Bone marrow histopathology in peripheral T-cell lymphomas


Dr Ahmet Dogan, Department of Histopathology, University College London, 21 University Street, Rockefeller Building, London WC1E 6JJ, UK. E-mail:


Peripheral T-cell lymphomas (PTCL) account for 10–15% of all lymphoproliferative disorders in the western hemisphere. In PTCL, bone marrow biopsy is performed to establish the diagnosis, rule out other pathology, assess the extent of disease and monitor treatment response. The frequency and histology of bone marrow involvement varies greatly between different clinicopathological entities recognized by the World Health Organisation (WHO) classification, reflecting the differences in the underlying biology. Some lymphomas, such as angioimmunoblastic T-cell lymphoma, show nodular and/or interstitial pattern of infiltration with accompanying reactive changes. Others, including hepatosplenic T-cell lymphoma and large granular lymphocyte leukaemia, are characterized by intrasinusoidal infiltration. In many instances the pathological features are subtle and immunohistochemical and molecular studies are required for the diagnosis. Histological appearances may overlap with a variety of reactive T-cell proliferations and other malignancies. Furthermore PTCL frequently induce secondary changes in the marrow that may obscure the neoplastic infiltrate. The diagnosis often requires critical integration of the information obtained from clinical features, peripheral blood, bone marrow aspirate and biopsy findings. In this article we review the histopathology of bone marrow biopsy in PTCL within the context of the new WHO classification.

Peripheral T-cell lymphomas (PTCL) account for 10–15% of all lymphoproliferative disorders in the western hemisphere (Melnyk et al, 1997; Rudiger et al, 2002). The current World Health Organization (WHO) classification of mature T-cell and natural killer (NK) cell neoplasms recognizes 16 entities (Jaffe et al, 2001). The most commonly encountered entities are PTCL unspecified, angioimmunoblastic T-cell lymphoma (AITL), anaplastic large cell lymphoma (ALCL) and cutaneous T-cell lymphomas (CTCL). In the United Kingdom, these four entities account for approximately 80% all PTCL seen, the others being relatively rare. In the context of PTCL, bone marrow biopsy is performed to establish the diagnosis, rule out other pathology, assess the extent of disease (staging) and monitor treatment response. The frequency of bone marrow involvement varies greatly between different clinicopathological entities, reflecting the differences in the underlying biology (Table I). In many cases of PTCL with systemic presentation, a bone marrow biopsy is the first diagnostic procedure performed and may be the only source of tissue available for assessment. However the assessment of the bone marrow histology is seldom straightforward. In many instances, the histological features are subtle and overlap with a variety of reactive T-cell proliferations and other malignant processes. Furthermore, T-cell lymphomas frequently induce secondary changes in the marrow microenvironment, such as haemaphagocytosis (Falini et al, 1990; Linn et al, 1995), granulocytic hyperplasia, eosinophilia, plasmacytosis, granulomatous inflammation, vascular proliferation and myelofibrosis (Rao et al, 2003; Uehara et al, 2003), all of which may obscure the neoplastic infiltrate. The diagnosis often requires critical integration of the information obtained from clinical features, peripheral blood and aspirate findings, and histopathological examination of the bone marrow biopsy. The latter has been greatly improved in recent years with developments in immunohistochemical detection methods and availability of wide range of antibodies for paraffin-embedded tissue samples (Table II).

Table I.  Bone marrow involvement in peripheral T-cell lymphomas.
Primary presentationLymphoma typeHistological bone marrow involvement (%)
  1. NK, natural killer; TCL, T-cell lymphoma; ALCL, anaplastic large cell lymphoma.

Leukaemic/disseminatedT-cell prolymphocytic leukaemia>95
T-cell large granular lymphocyte leukaemia>95
Adult T-cell lymphoma/leukaemia60–70
Aggressive NK-cell leukaemia>95
NodalAngioimmunoblastic TCL60–70
Peripheral TCL, unspecified20–40
ExtranodalMycosis fungoides<5
Primary cutaneous ALCL<5
Subcutaneous panniculitis-like TCL<5
Enteropathy-type TCL<5
Hepatosplenic TCL>95
Extranodal NK/TCL, nasal type<5
Table II.  Immunophenotypic markers that are suitable for staining T cells in paraffin-embedded bone marrow trephine biopsies.
MarkerExpression in normal cells
CD3T cells, NK cells
CD2T cells
CD5T cells
CD7T cells, myeloid cells
CD8Cytotoxic T cells
CD4Helper T cells, macrophages
CD10B-cell subsets, a subset of T cells
CD16NK cells
CD56T-cell, NK cells
CD57T cell subsets
TCR-beta chainMost T cells
CD25Activated T and B cells
CD30Activated T-cell and B cells
TIA-1Cytotoxic granules
Granzyme/perforinActivated cytotoxic T cells
Alk-1Not expressed in normal lymphoid cells
CD45ROMemory T cells, macrophages
CD43T cells, myeloid cells, a subset of B cells
TdTPrecursor T and B cells

The pathology literature on the morphology of PTCL in marrow biopsies is rather limited. This is partly the result of the relative rarity of these diseases and also partly because of the poorly reproducible classifications used until the publication of the current WHO classification (Jaffe et al, 2001). Many of the available studies lump together a number of fundamentally different clinicopathological entities, giving little help to the practicing pathologists for individual cases. The best resources for practical issues remain the two textbooks on bone marrow pathology published on either side of the Atlantic (Bain et al, 2001; Foucar, 2001). The aim of this article is to review the histopathology of bone marrow biopsy in PTCL within the context of the new WHO classification, to discuss some of the recent developments on the topic and to highlight some of the pitfalls in diagnosis.

