The incidence of and mortality in 8 of the 10 most common types of cancer in the United States are decreasing.[1,2] One of those in which they are increasing is non-Hodgkin's lymphoma, and it is estimated that 53,400 new cases of non-Hodgkin's lymphoma and 23,400 deaths due to it will occur in the United States in 2003. These increases have been most notable in patients older than 60 years, in whom 61% of the new cases of the disease occur. Combination chemotherapy with CHOP (cyclophosphamide [Cytoxan, Neosar], doxorubicin(Drug information on doxorubicin) HCl, vincristine [Oncovin], prednisone(Drug information on prednisone)), CHOP-like, or CHOP-R (plus rituximab(Drug information on rituximab) [Rituxan]) regimens is considered the standard of care in young and older patients alike with aggressive non-Hodgkin's lymphoma.[ 6-8] Non-Hodgkin's lymphoma is potentially curable and complete responses are achieved in many patients, but treatment failures and mortality are more common in older patients. In fact, age 60 years and older is a prognostic factor for adverse outcomes in patients with intermediate- grade non-Hodgkin's lymphoma.[ 9] The CHOP regimen produces complete responses in only 40% to 50% of elderly patients, with 3-year event-free and overall survival being 30% and 35% to 40%, respectively.[ 10] Nonmyeloablative, doseintensified chemotherapy regimens may improve the outcomes in younger patients, but they are generally not well tolerated in elderly patients. The lower survival in elderly patients has been attributed to the greater toxicity of standard full-dose chemotherapy in the presence of comorbidities and poor bone marrow reserves. Concern over elderly patients' tolerance of chemotherapy has led many physicians to treat them less aggressively. This may be accomplished either by administering lessmyelosuppressive (and potentially less-effective) chemotherapy or by administering substandard dose intensities. A retrospective analysis of data from a large survey of practice patterns in patients with non-Hodgkin's lymphoma found that older patients were significantly more likely than younger patients to be treated with a planned average relative dose intensity of 80% or less (28% vs 12%, P < .001). Elderly patients were also more likely to be treated with less-aggressive, and therefore potentially less-effective, regimens than younger patients. For example, non-anthracycline-containing regimens such as CNOP (cyclophosphamide, mitoxantrone [Novantrone], vincristine, prednisone) and CVP (cyclophosphamide, vincristine, prednisone) were more commonly used in patients aged 60 years and older than in younger patients (21% vs 7%, P < .001). This practice may be one of the reasons for the lower survival rates in the elderly. Routine undertreatment may be due to assumptions about their frailty or relative inability to tolerate chemotherapy. However, there is considerable evidence that elderly patients will benefit as much as younger patients when they are given equivalent chemotherapy doses. A classic example of this is seen in the results of a study by the Southwest Oncology Group (SWOG) in 307 patients with large cell lymphoma treated with CHOP. Of the 81 elderly patients in the study, 23 were given full-dose treatment in violation of the protocolstipulated 50% reduction in the chemotherapy dose in patients aged 65 years or older. Analysis of the results in these 23 patients found no significant differences between them and younger patients in the complete response rate, duration of complete response, or frequency of treatment-related complications. The investigators therefore concluded that the routine reduction in chemotherapy doses on the basis of age alone is inappropriate and may be a primary contributor to the poorer outcomes in elderly patients. In light of the results of this and other studies in non-Hodgkin's lymphoma, the relation between dose intensity and outcomes in this disease has become well established. Several retrospective studies have shown that maintaining the standard dose intensity of the chemotherapy increases the likelihood of a complete response and greater survival.[14-16] Kwak and colleagues reported that the strongest factor associated with greater survival was the delivery of more than 75% of the relative dose intensity of doxorubicin.