Breast cancer is the most common noncutaneous malignancy in women in industrialized countries; it has been estimated that it will affect 211,300 women in the United States in 2003, accounting for 32% of the new cases of cancer in women. Breast cancer incidence increases with age. The median age at onset is 63 years, with approximately 45% of those who die of the disease being 65 years old or older. Breast cancer is the leading cause of cancer death in women after lung cancer, but the 5-year survival rate is much higher than that in other cancers. Standard treatment for patients with early-stage breast cancer includes adjuvant chemotherapy, which confers a survival benefit, while chemotherapy is considered palliative in patients with advanced disease. The most recent analyses of the Early Breast Cancer Trialists' Collaborative Group found that treatment with adjuvant chemotherapy resulted in 23.5% reductions in the annual risk of the recurrence of breast cancer and 15.3% reductions in death due to breast cancer. A clear benefit was seen in patients up to 70 years old, with anthracycline-containing regimens appearing to be better than non-anthracycline-containing regimens. The impact of maintaining the dose intensity of the chemotherapy on disease- free and overall survival has been shown in trials of adjuvant chemotherapy. A retrospective study by Bonadonna and colleagues assessed survival with adjuvant chemotherapy with CMF (cyclophosphamide [Cytoxan, Neosar], methotrexate(Drug information on methotrexate), fluorouracil(Drug information on fluorouracil) [5-FU]); follow-up at 20 years found that patients receiving less than the standard dose had suboptimal outcomes (Figure 1). The investigators found that optimal long-term survival was associated with a delivered dose of at least 85% of the reference standard. The importance of relative dose intensity was confirmed in a prospective randomized trial in more than 1,500 patients with early-stage breast cancer treated with three different dose intensities (low, moderate, and standard) of adjuvant chemotherapy with cyclophosphamide(Drug information on cyclophosphamide), doxorubicin(Drug information on doxorubicin), and 5-FU. At a median follow-up of 9 years, both disease-free and overall survival were higher with the standard and moderate dose intensities than with the low dose intensity. A recent study by the Cancer and Leukemia Group B (CALGB) of adjuvant chemotherapy in nearly 6,500 patients, including those described above, showed similar dose-related benefits; the patients treated with what was considered at the time to be high-dose adjuvant chemotherapy were 12% more likely to remain alive and disease- free at a median follow-up of 9.6 years than were those treated with lowdose chemotherapy. Neutropenia and the Delivery of Chemotherapy Maintaining full-dose chemotherapy is often hampered by the occurrence of myelosuppression, with chemotherapy-induced neutropenia being the primary cause of dose delays and reductions in patients with early-stage breast cancer. A nationwide survey of more than 1,000 patients with early-stage breast cancer treated at 13 oncology practices found that 30% of the patients received less than 85% of the standard reference dose. The doses were delayed or reduced in 45% of the patients overall, and neutropenia was the cause in 61% of these dose modifications. A more recent and much larger survey of practice patterns, in more than 20,000 patients with early-stage breast cancer treated with adjuvant chemotherapy, found that 35% of patients had dose reductions of more than 15% and 25% had a treatment delay of more than 7 days. Overall, 56% of the patients were treated with a relative dose intensity of less than 85%, including 67% of those older than 65 years. Randomized clinical trials have shown that preemptive management with granulocyte colony-stimulating factor (G-CSF) can reduce the duration of severe neutropenia and the risk of its complications.[8,9] Such prophylaxis also facilitates the delivery of the planned dose of chemotherapy on time. In a trial by de Graaf and colleagues, 74% of patients treated with CSF 24 hours after chemotherapy were given at least 85% of the planned chemotherapy dose intensity, compared to only 45% of the controls. A meta-analysis of the results of eight randomized controlled trials of CSF found that chemotherapy dose reductions or delays were more than twice as common in patients who were given placebo. Furthermore, the use of CSF early in therapy affects the likelihood of neutropenic complications in both the initial and the subsequent cycles of therapy. Two randomized phase III trials that compared the G-CSFs filgrastim(Drug information on filgrastim) (Neupogen) and pegfilgrastim (Neulasta) in patients with breast cancer treated with docetaxel(Drug information on docetaxel) (Taxotere) and doxorubicin found that pegfilgrastim was comparable to filgrastim in reducing the incidence of febrile neutropenia and the duration of grade 4 neutropenia.[12,13] The pattern of less severe neutropenia in the later cycles was also seen in the pivotal trials of filgrastim, with the duration of grade 4 neutropenia being reduced from 3 days in cycle 1 to 1 day in the later cycles.