Angioimmunoblastic T-cell lymphoma

Angioimmunoblastic T-cell lymphoma was first described as a clinicopathological syndrome characterized by generalized lymphadenopathy, hepatosplenomegaly, anaemia and hypergammaglobulinaemia (Frizzera et al, 1974; Knecht et al, 1985). The lymph node histology showed a number of distinctive features, including partial effacement of normal architecture by a polymorphic inflammatory infiltrate that included large blasts and marked vascular proliferation. Based on these histological appearances, the disease was initially designated by variety of terms including immunoblastic lymphadenopathy, lymphogranulomatosis X and angioimmunoblastic lymphadenopathy with dysproteinaemia (AILD) (Dogan et al, 2003a). As the histological features of malignancy were not easily recognizable in many cases, AITL was initially thought to be an unusual reactive hyperplasia with a tendency to develop into a lymphoma rather than a frank neoplasm at the onset. However molecular clonality studies, either with Southern blotting or by polymerase chain reaction (PCR), have consistently shown the presence of monoclonal T-cell populations in vast majority of the cases strongly suggesting that AITL is a neoplastic process (Dogan et al, 2003a). This is further supported by limited number of cytogenetic studies demonstrating clonal chromosomal abnormalities within the T-cell subset (Schlegelberger et al, 1990, 1994). AITL was included in the Updated Kiel classification of lymphomas as T-cell lymphoma of AILD-type and the current WHO classification has opted for the term AITL to stress the neoplastic nature of this disease (Jaffe & Ralfkiaer, 2001).

The diagnosis of AITL is almost always determined by lymph node biopsy. The histological changes in the lymph node have been well characterized and the diagnosis is highly reproducible despite the difficulties in microscopic identification of the tumour cells. Typically, there is paracortical expansion with partial or complete effacement of the lymph node architecture. The infiltrate is polymorphic and comprises small lymphocytes, aggregates of clear cells, scattered large transformed blasts, occasional cells resembling Reed–Strernberg cells, plasma cells and eosinophils in a background of marked proliferation of small arborizing vessels. Follicles, where present, have a burnt out/depleted appearance.

Bone marrow involvement is seen in most cases and often mimics the features seen in the peripheral lymph nodes (Ghani & Krause, 1985; Gaulard et al, 1991). There is increased vascularity and a patchy polymorphic infiltrate comprising aggregates of small–medium sized lymphoid cells, scattered large transformed blasts, plasma cells, epithelioid histiocytes and eosinophils and, only occasionally, aggregates of clear cells (Fig 1). The infiltrate may be located at paratrabecular or intertrabecular locations. In some cases, plasma cell infiltrate can be very marked, mimicking a plasma cell neoplasm. Reticulin and collagen fibres are increased in the areas involved by the tumour and may mimic myelofibrosis.

Figure 1.

Bone marrow infiltration in AITL. (A) Bone marrow biopsy showing paratrabecular infiltrate of neoplastic lymphoid cells in a background rich in histiocytes (haematoxylin eosin, H & E). (B) Bone marrow biopsy showing a polymorphic infiltrate and marked proliferation of small vessels similar to that seen in lymph nodes involved by AITL (H & E). (C) Bone marrow biopsy showing a polymorphic infiltrate comprising aggregates of epithelioid histiocytes, lymphoid cells and eosinophils mimicking Hodgkin lymphoma (H & E). (D) and (E) High power view of bone marrow biopsy showing medium-sized atypical lymphoid cells with clear cytoplasm, a characteristic feature of AITL (H & E). (F) and (G) CD3 staining showing nodular (F) or interstitial infiltration (G) by AITL (immunoperoxidase). (H) High power view of CD3-stained bone marrow biopsy showing cytological atypia of the neoplastic infiltrate. (immunoperoxidase). (I) Occasionally neoplastic cells of AITL may express CD10 (immunoperoxidase). (J) A large intertrabecular nodular infiltrate in AITL. (K) and (L) The nodule is composed of a peripheral rim of CD3+ neoplastic T cells and a central zone containing CD20 positive large B cells (immunoperoxidase). (M) The B cells are positive for EBV (blue–black nuclear staining) with in situ hybridization. This is a secondary EBV-mediated proliferation of B cells presenting in bone marrow [nitroblue tetrazolium chloride/ 5-bromo-4-chloro-3-indolyl phosphate (NBT/BCIP)]. Original magnification: A, F ×40; B, G ×100; C, K, L ×200; D, E, H, I, M ×400; J ×15.

The neoplastic cells of AITL are typically CD3+ and CD4+ and negative for cytotoxic T-cell markers, including CD8. Recently it has been shown that the tumour cells aberrantly express CD10 in around 80% of the cases (Attygalle et al, 2002). When present, this is considered to be a very useful phenotypic marker for diagnosis of AITL. CD21 staining characteristically demonstrates marked proliferation of follicular dendritic cells, which usually wrap around small vessels. Numerous large B-cell expressing Epstein–Barr virus (EBV) products are seen in a third of cases.

