• Assess the efficacy of fludarabine and cyclophosphamide followed by GM-CSF and rituximab in inducing clinical and molecular response in patients with relapsed, refractory and newly diagnosed low-grade NHL. [ Time Frame: 6 months ] [ Designated as safety issue: Yes ]
  

• Determine the optimum dose of GM-CSF for CD20 antigen upregulation. [ Time Frame: 6 months ] [ Designated as safety issue: Yes ]
  

Patients with a low-grade, or indolent (slow-growing) form of non-Hodgkin's lymphoma (NHL) in which the usual survival is between 7-10 years are being asked to take part in this study. Although normally-used combinations of chemotherapy will cause NHL to disappear in 30-40% of patients (called complete response or complete remission), almost all will have their disease return.

When NHL is diagnosed, an abundance of white blood cells called B-lymphocytes (or B-cells) are found in the body. Almost all B-cells have a special protein on the surface called a CD20 antigen. Some anti-cancer drugs, called monoclonal antibodies, target cancer cells by binding, or "locking up", specific antigens found on their surfaces, which kills the cancer cells.

In this study, researchers will test a combination of anti-cancer agents to see if a better and more long-lasting response can be achieved. All of the medications are approved by the Food and Drug Administration (FDA) and are available on the market. The agents we will use are:

• Mitoxantrone and fludarabine, a combination of chemotherapy drugs that has been successfully used to treat NHL that has returned after treatment.
• Rituximab, a monoclonal antibody that kills cancer cells by binding the CD20 antigen found on the surface of B-cells, commonly used along with chemotherapy drugs to improve response rates in lymphoma treatment.
• GM-CSF (granulocyte-macrophage colony stimulating factor, also called sargramostim, GM, or Leukine), a growth factor which stimulates the development of new (stem) cells. GM-CSF encourages stem cells to divide, specialize, and become active. It is not a normal part of treatment for NHL.

Using GM-CSF in NHL treatment is the experimental part of this study. In studies done in the laboratory, GM-CSF caused an increase in the number of antigens, such as CD20, on the surface of B-cells. If more antigens are present, it may be easier to target cells that express CD20 or other antigens. Monoclonal antibodies (such as rituximab) might then be able to more effectively bind the antigens and kill the cancer cells.

The main purpose of this study is to see if giving GM-CSF along with a standard anti-cancer treatment will work better to reduce cancer, and to look at side effects of the treatment.

The researchers want to see if the laboratory results using GM-CSF (increased number of antigens on B-cells) hold true in human subjects and they also want to determine the best dose of GM-CSF and the best time to give rituximab.

Inclusion Criteria:

• To qualify for this study, the patient must have relapsed, refractory or previously untreated low-grade (indolent) non-Hodgkin lymphoma of the following subtypes: Follicular center cell lymphoma grade 1, lymphoplasmacytoid lymphoma, small lymphocytic lymphoma, splenic marginal-zone types lymphoma, monocytoid B-cell lymphoma and extranodal mucosa-associated lymphoid tissue (MALT) lymphomas. Final eligibility will be determined by the health professionals conducting this clinical trial.

Exclusion Criteria:

• Patients who have received prior treatment with purine analogs will be excluded from this study. Also, patients whose diagnostic/histologic subtype cannot be confirmed by our institution will not be able to participate in this study. Final eligibility will be determined by the health professionals conducting this clinical trial.