The evaluation of bone marrow (BM) status is a key process in the initial workup of patients diagnosed with lymphoma. Despite its importance, the complexity of the BM evaluation process has sometimes caused institutional heterogeneity and may be an obstacle in multi-center clinical trials involving patients with lymphoma.

In September 2011, the Korean Society of Hematology Lymphoma Working Party held a nationwide conference to establish a consensus for assessing BM involvement in patients with lymphoma. At this conference, many clinicians, hematopathologists, and diagnostic hematologists discussed various topics for a uniform consensus in the evaluation process to determine whether the BM is involved. It is now considered to have become sufficiently mature to be published. Hence, in this report, we describe the consensus reached and the limitations in current methods for assessing BM involvement in patients with lymphoma.

The BM is the most common site of extranodal involvement in lymphoid malignancies and the frequency of BM involvement varies according to the lymphoma subtype (Table 1). Compared to aggressive lymphoma, relatively high frequencies have been reported in indolent lymphomas, such as mantle cell lymphoma (MCL), follicular lymphoma (FL), and marginal-zone lymphoma [1]. Because BM involvement of lymphoma is definite evidence of disseminated disease [2,3], the assessment of BM status in patients with lymphoma provides important information for decisions regarding treatment. In particular, in the subtypes of lymphoma in which local treatment modalities, such as radiotherapy and concurrent chemoradiation, have a role in definitive therapy, knowledge about the presence or absence of BM involvement is vital. Hodgkin lymphoma, FL, mucosa-associated lymphoid tissue (MALT) lymphoma, and extranodal NK/T cell lymphoma, nasal type, are examples [4,5].

Currently, noninvasive clinical staging by computed tomography (CT), magnetic resonance imaging (MRI), or positron emission tomography (PET)-CT cannot fully assess BM involvement, which is possible only by pathological staging (aspiration and trephine biopsy). Thus, although clinical stage can suggest localized diseases (stage I or II), it is not uncommon for the final stage to change to stage IV due to positive BM involvement. For example, recent studies that have analyzed patients with MALT lymphoma have demonstrated that a considerable proportion of patients (10% to 44%) shows BM involvement despite single MALT organ disease [6-8].

Determining BM status is important for the prognosis of patients. The International Prognostic Index (IPI) consists of five clinical variables: age, serum levels of lactate dehydrogenase (LDH), performance status (PS), stage, and extranodal involvement. Of these, BM status can influence both stage and extranodal involvement. If a patient of clinical stage IE (e.g., gastric lymphoma) has only one positive variable among age, serum LDH, and PS, the prognosis may depend mainly on the BM status. With no BM involvement, the IPI score would be 1 and this patient would be in the low-risk group. In contrast, positive BM involvement makes the IPI score 3, placing the patient in the intermediate-high risk group. According to the IPI system, the low-risk group shows a response rate of 87%, 70% 5-year relapse-free survival, and 73% 5-year overall survival. However, the intermediate-high risk group shows a response rate of 55%, 49% 5-year relapse-free survival, and 43% 5-year overall survival [9]. A recent trial suggested that frontline autologous stem cell transplantation may provide clinical benefits to high-risk patients with aggressive non-Hodgkin lymphoma (NHL) compared to low-risk patients [10,11].

The pattern of infiltration is a key morphological feature to consider when generating a logical list of differential diagnostic considerations. These infiltration patterns, although not absolutely specific, may provide diagnostic clues both for the discrimination of benign and malignant infiltrates and for NHL subtyping. However, patterns can change during the course of the disease, and mixed patterns frequently occur. Five major patterns of BM infiltration can be discerned (Table 2) [12].

Nevertheless, the morphological features of lymphoma in BM involvement vary widely and frequently differ from these typical patterns. Particularly in cases with individual infiltration of lymphoma cells, it can be difficult to recognize the involvement of lymphoma in BM. Moreover, the histological features of BM involvement are sometimes discordant with those of extramedullary lesions. Thus, in addition to a morphological assessment, immunohistochemical studies are required for a complete diagnostic approach. Features of the morphological, immunohistochemical, and diagnostic clues are summarized in Tables 3 (B-NHL) and 4 (T-NHL) [12].