In the marrow, CD3 staining usually identifies many more T cells than that can be identified in haematoxylin eosin (H & E) sections (Fig 1). Interestingly, CD10, which is expressed by neoplastic T cells in the lymph nodes involved by AITL, is rarely expressed by the tumour cells infiltrating the marrow (Attygalle et al, 2004) (Fig 1). Similarly, proliferation of follicular dendritic cells, a characteristic feature of AITL in the lymph nodes, is seldom seen in the marrow (Attygalle et al, 2004). The plasma cell infiltrate, when present, is always polytypic for light chain expression. CD20 and/or CD79a positive large B cells, some resembling Reed–Sternberg cells, can be identified in a subset of cases. These are frequently positive for EBV and occasionally may be extensive, mimicking a large B-cell lymphoma (Fig 1).

Anaplastic large cell lymphoma

In the WHO classification, the term ALCL is used to define a number of clinicopathological entities presenting as either systemic, primarily nodal or cutaneous disease. These share, as a unifying feature, the presence of large cytologically atypical CD30+ cells. The primary cutaneous form of ALCL appears to be a clinically indolent disease that is usually limited to the skin and responds well to locally directed treatments. Bone marrow involvement is not seen in this form of ALCL. In contrast, primary nodal ALCL is a systemic disease characterized by large atypical lymphoid cells with abundant cytoplasm and characteristic horseshoe shaped nuclei (hallmark cells) (Stein et al, 2000). Most cases of primary nodal ALCL have translocations involving the ALK-1 gene, which leads to overexpression of the gene product (Lamant et al, 1996; Benharroch et al, 1998). A subset of the cases with identical histological and immunophenotypic features lack ALK-1 translocation.

Bone marrow involvement is seen only in around 25% systemic ALCL cases (Fraga et al, 1995). The infiltration is usually interstitial or, rarely, vaguely nodular. In some instances, hallmark cells are easily identified in a fibrotic background, in others the infiltrate may be too subtle to be seen in routinely stained sections. In particular, infiltration by the small cell variant of ALCL, which often presents with leukaemic dissemination without lymphadenopathy, may be impossible to diagnose on routinely stained sections (Bayle et al, 1999; Sadahira et al, 1999). The presence of haematological abnormalities, such as anaemia, is frequently associated with bone marrow involvement and warrant further investigations (Fraga et al, 1995).

The tumour cells of ALCL are virtually always positive for CD30 and cytotoxic T-cell markers but show variable expression of pan T-cell markers as well as CD45 and epithelial membrane antigen (EMA) (Stein et al, 2000). Nucleocytoplasmic or cytoplasmic Alk-1 protein expression is present in the majority of the cases. It is recommended that immunohistochemistry for Alk-1, CD30 and EMA is routinely performed on staging marrow to identify minimal infiltration. The bone marrow involvement appears to be associated with worse prognosis in ALCL (Fraga et al, 1995).

Adult T-cell leukaemia/lymphoma

Adult T-cell leukaemia/lymphoma (ATLL) is a PTCL which is endemic in several regions of the world including Japan and the Caribbean (Kikuchi et al, 2001). Its prevalence is closely linked to that of human T-cell lymphotrophic virus type I (HTLV-1) infection. HTLV-1 is thought to be the cause of this disease although the viral infection by itself it is not enough for neoplastic transformation. The time span between the infection and the development of lymphoma or leukaemia is often several decades. When the leukaemia/lymphoma develops, the disease can take a number of clinicopathological forms. At the one end of the spectrum, the patient may have mild lymphocytosis with a very indolent clinical course. At the other end, the patient may develop an aggressive clinical syndrome characterized by systemic disease involving peripheral blood, bone marrow, lymph nodes and sometimes the skin. The latter presentation accounts for most of the cases and is often referred as the ‘acute’ variant of ATLL (Kikuchi et al, 2001).

Medium to large size atypical lymphoid cells with deeply basophilic cytoplasm, polylobated nuclei and prominent nucleoli (so-called flower cells) are seen in the peripheral blood (Fig 2). When lymph nodes or other solid tissues are affected, a diffuse monomorphic infiltrate of similar cells destroying the underlying tissue architecture is seen.

Figure 2.

Blood and bone marrow involvement in ATLL (HTLV-1+). (A) Peripheral blood in ATLL. Atypical lymphoid cells of various sizes with irregular ‘flower-like’ nuclear contours (Wright–Giemsa). (B) Bone marrow biopsy showing a subtle interstitial infiltrate of atypical lymphoid cells with irregular nuclear contours (arrow heads and inset) similar to that seen in the peripheral blood (haematoxylin eosin, H & E). (C) CD3 staining highlights the subtle interstitial infiltration by ATLL that was not apparent on routinely stained sections. Some of the irregularities of the nuclear membrane can be best appreciated with CD3 staining (inset) (immunoperoxidase). (D) Marked paratrabecular infiltration by ATLL in a patient presenting with hypercalcaemia. Note the marked bone resorption with the moth-eaten scalloped appearance of the bony trabeculi (H & E). (E) In this case the tumour cells have large cell morphology (H & E). (F) The tumour cells express CD25 strongly, a characteristic feature of ATLL (immunoperoxidase). Original magnification: A ×1000; B, C ×400; D ×200; E, F ×650.

