMF compared to 1.1 for controls. Although most normal marrows (86%) averaged < 2 mast cells/HMF, 88% of HCL cases averaged > or = 2 mast cells/HMF. Splenic mast cells averaged < 1/HMF for both HCL and control cases. By electron microscopy, marrow mast cells in HCL demonstrated normal substructure with numerous granules. The cell surfaces of mast cells showed filopodia that often came in contact with those of hairy cells. These results indicate most HCL marrows, but not spleens, are associated with varying degrees of mast cell hyperplasia. Furthermore, this study suggests a biologic interaction between mast cells and hairy cells. The significance of marrow mast cell hyperplasia in HCL is unknown, but these mast cells may contribute to the increased reticulin fiber network observed in HCL marrows.

Acquired immunodeficiency syndrome-like illness associated with systemic Mycoplasma fermentans infection in a human immunodeficiency virus-negative homosexual man

A 35-year-old homosexual man developed a composite nodal Kaposi's sarcoma and peripheral T-cell lymphoma that were associated with a peripheral blood CD4-positive lymphocyte count of only 43/mm3. The patient subsequently developed Pneumocystis carinii pneumonitis and eventually died due to disseminated Cryptococcus neoformans. Numerous premortem tests for the presence of human immunodeficiency virus (HIV) types 1 and 2 were negative by the enzyme-linked immunosorbent assay, Western blot, viral isolation, and polymerase chain reaction techniques. Postmortem evaluations for HIV-1, HIV-2, human T-cell lymphotropic virus (HTLV)-I, and HTLV-II also were negative by polymerase chain reaction, immunofluorescence assays, and viral isolation. A systemic infection by Mycoplasma fermentans, however, was documented by immunohistochemistry and polymerase chain reaction in premortem and postmortem tissues. This recently recognized human pathogen has produced systemic infections in patients with the acquired immunodeficiency syndrome (AIDS) and in previously healthy non-AIDS patients who characteristically have a fulminant flu-like illness. Additionally, M fermentans has enhanced the cytopathic effect of HIV in in vitro studies and has produced fatal wasting illnesses with terminal lymphopenia in inoculated adult silvered leaf monkeys. This report is the first description of an association between M fermentans infection and an AIDS-like illness in an HIV-negative individual. The etiology of the severe immunosuppression in this patient and the associated role of M fermentans remain to be determined by further investigations.

Interleukin-4 may contribute to the abundant T-cell reaction and paucity of neoplastic B cells in T-cell-rich B-cell lymphomas

T-cell-rich B-cell lymphomas (TCRBCLs) are diffuse lymphomas that contain a minority of large neoplastic B cells amidst a majority of non-neoplastic T cells and numerous histiocytes, an unusually pronounced reactive component not seen in most diffuse large B-cell lymphomas (DLBCLs). This reaction may be influenced by various cytokines secreted by lymphoma or reactive cells; therefore, expression of interleukin (IL)-1 beta, IL-2, IL-4, IL-6, and IL-9 was evaluated immunohistochemically on paraffin-embedded sections of 18 TCRBCLs and was compared with that of 15 DLBCLs containing a minority of reactive T cells and to that of seven reactive lymph nodes. Moderate to intense expression of IL-4 was detected in variable numbers of tumor cells and in numerous histiocytes in 16 TCRBCLs. In contrast, intense IL-4 expression in numerous histiocytes was observed in only one of 15 DLBCLs with few T cells. In four other DLBCLs and three reactive nodes, moderate to intense staining for IL-4 was noted only in rare large transformed cells or in occasional histiocytes. Except for one IL-1 beta positive and another IL-9 positive TCRBCL, there was no marking or weak staining only with other cytokine antibodies in the neoplastic and reactive cases studied. The expression of IL-4 in most TCRBCLs, but not in other DLBCLs or in reactive nodes, suggests that this cytokine is one factor involved in the pathobiology of the abundant T-cell reaction and, perhaps, contributes to the paucity of neoplastic B cells in TCRBCLs.