[ 14] Similarly, Epelbaum and colleagues found that delivery of an average relative dose intensity of more than 80% resulted in a significantly higher complete response rate and that the delivery of more than 70% resulted in higher 5-year overall survival. Lepage and colleagues found that the response rate and overall survival were significantly lower in patients with aggressive non-Hodgkin's lymphoma who were treated with less than 70% of the planned dose intensity. Older patients were shown to have outcomes similar to those in younger patients when the dose intensity was maintained with the use of CSF. It is clear that maintaining the dose intensity of the chemotherapy is important in non-Hodgkin's lymphoma, and managing dose-limiting toxicities plays an important role in this. Chemotherapy-Induced Neutropenia Severe neutropenia and its complications are the major dose-limiting toxicities of chemotherapy for non- Hodgkin's lymphoma. Hospitalization for febrile neutropenia is associated with substantial costs, lower quality of life, and a greater risk of mortality, especially in elderly patients with comorbidities. In a study that compared data on hospitalizations for febrile neutropenia in patients with lymphoma and patients with solid tumors, the mean lengths of stay were longer (10.8 vs 8.5 days), the costs per episode greater ($19,061 vs $12,302), and treatment-related mortality higher (10% vs 9%) in patients with lymphoma.[ 18] The complications of neutropenia in patients with lymphoma may lead to treatment delays and dose reductions, seriously compromising the dose intensity in patients with this potentially curable cancer. As was noted above, the risk of neutropenia is a major factor in the choice of chemotherapy and in regimen modifications, and this can result in suboptimal treatment. The first occurrence of febrile neutropenia is most likely in the early cycles of chemotherapy. A randomized trial that assessed the toxicity of chemotherapy in 453 elderly patients with non-Hodgkin's lymphoma reported that 43% to 68% of all occurrences of grades 3 and 4 leukopenia were in the first cycle, depending on the chemotherapy regimen. A retrospective analysis of data on 577 patients with non-Hodgkin's lymphoma found that 62% of the initial occurrences of febrile neutropenia were in the first cycle (Figure 1). Another study found that more than 65% of the hospitalizations for febrile neutropenia were in cycles 1 and 2. Furthermore, a retrospective analysis of the data on 267 elderly patients treated with CHOP found that 63% of the treatment- related deaths occurred in the first cycle. These findings make it clear that neutropenia and its complications are best prevented early in the course of chemotherapy. It has been shown that first-cycle management with colony-stimulating factor (CSF) in both younger and older patients with non-Hodgkin's lymphoma reduces the incidence and duration of neutropenia and its complications and helps maintain the chemotherapy dose intensity. Four placebocontrolled randomized trials of firstcycle CSF use with standard-dose CHOP and regimens with similar myelosuppressive potential in elderly patients with non-Hodgkin's lymphoma have reported significantly lower incidences of neutropenia and its complications in the CSF arms, by 32% to 83% (Table 1).[23-26] In the trial of CHOP and CHOP-like regimens by Zagonel and colleagues the rates of grade 3 or 4 neutropenia and neutropenia-related infection were significantly lower in the patients who were given CSF than in those who received placebo; in addition, the use of CSF was also associated with fewer and shorter treatment delays and with shorter lengths of stay in patients hospitalized for febrile neutropenia. Jacobson and colleagues also reported that first-cycle CSF use made it possible to deliver full-dose standard chemotherapy in elderly patients with non-Hodgkin's lymphoma. Risk Models in Non-Hodgkin's Lymphoma Research is under way to determine risk factors in order to develop a risk model for chemotherapy-induced neutropenia in patients with non- Hodgkin's lymphoma. This will make it possible to target CSF to those at highest risk. Several risk factors have thus far been identified, mainly in retrospective studies. Since data indicate that most complications of neutropenia occur early in the first few cycles of chemotherapy, predictive risk studies in non-Hodgkin's lymphoma have assessed unconditional patient factors for the development of chemotherapyinduced neutropenia in the first cycle. A number of other risk factors have been identified, and predictive risk models that use these factors have been developed. The risk factors seen repeatedly in several studies[20, 28-31] include advanced age, poor performance status, planned high chemotherapy dose intensity, the presence of hepatic, renal, or cardiac comorbidities, and lack of primary prophylaxis with CSF (Table 2). Less-obvious pretreatment factors-including low serum albumin levels, elevated lactate dehydrogenase levels, bone marrow involvement, and high levels of soluble tumor necrosis factor receptor-have also been identified,[ 20,28,31,32] but most of these have not yet been validated in separate patient populations. Each risk factor for chemotherapyinduced neutropenia considered on its own has limited predictive power, and a clinically useful risk model should consider all of the risk factors in a patient. A pretreatment risk model developed through regression analysis found five