[8,9] Colony-stimulating factor has been shown to be clinically effective, but its use in all patients with breast cancer treated with myelosuppressive chemotherapy would not be cost-effective. It is important to devise strategies for using CSF in those patients who are at greatest risk for neutropenic complications and reduced dose intensity. These strategies will identify those patients who are most likely to benefit from CSF, which will reduce the risk of neutropenia and make it possible for full-dose-intensity chemotherapy to be maintained. Validated and reliable predictive models should be useful in guiding treatment decisions and selecting patients for prophylaxis with CSF. Risk Models for Neutropenia in Breast Cancer Silber and colleagues developed a risk model for neutropenic complications in patients treated with adjuvant chemotherapy for early-stage breast cancer. Logistic regression models were developed for unconditional (pretreatment) factors and conditional (based on the patients' initial hematologic response to the chemotherapy) factors. The pretreatment model was unsuccessful in accurately predicting neutropenia, dose reductions, or dose delays. The conditional model, however, found a significant association between the depth of the absolute neutrophil count (ANC) nadir in cycle 1 and subsequent neutropenic complications. The value of the first-cycle ANC nadir was validated as a predictor of complications of neutropenia in a second group of patients with early-stage breast cancer. This retrospective analysis used the same definition of neutropenic events as Silber and colleagues but also examined the incidence of febrile neutropenia. The firstcycle ANC nadir was the only risk factor that was found to be significantly predictive (P < .0001) of neutropenic events in subsequent cycles, with the rate of neutropenic complications being 30% in patients with a first-cycle ANC nadir of 0.5 × 109/L or less and 10% in those with higher nadirs (P = .04). The first-cycle ANC nadir also predicted the likelihood of the chemotherapy dose intensity being less than 85%. The rates of chemotherapy dose intensity of 85% or less were 55% in patients with a first-cycle ANC nadir of 0.5 × 109/L or less and 32% in those with higher nadirs (P = .05). The Silber model, in which G-CSF is used in all subsequent cycles in patients with a first-cycle ANC nadir of 0.5 × 109/L or less (categorized as high risk), was prospectively validated in a study by Rivera and colleagues. In this study, the rate of low delivered dose intensity-85% or less-in highrisk patients was only 5%, compared to 12.1% in matched historical controls (Figure 2). The incidences of febrile neutropenia and of hospitalization due to febrile neutropenia were similar in the high-risk group and the historical controls (10.9% and 9.4%, and 4.2% and 4.7%, respectively) (see Figure 2). Silber and colleagues then built a cost-effectiveness model based on the relation between chemotherapy dose intensity and disease-free survival in early-stage breast cancer that had been shown in CALGB 8541. They found a cost-effectiveness ratio of $34,297 per year of life saved when G-CSF was used in the 50% of patients who were classified as high risk by this model. A study of pretreatment and posttreatment factors in patients with early-stage breast cancer treated with adjuvant chemotherapy found that several variables affect the risk. The pretreatment factors that predicted hematologic complications in the first cycle of chemotherapy included age greater than 65 years, treatment with anthracycline-containing regimens, and low pretreatment blood cell counts. Posttreatment factors during the first cycle that were predictive of subsequent neutropenic events included ANC nadir less than 0.5 × 109/L, febrile neutropenia in the first cycle, and the extent of the drop in hemoglobin level. Risk models that have been developed thus far have been based on retrospective analyses and have defined neutropenic complications in various ways. A prospective registry that includes patients with breast cancer has been implemented in order to further identify pretreatment and early treatment risk factors for subsequent neutropenia to make early intervention with first-cycle CSF support possible.[ 19] Dose-Dense Chemotherapy in Early-Stage Breast Cancer A recent study with dose-dense chemotherapy (ie, using standard dose sizes but shortening the intervals between the cycles) has shown better patient outcomes. The observation that maintaining the dose intensity of standard chemotherapy is associated with better clinical outcomes led to the investigation of using greater dose intensity in adjuvant chemotherapy for breast cancer. Dose intensity (the amount of drug delivered per unit of time) can be increased either by increasing the dose (dose escalation) or by shortening the interval between cycles (dose density). The results with dose escalation have thus far been equivocal. Studies that compared conventional- dose chemotherapy and highdose chemotherapy followed by stem cell transplantation (HDCT/SCT) have mostly been negative in the adjuvant setting in high-risk patients with nodepositive disease.