The marrow is involved in most, but not all of the cases (Jaffe et al, 1984). The extent of involvement varies greatly from case to case. Some cases may have extensive diffuse infiltration mimicking an acute leukaemia, but more frequently the infiltrate is patchy and subtle and may not be detectable in routine examination of bone marrow aspirate and bone marrow biopsies (Fig 2). The cytology of the tumour cells varies greatly. In some instances only small lymphoid cells are seen, but in the majority of cases there are larger cells with cytological atypia. Other marrow features include increased vascularity, eosinophilia and, frequently, morphological evidence of increased bone resorption, which is considered to be the cause of hypercalcaemia often seen in these patients (Jaffe et al, 1984).

In ATLL, the tumour cells have a characteristic immunophenotype that can be demonstrated by either flow cytometry on peripheral blood or bone marrow aspirates or by immunohistochemistry on paraffin-embedded tissue sections (Fig 2). The tumour cells express pan T-cell markers CD3, CD2, CD5 but not CD7. Additionally, most cases are CD4+ and strongly CD25+ but are negative for cytotoxic T-cell markers including CD8. Large cell variants can mimic ALCL and can express CD30.

Peripheral T-cell lymphoma, unspecified

Historically PTCL, unspecified has been the most frequent type of T-cell lymphoma diagnosed by pathologists (Lopez-Guillermo et al, 1998). The tumours diagnosed as such appear to be biologically and clinically very heterogeneous and, in practice, the term is used to define a group of T-cell neoplasms that cannot be classified into one of the better defined clinicopathological entities in the WHO classification (Ralfkiaer et al, 2001). The pathological features and immunophenotype show considerable variation and reflect the biological heterogeneity within the group. For example in PTCL, unspecified presenting primarily in the lymph nodes, there may be cases showing pathological features overlapping with AITL at one end of the spectrum or ALCL at the other end.

The frequency of bone marrow involvement in PTCL, unspecified varies from study to study, possibly as a consequence of variations in diagnostic criteria. Overall, one-third of the PTCL, unspecified cases show evidence of bone marrow infiltration, which is best identified in biopsies rather than aspirates (Caulet et al, 1990; Chott et al, 1990; Gisselbrecht et al, 1998; Lopez-Guillermo et al, 1998). The histological pattern of bone marrow involvement, similar to extramedullary tissues, is quite variable but cytology frequently mimics the diagnostic specimen (Hanson et al, 1986; Gaulard et al, 1991). The patterns of involvement include varying intensities of interstitial infiltration, predominantly intrasinusoidal infiltration, and diffuse or nodular infiltration. None of these patterns are specific to PTCL, unspecified and can be seen other lymphoma subtypes.

Immunohistochemistry is very helpful in identifying subtle infiltrates and should be performed when possible (White et al, 1989; Gaulard et al, 1991). The phenotype of the bone marrow infiltrate is often identical to the extramedullary tumour. As most PTCL, unspecified show an aberrant T-cell phenotype, most frequently the absence of one or more pan-T-cell antigens, this could be used to confirm the neoplastic nature of the marrow infiltrate.

T-cell granular lymphocytic leukaemia

T-cell granular lymphocytic leukaemia (T-LGL) is an indolent lymphoproliferative disorder characterized by the presence of an immunophenotypically distinct, clonal cytotoxic T-cell population in the peripheral blood. The clinical manifestations of T-LGL are subtle and are usually related to the disease-associated neutropenia and/or anaemia (Loughran, 1993). In Wright–Giemsa stained peripheral blood smears, the neoplastic cells of T-LGL typically have small nuclei and abundant amphophilic cytoplasm containing variable numbers of azurophilic granules and therefore cannot be reliably distinguished from their non-neoplastic counterparts.

Flow cytometric immunophenotyping studies are extremely helpful in identifying and characterizing T-LGL cells and thereby establishing the diagnosis. Distinguishing flow cytometric features include the presence of a CD8+ T-cell population, frequently with abnormally diminished expression of CD5 and CD7, and with aberrant co-expression of a number of NK-cell-associated antigens including CD16, CD57 and CD94. Approximately 50% of T-LGL cases express NK-cell-associated killing inhibitory receptor (KIR) antigens, which include CD158a, CD158b and CD158e. In KIR-positive T-LGL cases, a single KIR antigen is typically expressed by all the neoplastic cells. This restricted pattern of KIR expression provides indirect evidence of clonality (Hoffmann et al, 2000; Morice et al, 2003). T-cell clonality can also be directly demonstrated either through the documentation of clonal patterns of T-cell receptor (TCR) beta chain variable region expression by flow cytometry or by molecular genetic (PCR or Southern blot) analysis of the T-cell antigen receptor genes (Langerak et al, 2001; Morice et al, 2004).

