Risk Factors and Predisposing Conditions

The sequence of events that results in NHL is unknown. Inherited genetic susceptibility and/or shared environmental exposures is suggested by registry-based studies that demonstrate slightly increased risk of developing NHL in first-degree relatives of patients with lymphoma and patients with monoclonal gammopathy of undetermined significance (MGUS).4,5 No germline mutations have been identified, and the majority of patients with NHL have no familial clustering. Many factors related to the host’s immune status combined with environmental exposures such as organic solvent exposure and wood products correlate with an increased risk of NHL (Box 106-1).^6,7 Conflicting reports exist on the connection between exposure to ultraviolet (UV) radiation, vitamin D levels, and the lifetime risk of NHL.⁸

A number of altered immunologic states greatly increase one’s lifetime risk of developing NHL, including both congenital and acquired immunodeficiency states. Most NHLs that occur in association with immunodeficiency states are B-cell lymphomas, with the exception of the increased risk of T-cell lymphoma seen in ataxia telangiectasia and a small percentage of posttransplant lymphoproliferative disorders (PTLDs), which are of T-cell/natural killer (NK)-cell origin. A common theme in these lymphomas is lost or impaired cellular immunity (particularly T cell) that allows for unregulated B-cell proliferation in lymphoid tissues.

Rare, extranodal T-cell lymphomas are also encountered in the setting of immunologic defects. Enteropathy-associated T-cell lymphoma (EATCL) often occurs in the setting of celiac sprue, and hepatosplenic T-cell lymphoma (HSTCL) occurs in young men with a history of solid-organ transplantation or other immune defects.⁹ The current World Health Organization (WHO) classification of lymphomas recognizes a subset of diffuse large B-cell lymphomas (DLBCL) that arise as a consequence of chronic inflammation and are frequently associated with Epstein-Barr virus (EBV).¹⁰ The classic example is pyothorax-associated lymphoma, which was first reported in 1987 in patients treated for tuberculosis by artificial pneumothorax. Other cases of DLBCL may occur in the setting of chronic inflammation, such as chronic skin ulcers or osteomyelitis.¹¹

The role of chronic immune stimulation is less ambiguous in infection-associated lymphomas, such as in lymphoma of mucosa-associated lymphoid tissue (MALT) in which the lymphoproliferation is frequently antigen driven.12–15 Evidence for the essential role of the chronic immune stimulation is that eradication of the underlying infection, such as chronic hepatitis C or Helicobacter pylori, can result in remission in many cases.^18–18 Viruses may also function as co-factors in lymphomagenesis, whereby they may exert their effect through genomic integration, which leads to alterations in gene expression, and/or by directly affecting cellular proliferation. EBV, human T-cell leukemia/lymphoma virus type I (HTLV-I), and human herpesvirus–8 (HHV-8), in particular, have well-established oncogenic roles in subtypes of NHL.^19–22 In the case of EBV, it appears to have a direct oncogenic role in lymphomas of immunocompromised patients, such as those infected with HIV and in the posttransplant period.¹¹ HTLV-I has a direct role in adult T-cell leukemia/lymphoma (ATLL), but carriers have only a 2% to 5% lifetime risk of developing disease with a latency period of 30 to 40 years. HHV-8, also known as Kaposi sarcoma herpes virus (KSHV), has been associated with primary effusion lymphoma (PEL), which is a rare B-cell lymphoma that occurs primarily in highly immunosuppressed patients with AIDS who are often co-infected with EBV.²³

Classification of Lymphomas

Lymphomas are a heterogeneous group of diseases with a variety of natural histories. Histologic classification schemes have been developed to organize lymphomas into groups with shared pathogenesis and clinical behavior with an aim to guide treatment. The classification systems for lymphomas have changed frequently and dramatically since they were first introduced in the 1950s and continue to evolve with technologic advances and scientific discovery. These classifications have evolved from exclusively morphologic to the current working system that incorporates immunophenotype and hallmark genetic abnormalities. The most recent version of the WHO classification system in 2008 places great emphasis on molecular and cytogenetic abnormalities in conjunction with clinical variables to define individual subtypes (Box 106-2).

History of Lymphoma Classification Systems

In 1956, Henry Rappaport of the U.S. Armed Forces Institute of Pathology proposed a simple system of lymphomas based on the growth pattern of the disease (nodular vs. diffuse) as well as the predominant cell’s well-differentiated, poorly differentiated, undifferentiated, or histiocytic appearance. With the use of this system, nodular lymphomas composed of small lymphocytes were generally considered indolent disorders whereas so-called histiocytic lymphomas were aggressive. The Rappaport system was used until the 1970s when the American pathologists L.J. Lukes and R.D. Collins and the German pathologist Karl Lennert each proposed classification systems that reflected advances in cellular immunology, cell morphology, and lymphocyte lineage by dividing entities into B-cell and T-cell disorders on the basis of their cell surface markers.26,27 The immune phenotype proved fundamental for the accurate classification of lymphoma, with early studies showing that most lymphomas are of B-cell origin (Table 106-1), but these studies still did not address clinical concerns and were not uniformly embraced.

