Overview of Prognostic Factors in Non-Hodgkin’s Lymphoma
Overview of Prognostic Factors in Non-Hodgkin’s Lymphoma
The non-Hodgkin’s lymphomas are a biologically heterogeneous group of diseases with varying clinical presentations and outcomes. A number of studies have identified variables that carried independent prognostic significance. Although several staging systems had evolved that incorporated these prognostic variables, they were still unable to predict outcome. Ideally, the object of a staging system is to predict the likelihood of treatment response, time to progression or disease-free survival, and overall survival, and to provide a way to compare the outcome of similar groups of patients among various clinical trials. The need for such a system led to the creation of prognostic models such as the M. D. Anderson Tumor Score and, more recently, the International Prognostic Index. These prognostic models may identify those patients at highest risk for treatment failure, thereby identifying those patients who may require different therapeutic approaches. [ONCOLOGY 12(Suppl 8):17-24, 1998]
There are many clinically important prognostic factors in non-Hodgkin’s lymphoma, though only a few have consistently been shown to correlate with prognosis and outcome in multivariate analyses. These prognostic parameters (Table 1) have been incorporated into various proposed staging systems, such as the M. D. Anderson Tumor Score and the International Prognostic Index (IPI).[1,2] Unlike in Hodgkin’s disease, the Ann Arbor staging system is of limited use in predicting outcome in patients with non-Hodgkin’s lymphoma, primarily due to its inability to estimate tumor burden. The Ann Arbor staging system becomes more useful when used in combination with other parameters that estimate tumor burden and which might also reflect, directly or indirectly, the tumor’s biologic features.
Surrogate parameters for tumor burden include an elevated beta-2-microglobulin (ß2M), bulky disease, and the number of extranodal sites .[3-5] Bulky disease in the M. D. Anderson Tumor Score System is defined as any peripheral lymph node or mass ³ 7 cm, a T3 or T4 lesion (by the TNM system) in the sinus cavity or nasopharynx, or > two-thirds infiltration of the stomach, liver, or other extranodal sites that are difficult to measure. Such estimates of tumor burden, when combined with serum lactic dehydrogenase (LDH) levels and used as a prognostic model, have proven useful in predicting 5-year survival. For example, patients with poor-risk disease as defined by extensive nodal and/or extranodal disease and elevated LDH levels had a 5-year survival of 20%, in comparison with 87% for those good-risk patients with less extensive disease and normal LDH levels.
Beta-2-microglobulin seems to correlate with tumor burden as well, and is an independently significant and easily measured prognostic parameter. The ß2M level has been used together with LDH to identify patients with a poor prognosis; in combination with an elevated serum LDH, an elevated ß2M correlates with an inferior time to treatment failure and survival among patients with aggressive non-Hodgkin’s lymphoma, regardless of Ann Arbor stage.
Swan et al correlated these variables to identify a distinct high-risk group of patients with large-cell lymphoma whose 2-year survival was 19% compared with a 2-year survival of 100% in a low-risk group of patients. In this system, an intermediate prognostic category was observed and consisted of cases with either ß2M or LDH elevation, but not both. This intermediate category is considered undesirable for treatment planning because it is much easier to design or select treatment for patients with either a very favorable or a very unfavorable prognosis, but is more difficult for patients with an intermediate prognosis. The former could be treated with conventional regimens whereas the unfavorable group could be entered on experimental programs.
Estimates of tumor burden have proven to be significant in multivariate analyses and have found their way into some of the more recent staging systems. Ann Arbor stage, an increased ß2M and LDH, the presence of constitutional “B” symptoms and bulky disease were incorporated into a staging system known as the M. D. Anderson Tumor Score. One point is assigned for each adverse prognostic feature with the sum representing a “tumor score.” An estimate of tumor burden, the tumor score, correlates with response rates (Table 2).
In a study of 144 patients who were uniformly treated with the CHOP regimen (cyclophosphamide [Cytoxan, Neosar], doxorubicin [Adriamycin], vincristine [Oncovin], and prednisone), those patients with a tumor score of 0 to 2 had a time to treatment failure rate of 83% compared to 24% for those patients with a tumor score of 3 or more. This system incorporated tumor-dependent variables that had been found to be significant in multivariate analyses, were easily reproducible, and correlated with outcome. A major advantage of this system is the definition of only two prognostic categories with no intermediate categories.