Flow cytometry is useful and may be essential for the diagnosis of leukemic manifestations of a lymphoma. The leukemic phase indicates lymphoma cells ≥ 25% among nucleated cells in the peripheral blood or BM aspirate [25]. Usually, lymphoma cells are gated as CD-45^bright/side-scattered light (SSC)^low by flow cytometry, but immature lymphoid cells are CD45^intermediate or negative, TdT⁺, and CD34⁺. The lineage markers on lymphocytes are CD19 for the B cell lineage, CD3 for the T cell lineage, and CD56 for the NK cell lineage. According to European LeukemiaNet, the B-oriented panel includes CD19⁺ cells with other markers such as CD5, CD20, CD22, CD23, CD103, CD10, κ and λ light chains, surface Ig, CD25, CD79b, and CD38; the T-oriented panel includes CD3+ cells with other markers such as CD2, CD4, CD5, CD8, and CD7; and the NK-oriented panel includes CD56⁺ cells with other markers such as CD16, CD57, CD45, CD158, CD94, and cytotoxic granule-associated proteins (TIA-1, granzyme-B, perforin) as an activated cytotoxic phenotype [26]. Analyses of leukemic BM involvement in Korean patients with NHL have shown that the leukemic phase is associated with a poor prognosis, such as low complete remission, low 5-year survival rate, and a high death rate [27].

Recently, EuroFlow suggested eight-colored antibody panels for standardized n-dimensional flow cytometric immunophenotyping of normal, reactive, and malignant leukocytes. The lymphoid screening tube detects aberrant mature lymphocyte populations of B, T, and NK lineages, but this eight-color tube does not allow the precise diagnosis and classification of the aberrant lymphocyte populations detected. This needs further characterization with plasma cell dyscrasia, B-chronic lymphoproliferative disease (B-CLPD), T-CLPD, and NK-CLPD panels [28].

In February 2006, the Korean Society of Diagnostic Hematology recommended a minimum antibody panel for the diagnosis and classification of CLPDs [29]. Recently, this recommendation was modified, considering recent progress in the flow cytometric analysis of lymphoid malignancies (Table 6). After gating on CD45^bright/SSC^low, combinations of κ/λ/CD20 (or CD19)/CD45, CD5/CD10/CD20 (CD19)/CD45, or CD4/CD8/CD3 (CD7)/CD45 are useful for the interpretation of flow cytometric results. For example, because a mature B cell lymphoma shows immunoglobulin light chain restriction, the antibody combination of κ/λ/CD19 (or CD20)/CD45 is useful. To see aberrant antigen expression, the combination of CD5/CD10/CD19 (or CD20)/CD45 is also useful. In the panel of CD5/CD10/CD19 (or CD20)/CD45, after the serial gating of CD45^+bright/low and CD19 (or CD20)⁺ gating, the flow cytometric cytograms of CD5⁺ CD10^–, CD5^– CD10⁺, CD5⁺ CD10⁺, and CD5^– CD10^– patterns indicate specific subtypes of mature B cell lymphomas. Mature T&NK cell lymphomas frequently show aberrant T cell antigen expression, such as a lack of pan-T antigens (CD5, CD7). To identify the restricted populations of T cells, the antibody combination of CD4/CD8/CD3/CD45 is used, and serial gating of CD45^+bright/low and CD3⁺ identify specific subtypes of mature T&NK cell lymphomas such as CD4⁺ CD8^–, CD4^– CD8⁺, CD4⁺ CD8⁺, and CD4^– CD8^– patterns. The immunophenotypes of the specific subtypes of lymphoid malignancies are summarized in Tables 7 and 8.

Cytogenetic analyses can provide important information for discriminating lymphomatous BM and predicting prognoses in certain types of lymphoma. For example, del(11q) or del(17p) in CLL indicate a worse prognosis, and thus aggressive treatment such as allogeneic stem cell transplantation should be considered early in the course of therapy. Fluorescence in situ hybridization (FISH) is particularly useful for detecting clonal lymphoma cells in BM. Types of probes for FISH that are validated and of clinical importance are summarized in Table 9.

Currently, a pathological examination does not assess the whole BM and the interpretation of a BM specimen usually requires a highly experienced hematopathologist. Although the revised response criteria for malignant lymphoma, published in 2007, indicates that uninvolved BM should be normal histologically with a small (< 2%) clonal B cell population detected by flow cytometry [30], there are still some diagnostic challenges.