In the early 1980s, classifications emerged that incorporated the putative cell of origin or tumor aggressiveness. The Kiel classification incorporated information on the cell of origin and its relationship to normal lymphoid cells, whereas the Working Formulation provided clinical groupings and pathological divisions largely based on cell size (large vs. small), cell shape (round vs. not round), and growth pattern (follicular vs. diffuse). The Working Formulation was the first system to define broad categories of lymphoma based on general clinical prognosis of either low grade, intermediate grade, or high grade to assist the clinician. The system, however, did not include immunophenotype and did not foster recognition of new entities.

In 1994, the International Lymphoma Study group developed a consensus list of diseases that could be recognized by pathologists and was codified in the Revised European–American Classification of Lymphoid Neoplasms (REAL) classification system; this ultimately became the WHO classification system and is the recognized standard today.²⁸ The first WHO classification, published in 2001, defined diseases by four features: morphology, immunophenotype, molecular genetic characteristics, and clinical information,²⁹ whereas the updated 2008 version places greater emphasis on distinctive immunologic and molecular profiles of lymphomas and emphasizes the importance of clinical characteristics.³⁰

Molecular Genetics of Non-Hodgkin Lymphoma

Insight into the genetic features that characterize tumor subtypes has aided in the diagnosis of lymphomas and identification of potential therapeutic targets. A variety of complementary technologies can measure the expression of thousands of genes on a solid platform, termed molecular profiling, and can link biology to a genetic expression signature. The application of gene expression profiling (GEP) to lymphoma has provided insights into unique molecular signatures of distinct types of B-cell malignancies³¹ and can relate lymphoid neoplasms to normal stages in B-cell development and physiology. This approach has provided a new way to classify lymphomas and predict clinical outcome.

Molecular profiling has advanced the understanding of lymphomagenesis through identification of genes important in cellular proliferation, differentiation, and apoptosis.32–36 At a molecular level, genetic lesions identified in lymphomas include oncogene activation or loss of tumor suppressor genes caused by chromosomal translocation, deletion or mutation, or the integration of viral genomes (Table 106-2). Some lymphomas rely on the continuous signaling provided by these oncogenes, termed oncogene addiction, whereas others seem more reliant on signaling pathways not controlled by oncogenes, termed non-oncogene addiction.³⁷ GEP has also helped define the molecular makeup of the nonmalignant cells in tumor specimens, the so-called microenvironment that is important for pathogenesis and may include potential therapeutic targets.^38,39

Gray Zone Lymphomas

Cases exist that cannot be classified with certainty. The term gray zone lymphoma was first introduced in 1998 for cases that shared morphologic and immunophenotypic features of Hodgkin lymphoma and DLBCL.⁴⁰ It has been demonstrated that primary mediastinal B-cell lymphoma (PMBL) has an overlapping gene expression profile with nodular-sclerosing Hodgkin lymphoma (NSHL).^41,42 Given that PMBL and NSHL exhibit clinical and biological overlap, the term mediastinal gray zone lymphoma (MGZL) was first used in 2005 by Traverse-Glehen and colleagues to identify those cases with features intermediate between PMBL and NSHL.⁴³ The term has more than theoretical interest because MGZL has an inferior prognosis with contemporary therapies.^44,45 As such, the 2008 version of the WHO classification system includes a provisional category of B-cell lymphoma, unclassifiable, with features intermediate between diffuse large B-cell lymphoma and classic Hodgkin lymphoma that includes these cases of MGZL.

Similarly, the distinction between high-grade lymphomas such as DLBCL and Burkitt lymphoma (BL) can be challenging on morphologic and immunophenotypic grounds.⁴⁶ The distinction between the two entities is critical, because patients with BL require high-intensity chemotherapy. Two studies have demonstrated that BL has a unique molecular profile and that a subset of cases classified as DLBCL have a BL molecular profile.^47,48 It is now appreciated that between 8% and 10% of cases of newly diagnosed DLBCL will harbor a mutation in the MYC oncogene and have a poor outcome with R-CHOP (rituximab + cyclophosphamide, doxorubicin, vincristine, and prednisone).^49,50 Some cases of MYC+ DLBCL may have dual translocations affecting both MYC and BCL-2, which portends an even worse outcome.^51–54 The WHO 2008 classification includes a provisional category for these cases that is referred to as “B-cell lymphoma, unclassifiable, with features intermediate between diffuse large B-cell lymphoma and Burkitt lymphoma.”