As in peripheral blood specimens, the neoplastic cells of T-LGL cannot be reliably identified in bone marrow aspirates and biopsy specimens by morphological assessment alone. In bone marrow aspirates, the cytoplasm of the granular lymphocytes becomes contracted, which interferes with the visualization of cytoplasmic granules. Furthermore, the distinctive cytological features of cytotoxic lymphocytes are not readily apparent in H & E stained bone marrow biopsy specimens. Given the lack of distinguishing morphological features, early descriptions of bone marrow involvement by T-LGL focused on the presence of identifiable interstitial lymphoid aggregates. However, the advent of antibody reagents that enable the identification of cytotoxic lymphocytes in paraffin-embedded tissues have proved invaluable in elucidating disease-specific patterns of marrow infiltration by this disorder (Fig 3). Immunoperoxidase staining of bone marrow specimens from T-LGL patients with antibodies to CD3, CD8, and the cytotoxic granule proteins TIA-1 and granzyme B frequently reveals distinctive interstitial clusters of cells positive for one or more of these antigens (Fig 3). In addition, careful review also commonly reveals linear staining of antigen positive cells because of their accumulation in marrow sinusoids and microvascular structures. At least one of these immunoperoxidase features (interstitial clusters or intravascular staining) is present in the majority of T-LGL cases (>80%). These stains also demonstrate that the neoplastic cells of T-LGL are present in subtle interstitial foci that are not readily apparent in H & E stained slides and that they are not associated with the discrete lymphoid aggregates that may be seen in some T-LGL marrows but do not represent a disease-specific feature (Fig 3). Moreover, while increased numbers of CD3+ cytotoxic T cells may be seen by immunoperoxidase stains in reactive conditions associated with numbers of granular lymphocytes, the distinctive interstitial clusters and intravascular staining are not present in these cases (Morice et al, 2002). Hence, bone marrow immunohistochemistry using these antibodies provides both a sensitive and specific tool for establishing a diagnosis of T-LGL.

Figure 3.

Bone marrow involvement in T-LGL. (A) Non-neoplastic interstitial lymphoid aggregate in a T-LGL case (haematoxylin eosin, H & E). (B) and (C) The aggregate contains a mixture of CD3+ T cells (B) and CD20+ B cells (C) (immunoperoxidase). (D) The T cells in the lymphoid aggregate are TIA-1-negative. However intravascular staining of TIA-1-positive cells is present in vessels at the periphery of the aggregate as well as in one transecting the aggregate (immunoperoxidase). (E) In some cases, T-LGL cells with round nucleus and a clear cytoplasmic halo can be seen within the vessels (H & E). (F) and (G) These are intravascular lymphoid cells are positive for CD3 (F) and CD8 (G) (immunoperoxidase). (H–J) CD8 staining also highlight interstitial infiltration by LGL (immunoperoxidase). (I) and (J) The T-LGL cells are typically positive for cytotoxic granule marker TIA-1 (I) and granzyme B (J) (immunoperoxidase). Original magnification: A–D ×200; E–I ×650, J ×400.

Hepatosplenic T-cell lymphoma

Hepatosplenic T-cell lymphoma (HSTCL) is a rare, aggressive malignancy of cytotoxic T cells that involves the splenic red pulp, the hepatic sinusoids, and the bone marrow sinusoids and interstitium. This disorder typically affects males in late adolescence and early adulthood; frequent clinical manifestations include hepatosplenomegaly, thrombocytopenia and anaemia, and B symptoms (fever, night sweats, weight loss). HSTCL has a dire prognosis, even in those patients which initially respond to treatment, with a median survival of <1 year (Farcet et al, 1990; Cooke et al, 1996; Weidmann, 2000; Belhadj et al, 2003).

Typically in HSTCL there are relatively few circulating malignant cells, however the peripheral blood disease burden may increase with disease progression. When identified in Wright–Giemsa stained peripheral blood smears the neoplastic cells usually have intermediate to large-sized, variably irregular nuclei and moderate amounts of basophilic cytoplasm that may or may not contain azurophilic granules. In contrast to most other T-cell malignancies, HSTCL frequently expresses gamma-delta TCR heterodimers (Farcet et al, 1990; Cooke et al, 1996). This is not exclusively a disorder of gamma-delta T cells, however, as a subset of cases expressing alpha-beta TCR heterodimers have also been described (Macon et al, 2001). Gamma-delta positive HSTCLs have immunophenotypic features commonly found in non-neoplastic gamma-delta T cells, including the absence of CD4 and CD8 expression (double-negative T cells) and absence of CD5 expression (Haas et al, 1993). Absence of CD5 expression is also common in alpha-beta positive HSTCL and both subtypes of HSTCL frequently show loss of CD7 expression and co-expression of NK-cell-associated antigens such as CD16 and CD56. Clonal rearrangement of either the TCR gamma chain or beta chain genes can be detected in virtually all the gamma-delta and alpha-beta HSTCLs, respectively. However, the presence of these gene rearrangements does not define the immunophenotypic subtype (alpha-beta versus gamma-delta), as gamma chain gene rearrangements are present in 75% of alpha-beta HSTCLs and conversely beta chain gene rearrangements are present in 62% of gamma-delta HSTCLs (Przybylski et al, 2000; Macon et al, 2001). A clonal isochromosome 7q abnormality (either in isolation or associated with other clonal cytogenetic abnormalities including trisomy 8) can be identified by karyotypic analysis in many HSTCL cases (Cooke et al, 1996; Weidmann, 2000; Macon et al, 2001; Belhadj et al, 2003).