Coiffier et al prospectively applied several of these prognostic systems from various single institutions to patients receiving the LNH-84 regimen and identified groups of good- and poor-risk patients. (The LNH-84 regimen includes doxorubicin, cyclophosphamide, vindesine [Eldisone, Enisone], bleomycin [Blenoxane], prednisone, and methotrexate.) Although the ability of each system to predict prognosis was roughly equivalent, a commonly accepted definition of tumor burden remained elusive. Serum LDH levels and Ann Arbor stage remained the most important variables, although ß2M was not included as a variable. Coiffier then proposed a prognostic index based on LDH and three measurements of tumor burden (tumor size, the number of extranodal sites, and Ann Arbor stage). Other prognostic variables such as serum albumin levels and bone marrow involvement did not seem to add to the prognostic capability beyond the aforementioned variables.
The International Non-Hodgkin’s Lymphoma Prognostic Factors Project also reviewed pretreatment clinical prognostic factors in order to develop a system that might correlate with future outcome for newly diagnosed patients with non-Hodgkin’s lymphoma. The resulting International Prognostic Index (IPI) has proven to be particularly useful in predicting outcome, and is widely used. Both prospective and retrospective analyses of other data have confirmed the utility of this prognostic model.
The IPI incorporates advanced age (defined as greater than 60 years old), advanced stage (Ann Arbor III or IV), elevated LDH, poor performance status, and greater than one extranodal disease site. For example, patients with only one or fewer adverse prognostic features are classified as being low risk, with a complete response rate of 87%, and a 5-year overall survival rate of 73%. Patients with low-intermediate and high-intermediate risk have complete response rates of 67% and 55%, and 5-year survival rates of 51% and 43%, respectively. Patients with high risk disease have a projected complete response rate of only 44%, with a 5-year survival rate of 26%. Variations of the IPI scale, such as an age-adjusted IPI score, have also been found to be quite sensitive.
The IPI has proven useful in predicting outcome in patients with both aggressive and low-grade lymphoma and has even proven to be useful among patients with mantle cell lymphoma. Both the M.D. Anderson tumor score and the IPI are shown in Table 2 and Table 3 along with projected response rates and survival estimates. The only drawback of the IPI is the fact that there are a large number of cases that fit into the intermediate risk categories, making treatment selection problematic.
Bone Marrow and Central Nervous System Involvement
Patients with lymphoma involving the bone marrow, testes, or those with a lymphoblastic or Burkitt’s lymphoma are at greatest risk for central nervous system (CNS) involvement. Given the poor prognosis of CNS or lepto-meningeal lymphoma, it is important to search for central nervous system involvement in such patients at presentation and to incorporate CNS prophylaxis into the treatment plan whenever indicated. Similarly, bone marrow involvement is considered to be an adverse feature and may occur in approximately 10% of all patients with histologically aggressive non-Hodgkin’s lymphoma at the initial staging evaluation. While bone marrow involvement has been shown to be prognostically important, it does not significantly add to the utility of existing prognostic models such as the IPI, and is therefore not generally included.
The presence of divergent histology in the marrow is an interesting phenomenon with prognostic ramifications. Robertson et al described 50 patients with diffuse large-cell lymphoma of lymph nodes and evidence of bone marrow involvement. Forty-eight percent had large-cell lymphoma, 38% had small cleaved-cell lymphoma, and 14% had mixed histology in the marrow. Those with large-cell lymphoma in the bone marrow had a low complete remission rate (16.7%), a high risk of CNS involvement (33%), and a poor 5-year overall survival (12%). Those with small-cleaved-cell lymphoma in the bone marrow had a higher complete remission rate (89.4%), a lower rate of CNS involvement (5%), and a better 5-year survival rate (79%). However, those patients with small-cleaved-cell lymphoma in the marrow had a continuous rate of relapse with a progression-free survival of only 30% at 5 years and 15% at 8 years, reminiscent of the situation with indolent non-Hodgkin’s lymphoma.