T-cell-rich B-cell lymphomas. A clinicopathologic study of 19 cases

T-cell-rich B-cell lymphomas (TCRBCLs) are recently described, unusual non-Hodgkin's lymphomas that have a diffuse morphology, a predominance of reactive T-cells, and a minority of neoplastic B-cells. The clinical and pathological features of 19 TCRBCLs, all of which demonstrated B-cell clonality, are presented. These lymphomas generally affected older patients by widespread disease and usually were nodal in origin. Treatment varied, but continuous complete remissions (eight patients) were achieved only in those receiving chemotherapy directed at intermediate-grade lymphomas. Although morphologically heterogeneous, all cases resembled peripheral T-cell lymphomas (PTCLs); several TCRBCLs also contained Reed-Sternberg-like cells. Flow cytometry or frozen-section immunoperoxidase failed to detect monotypic immunoglobulin (Ig) in eight of eight cases tested. In contrast, paraffin immunoperoxidase was very useful diagnostically, showing large L26 (CD20-associated) positive cells scattered singly or in small clusters among numerous small T-cells (UCHL1[CD45RO] positive) in all cases. Monotypic cytoplasmic Ig was present in 16 of 19 cases, one of which exhibited plasmacytic differentiation. Southern blot analysis demonstrated relatively faint Ig JH and/or JK bands, indicating a small monoclonal B-cell population in nine of 11 cases, one of which also showed a bcl-2 rearrangement. No T-cell receptor gene rearrangements were observed. These results showed that TCRBCLs may be easily confused with PTCLs or occasionally confused with Hodgkin's disease. TCRBCLs are probably heterogeneous biologically; some cases are of follicular center cell origin. These lymphomas respond to chemotherapy directed at intermediate-grade lymphomas, apparently have a better prognosis than PTCLs, and seem to represent morphological variants of different types of large B-cell lymphomas.

T-cell lymphoblastic leukemia/lymphoma presenting in a recurrent thymoma

A lymphoblastic malignancy presented as the lymphoid component of a recurrent invasive lymphocyte-rich thymoma that had been previously resected and treated with chemotherapy. This high grade lymphoid neoplasm concurrently involved mediastinal nodes and subsequently disseminated to pleural fluid and peripheral blood. Lymphocytes with convoluted nuclei demonstrated a T-lymphoblastic phenotype (UCHL1+, Leu-22+, TdT+) by immunohistochemical studies. T-cell lymphoblastic leukemia/lymphomas in this clinical setting have not been reported, and this case suggests that neoplastic transformation of lymphocytes may occur in longstanding refractory thymomas.

Leu-22 (L60). A more sensitive marker than UCHL1 for peripheral T-cell lymphomas, particularly large-cell types

Paraffin-embedded sections of 77 peripheral T-cell lymphomas (PTCLs) were stained with several monoclonal antibodies, including the preferential T-cell markers Leu-22 (L60[CD43]) and UCHL1 (CD45RO). The staining characteristics of L60 and UCHL1 were compared to determine the value of each in the immunophenotypic analysis of PTCLs. Lineage specificity was evaluated among 39 B-cell lymphomas and 33 cases of Hodgkin's disease (HD). L60 and/or UCHL1 stained 95% of PTCLs, whereas L60 and UCHL1 alone stained 90% and 69% of cases, respectively. L60 demonstrated significantly greater numbers of immunopositive tumor cells than UCHL1 in 37% of the PTCL cases, principally because of enhanced marking of large, neoplastic cells. UCHL1 was a better marker in only 10% of the PTCL cases. L60 stained 33% of B-cell lymphomas, usually small lymphocytic or lymphoplasmacytic types. UCHL1 stained only 8% of B-cell lymphomas, all large-cell types. L60 and UCHL1 stained Reed-Sternberg cells and variants in three cases of nodular sclerosing HD. These results suggest that both L60 and UCHL1 are useful markers of PTCLs in routinely processed tissue. L60 is a more sensitive marker of large neoplastic T-cells than UCHL1 but is less lineage-specific. These antibodies are most effective when used as part of a panel of monoclonal antibodies.