Bone marrow involvement by HSTCL is present in most, if not all, cases (Vega et al, 2001; Belhadj et al, 2003). The cytologic features of the majority of the neoplastic cells are similar to those in the peripheral blood, although a subset of the cells may have larger nuclei with more immature-appearing, reticulated chromatin (Fig 4). In H & E stained sections, the neoplastic cells most often have medium-sized rounded or slightly irregular nuclei with slightly dispersed chromatin and small, distinct nucleoli. These cells infiltrate the marrow in an intrasinusoidal and subtle interstitial pattern that, like T-LGL, can be very difficult to recognize in H & E stained biopsy sections. As in T-LGL, immunoperoxidase stains using antibodies to T-cell-restricted intracellular antigen 1 (TIA-1), which is universally expressed by HSTCL, and T-cell-associated antigens, such as CD3, are of great utility in revealing the presence of neoplastic infiltrates that may be missed by routine morphological evaluation (Cooke et al, 1996; Belhadj et al, 2003). Interestingly, HSTCL usually is granzyme B-negative; this TIA-1-positive, granzyme B-negative immunophenotype putatively reflects the unprimed or non-activated cytotoxic phenotype of the neoplastic cells (Fig 4). With disease progression, increased numbers of large atypical cells become evident in the bone marrow aspirate and biopsy and interstitial marrow infiltration becomes a more prominent histological feature, allowing for more ready recognition of the marrow involvement on morphological review (Vega et al, 2001; Belhadj et al, 2003). However, even in this setting, the degree of infiltration as demonstrated by immunoperoxidase staining may exceed that expected from the H & E stained biopsy slides.

Figure 4.

Bone marrow involvement in HSTCL. (A) and (B) The marrow appears hypercellular without a discrete lymphoid infiltrate (A). However, at higher magnification, subtle interstitial infiltration by lymphoid cells with irregular nuclei and pale staining eosinophilic cytoplasm is evident (B) (haematoxylin eosin, H & E). (C) and (D) In many cases tumour cells can be readily identified within the bone marrow sinusoids (H & E). (E) CD3 staining often shows the presence of an interstitial and intersinusoidal CD3 tumours cells (immunoperoxidase). (F) In this αβ subtype of HSTCL, CD3+ tumour cells are restricted to the sinusoids (immunoperoxidase). (G) and (H) The tumour cells typically express cytotoxic granule marker TIA-1 (G) but not granzyme B (H) (immunoperoxidase). Original magnification: A, E, F ×400; B, C ×1000; D, G, H ×650.

T-cell prolymphocytic leukaemia

T-cell prolymphocytic leukaemia (T-PLL) is an uncommon, bone marrow-based lymphoproliferative disorder of mature, post-thymic T cells. T-PLL represents <5% of chronic lymphocytic leukaemias and has been associated with Ataxia Telangectasia (Simpkins et al, 1985; Matutes et al, 1991; Taylor et al, 1996). This disease most often affects older adults (median age >55 years) without a sex predilection; typical presenting features include marked lymphocytosis (frequently exceeding 100 × 109/l), anaemia and thrombocytopenia, and splenomegaly. Hepatomegaly, generalized lymphadenopathy, and localized rash secondary to cutaneous involvement may also be present. T-PLL is aggressive, with a median survival of <1 year when treated with conventional chemotherapy. A minority of patients may experience a more indolent clinical course, and the disease may respond to treatment with anti-CD52 monoclonal antibodies (Dearden et al, 2001, 2002; Pawson et al, 1997).

In the majority of T-PLL cases (approximately 80%), the peripheral blood lymphocytes have the cytological features of prolymphocytes with high nucleus:cytoplasm ratios and intermediate-sized nuclei having moderately condensed chromatin and a single, prominent, central nucleolus. In the remaining cases, the cells have the appearance of mature lymphocytes in Wright–Giemsa stained slides with scant agranular cytoplasm, small nuclei, condensed chromatin, inconspicuous nucleoli and varying degrees of nuclear irregularity (Matutes et al, 1986, 1991; Hoyer et al, 1995). A minority of these cases in both groups have markedly convoluted, Sezary cell-like nuclei and cytoplasmic protrusions or blebs may also be present (Brito-Babapulle et al, 1997). The most appropriate nomenclature for the cases with mature lymphocyte cytology has remained somewhat controversial as some authors argue that the moniker ‘T-cell chronic lymphocytic leukaemia’ is most appropriate whereas others refer to these cases as ‘small cell variants’ of T-PLL (Matutes et al, 1991; Hoyer et al, 1995). Except for possibly lesser degrees of hepatosplenomegaly in cases with small cell cytological features, the clinical and laboratory features appear to be similar in both groups and in both the disease is aggressive and the prognosis is poor (Matutes & Catovsky, 1996). Therefore these entities appear to represent varying morphological types of a single common disease entity, regardless of the nomenclature applied.

The T-PLL has immunophenotypic features of mature T cells by flow cytometric analysis with expression of CD2, CD3, CD5, and CD7 and absence of T-precursor-associated antigens, such as CD1a, CD34 and TdT. In most cases no aberrancies of T-antigen expression are present and a minority of cases may show diminished expression of CD3. The T-PLL cells are most often (>60%) CD4+ and CD8, although CD4 and CD8 double positive and CD8+, CD4 cases also occur (Matutes et al, 1991; Hoyer et al, 1995). Even in those cases that are CD8+, neither NK-associated antigens, such as CD16 and CD56, nor cytotoxic granule proteins, such as TIA-1, and granzyme B are expressed. T-cell clonality can be demonstrated by flow cytometric analysis of TCR beta chain variable region expression and by molecular genetic analysis of TCR gene rearrangements in all cases (Langerak et al, 2001; Morice et al, 2004). Approximately 80–90% of T-PLL cases contain a clonal cytogenetic abnormality involving juxtaposition of the TCL1 and TCL1b loci on chromosome 14q32.1 and the TCR α/δ on chromosome 14q11 (Brito-Babapulle et al, 1987; Brito-Babapulle & Catovsky, 1991). This fusion can result either from a chromosome 14 inversion or reciprocal translocation and leads to constitutive overexpression of the TCL1 oncogene (Pekarsky et al, 2001). These chromosome 14 abnormalities are frequently associated with an isochromosome 8q. The Ataxia Telangectasia Mutated (ATM) gene on chromosome 11q23 is abnormal in approximately 40–50% of cases and this probably plays a role in disease pathogenesis. However most ATM gene abnormalities in T-PLL are detectable only by molecular genetic and/or fluorescence in situ hybridization (FISH) analysis (Yuille et al, 1998).