First, the differential diagnosis of reactive lymphoid infiltrates from nodular infiltrates of low-grade lymphomas remains a common problem. Although some diagnostic clues based on morphology have been suggested [16,31], some cases may remain unresolved. To clarify this uncertainty, molecular studies using DNA- or RNA-based polymerase chain reaction should be considered. Although this technique still does not cover the whole range of problems, it should be considered whether some proven tests, such as immunoglobulin heavy chain gene rearrangement analysis and T cell receptor analysis, should be involved in the standard. The next issue associated with BM morphology concerns the clinical meaning of concordance with the morphology of the lymphoma cells in the lymph node. In patients with DLBCL, some have shown infiltrates of small B cell lymphoma cell in BM, instead of large B cells (discordant involvement). Under this condition, it is presumed that DLBCL developed as a transformation from an occult small B cell lymphoma, or alternatively, two unrelated lymphomas are present [21]. Recently, it has been reported that in multivariate analyses, including individual factors of the IPI system, concordant BM involvement has an independently negative impact on prognosis whereas discordant BM involvement does not [32]. If this can be validated in a prospective cohort, the morphology of lymphoma cells in BM involvement should be considered when predicting prognosis in patients with DLBCL.

Another issue is flow cytometry analyses. As mentioned above, flow cytometry is essential when assessing BM involvement. Current guidelines regard a small clonal population (< 2%) as uninvolved BM [30]. However, to the best of our knowledge, no clinical study has addressed this issue. Moreover, recent advances in flow cytometry technology have caused a wide range of variation in the sensitivity of flow cytometric analyses among institutions. Thus, it is desirable to document the minimal criteria for flow cytometry technology in a standard guideline. The last issue in flow cytometry is the meaning of discordance between a bone marrow biopsy (BMB) and flow cytometry. Recently, in a study that included 757 NHL patients, there was considerable discordance between BMB and flow cytometry [33]. For example, in FL and LPLs, the discordance rates were up to 22% and 24%, respectively. Based on BMB and flow cytometry results, four subsets can be created in patients with NHL: positive BMB and positive flow cytometry, positive BMB and negative flow cytometry, negative BMB and positive flow cytometry, and negative BMB and negative flow cytometry. To date, no study has analyzed the prognostic significance of each group. It could be valuable to investigate whether there are prognostic differences among these subgroups. Moreover, the relationship between prognosis and the degree of involvement remains unclear. Although some retrospective analyses have revealed significant correlations between degree of BM involvement and prognosis [34-37], this has not been validated. To clarify this issue, it is important to establish uniform criteria for quantifying the degree of BM involvement objectively, so as to minimize differences among interpreting pathologists.

Finally, we discuss whether BM status can be evaluated by clinical staging using CT, MRI, or PET-CT. Because of technical limitations of these imaging modalities, BM assessment is currently possible only by a pathological examination. Because a BM biopsy at multiple sites is impossible in routine clinical practice, hematopathologists and clinicians have accepted results from BM examinations at the iliac crest on the assumption that this area may reflect whole BM status. However, technical advances in imaging modalities have shown that this is not true. A recent prospective study that compared the performance of BMB, whole-body MRI, and PET-CT to evaluate lymphomatous BM invasion demonstrated that noninvasive morphological procedures could be superior to a pathological BM examination [38]. In that study, most patients with abnormal BM findings on MRI or PET-CT showed BM infiltration in areas other than the iliac crest. Recently, resolution and sequence protocols of MRI have made rapid progress, sufficient to detect lymphomatous BM invasion [39]. Usually, MRI is considered to have high sensitivity, but specificity may be suboptimal for assessing lymphomatous BM invasion [40]. Nevertheless, it is expected that ongoing advances in technical aspects will make it possible to assess BM status by imaging without a pathological exam. Moreover, it is considered that PET-MRI, a new, developing imaging modality combining PET and MRI, may be promising for BM evaluations and applicable in the near future [39].

BM involvement in patients with lymphoma provides important information in clinical practice. Despite the limitations of current BM evaluation methods, standardization of the BM evaluation process is important to improve patient care. Based on standardized protocols, discussions about how to define positive lymphomatous BM should be in progress. Furthermore, because recent technological advances in flow cytometry and imaging modalities are expected to impact clinical practice in the near future, interdepartmental collaboration including the areas of medicine, hematopathology, radiology, and nuclear medicine is more important than ever.