significant independent predictors for the first episode of febrile neutropenia. These predictors of early febrile neutropenia were age greater than 65 years, serum albumin level less than 35 g/L, planned dose intensity greater than 80% of the standard, absolute neutrophil count less than 1.5 × 109/L, and the presence of hepatic disease. A composite risk score of the number of risk factors present differentiated patients as being at low (1 or 2 factors) or high (≥ 3 factors) risk for hospitalization for febrile neutropenia. A clinical decision model was then developed to assess the cost implications of three strategies for using firstcycle CSF in patients treated with CHOP or CHOP-like regimens for non-Hodgkin's lymphoma. The first treatment strategy was "control," in which CSF was not used in any pa tients; the second was "universal," in which the CSF pegfilgrastim (Neulasta) was used in all patients in all cycles; and the third was "model," in which pegfilgrastim was used in all cycles in those patients who were at high risk according to the model described above. The "model" strategy was found to be the most cost-effective, because it targets pegfilgrastim prophylaxis to those who are most likely to benefit from it. Dose-Dense Chemotherapy in Non-Hodgkin's Lymphoma Maintaining the dose intensity of chemotherapy is associated with greater overall survival in non- Hodgkin's lymphoma; therefore, it is logical to assume that increasing the dose intensity would further improve the outcomes. Recent studies have tested this hypothesis, finding that the dose intensity can be increased by shortening the time between the chemotherapy cycles, which also requires prophylactic CSF in all cycles. A phase II study in 120 patients with non- Hodgkin's lymphoma evaluated the efficacy of filgrastim(Drug information on filgrastim) in CHOP with CSF support given in 14-day cycles. This was compared with historical data for CHOP given in the standard 21- day cycles. In this study 90% of the cycles were given at full dose and on time and 86% of the patients were treated with the planned six cycles. The preliminary results of a phase III trial of dose-dense (14-day) CHOP with CSF support vs 21-day CHOP in 738 patients older than 60 years have also been reported. At a median follow-up of 49 months, CHOP-14 resulted in a significantly higher complete response rate (77.1% vs 63.5%, P = .028), greater overall survival (58.6% vs 44.6%, P = .002), and greater disease-free survival (47% vs 39.2%, P = .024) (Figure 2). The CHOP-14 plus CSF regimen was also well tolerated, with a lower rate of grade 4 leukopenia than the CHOP- 21 regimen (24% vs 44%). Furthermore, appropriate supportive care with CSF was required with dosedense regimens; the rates of grades 3 and 4 infections when filgrastim was given for 10 days were half those when it was given for only 7 days (11% vs 21%). The results from these studies are promising, and suggest that the routine use of dose-dense regimens with prophylactic CSF may prove to be an important strategy for improving the outcomes in patients with non- Hodgkin's lymphoma. Conclusions The incidence of and mortality in non-Hodgkin's lymphoma are increasing. This disease is potentially curable, but treatment failures are common and can often be attributed, at least in part, to a failure to deliver adequate doses of chemotherapy because of perceived or actual problems with myelosuppression. Dose reductions and delays are most common in elderly patients, and therefore, not unexpectedly, the response rates and overall survival are lower than those in younger patients. Several risk factors for chemotherapyinduced neutropenia and its complications (eg, hospitalization for febrile neutropenia and dose delays or reductions) have been identified in patients with non-Hodgkin's lymphoma treated with standard-dose regimens. Incorporating these risk factors into a predictive model would make it possible to determine those patients at highest risk for chemotherapy-induced neutropenia and to appropriately target first-cycle CSF to them. Furthermore, a recent study assessed three strategies for prophylaxis with CSF and found that the "model" strategy, in which the CSF pegfilgrastim is used only in those patients at high risk on the basis of their composite risk score, was the most costeffective. The promising results with dose-dense chemotherapy with CSF prophylaxis, providing greater overall survival in the elderly, should be confirmed in additional studies. Until that time, a pretreatment predictive risk model for identifying patients at risk for chemotherapy-induced neutropenia and its complications should reduce the incidence and severity of chemotherapy- induced neutropenia by making the targeted use of CSF possible in those most likely to benefit from it. Furthermore, it will facilitate the cost-effective delivery of standard full-dose chemotherapy on time.