Large-cell lymphomas: clinical and prognostic features

We reviewed the clinical and pathologic features in 186 patients with large-cell lymphomas seen at Vanderbilt University Hospital between 1970 and 1986. Ninety-two cases (49%) were large noncleaved-cell lymphoma (LNCCL), 61 cases (33%) were large-cleaved-cell lymphoma (LCCL), 17 cases (9%) were peripheral T-cell lymphoma (PTCL), and 16 cases (9%) were immunoblastic sarcoma of B cells (IBS-B). These subsets of large-cell lymphoma did not differ with respect to median age, distribution by stage, or incidence of bone marrow involvement. Significant differences between groups were noted with regard to male:female ratio, incidence of symptoms, incidence of extranodal disease, and pattern of adenopathy. However, when LCCL was excluded from the analysis, none of these differences were significant. By univariate analysis, age, stage, marrow involvement, extranodal disease, B symptoms, elevated serum lactic dehydrogenase (LDH), and diffuse pattern were unfavorable prognostic features in large-cell lymphoma. However, when cases were stratified by cell of origin, nodular versus diffuse pattern was of no prognostic significance. Nodularity was favorable only because 71% of nodular and nodular-diffuse cases were LCCL, while the majority of diffuse cases were LNCCL. Although IBS-B is considered a "high-grade" lymphoma, we found no evidence for inferior survival in these patients compared with LNCCL or LCCL. In fact, survival was better in IBS-B than in LNCCL or LCCL, although this difference was not significant. However, survival was significantly inferior in PTCL (median, 11 months) compared with the other subsets of large-cell lymphoma (median, 46 months; P = .038, log-rank test). Since the association of PTCL and an inferior survival has most often been noted in the context of "second-generation" chemotherapy, we believe that this association may be therapy-dependent and may be minimized by the use of more aggressive chemotherapy regimens.

Silver stain for ringed sideroblasts. A sensitive method that differs from Perls' reaction in mechanism and clinical application

A modified silver stain is described for the demonstration of ringed sideroblasts in bone marrow. It is more sensitive than Perls' reaction for that purpose, especially when iron stores of marrow are low or absent. Ringed sideroblasts may still be demonstrable by silver stain in cases of sideroblastic anemia without ringed sideroblasts, for which severe iron deficiency prevents detection of the abnormal sideroblasts by Perls' reaction. As iron has been reported to be present in mitochondria of ringed sideroblasts in the form of ferric phosphate, it is possible that the silver stain demonstrates the phosphate moiety and not the iron, thus explaining its greater sensitivity in iron deficiency as compared to Perls' reaction. Further study is necessary to confirm the staining mechanism, to elucidate the composition of iron deposits, and to explain the pathophysiology of sideroblastic anemia.

Atypical carcinoid of the lung. A distinct clinicopathologic entity

Eleven cases of atypical carcinoid (AC) of the lung were identified during an eight-year period. Their clinical features and treatment responses were contrasted with our experience at Vanderbilt with small cell lung cancer (SCLC) and a literature review of typical bronchial carcinoids (TC). Clinically, there were no features to distinguish AC from TC except for age at diagnosis (59 vs 49 years). Atypical carcinoid was similar to SCLC with respect to many clinical features, although female sex, absence of smoking history and localized disease at presentation were more common in AC. Pathologically, these tumors were distinguished by cellular atypia, necrosis, architectural disorder, or increased mitotic rate in the presence of a recognizable carcinoid pattern. Immunoperoxidase staining revealed no difference between AC and TC or SCLC. Atypical carcinoid of the lung represents a distinct clinicopathologic disease.