Bone marrow involvement is a universal feature of T-PLL (Nieto et al, 1989; Hoyer et al, 1995). In bone marrow aspirates the cytological features of the T-PLL cells are similar to those in the peripheral blood (Fig 5). The pattern of bone marrow biopsy infiltration by T-PLL varies, ranging from cases with nodular and interstitial lymphoid infiltrates that comprise a minority of the marrow cellularity to those with diffuse marrow involvement and effacement of the normal marrow architecture. While bone marrow biopsy involvement by T-PLL is usually readily identified by morphological evaluation, the cytological features of the neoplastic infiltrates do not distinguish them from other chronic lymphoproliferative disorders. Therefore, immunoperoxidase studies using antibodies to T-cell and B-cell-associated antigens are essential in establishing the infiltrates to be of T-cell lineage where prior immunophenotyping analysis has not been performed. Recently TCL1 overexpression has been demonstrated in 70% of the T-PLL cases but not other T-cell lymphoproliferative disorders (Herling et al, 2004). As anti-TCL1 antibodies that are reactive in paraffin-embedded material are available, it is likely that staining for TCL1 in bone marrow biopsies will prove to be useful in the interpretation of bone marrow involvement. The degree or pattern of marrow involvement by T-PLL has not been found to be of prognostic significance, although there is relatively little data in this regard.

Figure 5.

Blood and bone marrow involvement in T-cell prolymphocytic leukaemia (T-PLL). (A) and (B) Peripheral blood smears from a case of T-PLL (A) and small cell variant T-PLL (B). In the T-PLL a single, pale staining central nucleolus is present. In comparison, the small cell variant has smaller nuclei, more condensed chromatin, and smaller less prominent nucleoli (Wright–Giemsa). (C) and (D) Bone marrow biopsy shows a subtle interstitial infiltration by neoplastic lymphoid cells, with small to intermediate-sized, variably irregular nuclei and open chromatin (haematoxylin eosin). (E) The neoplastic infiltrates are CD3+. The staining also highlights the nuclear irregularity of some of the malignant lymphoid cells (immunoperoxidase). Original magnification: A, B, D ×1000, C, E ×400.

Cutaneous T-cell lymphoma (mycosis fungoides and Sezary syndrome)

Cutaneous T-cell lymphoma are a somewhat heterogeneous group of lymphoproliferative disorders, the majority of which fall into the clinicopathological entities of mycosis fungoides (MF) and Sezary's syndrome (SS) (Willemze et al, 1997). MF is an indolent disorder characterized early in the disease course by localized patches that evolve slowly over time into plaques and, ultimately, tumour nodules. Lymph nodes and other visceral organs may be involved, usually later in the disease course. In contrast, SS is a clinically aggressive disorder, classically defined by the triad of erythroderma (usually diffuse), widespread lymphadenopathy, and atypical cells with cerebriform nuclei (Sezary cells) in the peripheral blood. MF and SS have similar histopathological and immunophenotypic features, with epidermotropic lymphoid infiltrates comprised of cells that usually have small- to intermediate-sized cerebriform nuclei, which are mature CD4+ T cells by immunophenotyping analysis, frequently with diminished expression of CD7 (Sentis et al, 1986; Nickoloff, 1988; Bogen et al, 1996; Kamarashev et al, 1998). Other observed immunophenotypic abnormalities in CTCL include diminished expression of CD2 and CD3 (Edelman & Meyerson, 2000; Morice et al, 2004). T-cell clonality can be demonstrated by molecular genetic TCR analysis in most cases; clonality can also be documented by TCR V-beta flow cytometry in those with peripheral blood involvement (Schwab et al, 2002).

There is a paucity of published information regarding patterns of marrow involvement by CTCL. This is, in part, the result of the fact that bone marrow examination is not a standard element of CTCL staging (Lamberg & Bunn, 1979; Lamberg et al, 1984). The largest published series indicate that CTCL involves the marrow in approximately 15–20% of cases and may be the only extracutaneous site involved (Salhany et al, 1989; Graham et al, 1993; Marti et al, 1996; Sibaud et al, 2003). In most of the reported cases the involved marrows were normocellular and contained nodular and/or subtle interstitial lymphoid infiltrates, although in isolated cases there was dense marrow infiltration with effacement of normal marrow architecture (Fig 6). Bone marrow eosinophilia was described as a common secondary feature. The conclusions of these studies are somewhat limited, however, by the fact that most were performed prior to the widespread availability of antibodies to T-cell-associated antigens reactive in paraffin-embedded tissues (Salhany et al, 1989; Graham et al, 1993). For this reason they focused on the presence or absence of cytological atypia in the marrow lymphoid cells as a discriminating feature between benign, reactive lymph aggregates and neoplastic lymphoid aggregates/infiltrates. In these studies, morphological evidence of marrow involvement by CTCL was more frequent with advanced clinical stage (widespread rash) and was associated with adverse prognosis, although this was not an independently predictive factor when multivariate analysis including degree of skin involvement was performed (Bunn et al, 1980; Graham et al, 1993).