Prognostic Factors for Lymphoma

Clinical variables are powerful and independent predictors of outcome. A number of prognostic indices aid in the prognosis of individual NHL subtypes.64–67 An international project identified that inferior outcome was related to age older than 60 years, stage III or IV disease, serum LDH value above normal range, Eastern Cooperative Oncology Group (ECOG) performance status of 2 or higher, and involvement of two or more extranodal sites. A clinical prognostic model, termed the International Prognostic Index (IPI), was developed with these five factors and stratifies patients into quartiles with differing disease-free survival rate at 5 years (Table 106-5). The IPI is the standard for assessing prognosis in DLBCL, as well as for comparison between clinical trials. Because the IPI was developed before use of rituximab, a revised IPI was suggested, but it was not validated in a larger prospective study.^68,69

Gene expression profiling (GEP) is also an important prognostic tool that demonstrates molecular heterogeneity within tumors. In DLBCL, for example, morphologically indistinguishable tumors show marked heterogeneity in gene expression and these patterns of expression correspond to the cellular origin of the lymphoma according to its stage of differentiation.⁷⁰ DLBCL can thereby be divided into at least three different subtypes: germinal center B-cell (GCB)-like, activated B-cell (ABC)-like, and PMBL that arise by distinct pathogenetic mechanisms and have a very different prognosis.^38,42,71,72 A molecular prognostic model of survival based on signatures of germinal center B cells, proliferating cells, reactive stromal and immune cells in the lymph node, and major histocompatibility complex class II cells, was developed for CHOP-treated DLBCL.⁷¹ Overall survival (OS) was superior in patients with the GCB-type compared with the ABC-type DLBCL, a finding that was independent of the IPI (Fig. 106-2). The distinction between GCB-like and ABC-like subtypes of DLBCL carries prognostic information in patients treated with rituximab-containing regimens.³⁴ Wilson and associates demonstrated that treatment with dose-adjusted etoposide, prednisone, vincristine, cyclophosphamide, doxorubicin, and rituximab (DA-EPOCH-R) resulted in a 5-year progression-free survival rate (PFS) of 100% for GCB-like DLBCL compared with 67% for non–GCB-type DLBCL.⁷³ In relapsed DLBCL, different responses were observed in bortezomib-containing regimens.⁷⁴ The gene expression profiles of follicular lymphoma (FL) and mantle cell lymphoma (MCL) have also identified subgroups with different prognosis on the basis of cellular proliferation⁷⁵ and tumor-infiltrating cells.³⁹

GEP is impractical to perform on all patients with newly diagnosed lymphomas, so multiple immunohistochemical (IHC) models using combinations of antibodies have been developed to predict GCB and non-GCB subtypes of DLBCL.76–80 The concordance rate with GEP is variable, however, ranging between 74% and 93%.⁸¹ The “Tally” algorithm in which results are not analyzed in a specific order demonstrated the best concordance with GEP at 93% (Fig. 106-3). IHC models will require further validation and standardization but have the advantage of being widely available.

MicroRNAs (miRNA) are short noncoding regions of the RNA that regulate gene expression and have been used to discriminate between normal cells and cancer cells.⁸² Recently a 9-miRNA signature was shown to separate cell lines derived from GCB and ABC-like subtypes of DLBCL.⁸³ MiRNA profiling demonstrates promise in predicting outcomes in DLBCL,⁸⁴ FL,⁸⁵ and MCL⁸⁶ and will require further study and validation.

Response Assessment

Accurate assessment of response to therapy is critical, particularly in curable lymphomas. Treatment response should be documented by physical findings, and all abnormal tests should be repeated. End-of-therapy assessment is usually performed 3 to 6 weeks after completion of therapy unless progression is suspected earlier. An International Working Group (IWG) workshop held under the auspices of the National Cancer Institute in 1998 standardized response criteria based on the bi-dimensional measurements of involved nodal groups.⁸⁷ The spleen is considered to be a nodal site and is assessed by CT. Focal lesions in the liver are considered measurable. Patients with bone marrow involvement before initiating therapy are required to be morphologically free of lymphoma to be considered in complete remission.

The response criteria were updated in 2007 and state that if all involved nodes have regressed to 1 cm or less, symptoms of disease have disappeared, and bone marrow biopsy is without involvement, the patient is in complete remission.⁵⁶ Because residual necrotic tissue and inflammatory cells may prevent bulky nodes from reducing to 1 cm or less, FDG-PET can help distinguish residual tumor cells from tissue necrosis. The timing of the imaging after therapy affects the positive predictive value, and it is recommended that FDG-PET be performed 3 to 8 weeks after chemotherapy and 8 to 12 weeks after radiation therapy.⁸⁸ In patients with bulky mediastinal lymph nodes, FDG-PET confirmation of a complete response is essential because these nodes frequently do not completely regress.