Figure 6.

Blood and bone marrow involvement in Sezary syndrome/CTCL. (A) Atypical lymphoid cells with intermediate to large sized nuclei and markedly convoluted nuclear membranes are present in the peripheral blood smear (Wright–Giemsa). (B) Bone marrow biopsy from a Sezary syndrome case with marrow involvement. Focal interstitial infiltration by atypical, cerebriform lymphocytes is present (arrow) (haematoxylin eosin). (C) CD3 staining greatly facilitates detection of marrow involvement and highlights the marked nuclear irregularity (arrow) (immunoperoxidase). Original magnification: A, B, C ×1000.

NK-cell neoplasms

True NK-cell neoplasms in the current WHO classification include extranodal NK/T-cell lymphoma of nasal type (which may also occur outside the nasal cavity) and a rare, biologically linked but clinically distinct entity, aggressive NK-cell leukaemia (Jaffe et al, 2001). A third entity, referred to as ‘blastic NK-cell lymphoma’ in the WHO classification, is now considered to be not a neoplasm of NK/T cells and will not be discussed further (Chaperot et al, 2001; Feuillard et al, 2002; Jacob et al, 2003). These neoplasms show a limited geographical distribution, with the vast majority of cases being reported from South-East Asia and South America. The tumours typically arise in the nasal cavity and other mucosal sites of the upper aerodigestive tract. Virtually all cases show evidence of clonal EBV infection and the clinical behaviour is aggressive in most cases (Kwong et al, 1997; Cheung et al, 2003). Histologically, the lymphoma shows a broad cytological spectrum, but often there is diffuse infiltration by medium-sized atypical lymphoid cells with a high rate of apoptosis and mitosis. Typically the tumour cells infiltrate vessels and there is often widespread necrosis. The tumour cells are usually positive for CD56, and cytoplasmic CD3, but not surface CD3. This discrepancy in CD3 immunoreactivity is probably because of the specificity of the antibody reagents used. Those reactive to cytoplasmic CD3 usually bind a subunit of the CD3 complex that is expressed by normal NK cells. In contrast, those used to detect surface CD3 expression usually recognize the fully assembled CD3 complex, which NK cells lack as their TCR genes in these cells are in the germline (un-rearranged) configuration. Bone marrow infiltration is seen in the leukaemic variant but is reported to be rare in lymphomatous variants (Wong et al, 2001; Sung & Ko, 2004). When the marrow is involved, the infiltration may be subtle and immunohistochemistry for CD56 and, more importantly, in situ hybridization for EBV are helpful in identifying tumour cells.

Aggressive NK-cell leukaemia must be distinguished from T-cell large granular lymphocyte leukaemia and indolent NK-cell lymphoproliferative disorder (also termed chronic NK-cell lymphocytosis), both of which are clinically indolent (Cheung et al, 2003).

Molecular and cytogenetics tests for diagnosis of PTCL in paraffin-embedded bone marrow biopsies

Most of the molecular tests that are available for the diagnosis of PTCL in paraffin-embedded extramedullary sites, such as PCR, reverse transcription (RT)-PCR or interphase FISH, are also readily applicable to the bone marrow (Hanamura et al, 1999; Gebhard et al, 2001; Brown et al, 2002; Liu et al, 2002; Cotelingam, 2003). The most helpful molecular tests are the PCR-based clonality analyses targeting TCR gene rearrangements (Crotty et al, 1998; Attygalle et al, 2004). The DNA quality and the efficiency of amplification depend largely on the type and length of fixation and decalcification methods and is usually lower than non-decalcified paraffin-embedded biopsies. In staging biopsies, the demonstration of a monoclonal T-cell population by PCR in the marrow is particularly helpful if the presence of same dominant clone has been shown in the diagnostic biopsy. In contrast, caution is required if the primary diagnosis of PTCL is based on the molecular clonality analysis performed solely on the marrow biopsy. There may be false positive results because of small numbers of T cells present in the sample or reactive oligoclonal expansion of T cells.

Histological mimics of PTCL in marrow biopsies

A number of benign and malignant conditions of the marrow can be associated with dense T-cell infiltrates that can mimic PTCL involvement. The differential diagnosis on histology alone may be impossible in the vast majority of the cases and immunohistochemistry would be of limited help. Correct diagnosis often requires the integration of clinical features, peripheral blood findings, bone marrow aspirate cytology flow cytometric immunophenotyping, cytogenetic and molecular analysis.

Benign mimics include autoimmune diseases, such as rheumatoid arthritis and polymyalgia, a variety of viral infections, such as infectious mononucleosis, cytomegalovirus and human immunodeficiency virus/acquired immunodeficiency syndrome, bacterial infections, such as tuberculosis, bone marrow or peripheral stem cell transplantation.

The neoplastic conditions that are associated with non-neoplastic T-cell infiltrates in the marrow include small B-cell neoplasms, classical Hodgkin's lymphoma, nodular lymphocyte predominant Hodgkin's lymphoma, T-cell rich large B-cell lymphoma (Dogan et al, 2003b) and the myelodysplastic syndromes.