[21-24] A few trials have shown a benefit of HDCT/SCT in a subset of patients. Roche et al showed an improvement in diseasefree survival in those with 7 to 10 positive nodes. Recently, Tallman et al showed a significantly longer time to recurrence in those assigned to HDCT/SCT who met strict eligibility criteria. Rodenhuis et al showed greater relapse-free survival in those with 10 or more positive nodes and those with tumors that did not overexpress HER2/neu. None of these trials has shown a benefit in overall survival in all patients. The dose-dense approach to increasing dose intensity is based on preclinical models of the growth of cancer cells by nonexponential Gompertzian kinetics. In volume-reduced Gompertzian cancer models the regrowth of cancer cells between the cycles of cytoreduction is more rapid than in exponential models. Thus, it is unclear whether simple dose escalation is enough for the success of adequately planned multicycle regimens, since other strategies, such as dose density, may prove to be more potent as a therapeutic manipulation.[ 29] This hypothesis was initially investigated in pilot studies. Hudis and colleagues reported the results in a trial in which 42 patients with resected breast cancer that involved more than three positive nodes were treated with sequential dose-dense chemotherapy, consisting of three cycles of doxorubicin, followed by three cycles of paclitaxel(Drug information on paclitaxel), and then three cycles of cyclophosphamide (A → T → C) in 14-day cycles with G-CSF support. This regimen proved to be feasible and promising. The reported actuarial disease- free survival rate was 78% after a median of 4 years, with only 4 deaths due to metastatic disease. Because combination chemotherapy is often more toxic than single agents, Fornier and colleagues compared the toxicity of a sequential dosedense regimen of doxorubicin, paclitaxel, and cyclophosphamide (A → T → C) and that of another dosedense regimen with the same schedule of doxorubicin, followed by three cycles of concurrent paclitaxel and cyclophosphamide (A → TC). There was greater toxicity in the concurrent arm, with more hospitalizations for febrile neutropenia and more red blood cell transfusions for anemia. Furthermore, the mean delivered dose intensities of the paclitaxel and cyclophosphamide were significantly greater in the sequential arm than in the concurrent one (P = .01 and P = .05, respectively). Thus, dosedense sequential chemotherapy is more feasible than doxorubicin followed by concurrent paclitaxel and cyclophosphamide. The effect of dose-dense chemotherapy was then tested in a large prospective phase III study coordinated by the CALGB for the National Cancer Institute's Breast Intergroup, INT C9741. The study compared sequential doxorubicin, paclitaxel, and cyclophosphamide (A → T → C) with concurrent doxorubicin and cyclophosphamide followed by paclitaxel (AC → T), using dose-dense (2- weekly) and conventional (3-weekly) schedules, as adjuvant chemotherapy in 1,973 patients with breast cancer. The dose-dense schedule used G-CSF in all patients to make the 2-week cycles possible by lessening neutropenic complications (Figure 3). The dose-dense regimens resulted in significantly higher 3-year diseasefree survival (85% vs 81%; P = .01) and 3-year overall survival (92% vs 90%; P = .013), regardless of predictive factors such as the number of positive nodes, tumor size, menopausal status, and tumor estrogen receptor status (Figure 4). There was no difference in disease-free survival or overall survival between the sequential and concurrent arms. Grade 4 neutropenia was more common with conventional therapy, occurring in 33% of patients treated with conventional regimens and 6% of those treated with dosedense regimens (P < .0001). In addition, fewer cycles were delayed because of hematologic toxicity with dose-dense than with conventional therapy (15% vs 38%; P < .0001). In order to deliver conventional chemotherapy in a dose-dense schedule, CSF support is required to reduce neutropenic complications. There was more grade 4 neutropenia in the 3-week-cycle arm, but more patients (13%) required red blood cell transfusions in the 2-week-cycle arm of AC → T than in the three other arms. This was not because of significant grade 3 or 4 anemia, but was perhaps caused by grade 2 anemia (data not reported). With the availability of erythropoietin, the need for red blood cell transfusions should be diminished. Furthermore, dose-dense chemotherapy significantly reduced the occurrence of contralateral breast cancer (0.3% vs 1.5%, P = .0004). CALGB 9741 showed not only the feasibility of this approach, but also the superiority of dose-dense over conventional chemotherapy. These findings are exciting and are consistent with previous mathematical model predictions that shortening the interval between chemotherapy cycles could result in more-effective eradication of malignant cells, potentially improving survival. The sequential approach tested in CALGB 9741 failed to show superiority of singleagent sequential therapy over combined doxorubicin/cyclophosphamide sequenced into paclitaxel, but showed no disadvantage for uncoupling agents from one another, either. On the basis of current data, practicing oncologists may consider treating patients with breast cancer in this dose-dense fashion. However, extrapolating these data to all regimens outside of a clinical trial setting should be done with caution, as unexpected toxicities may emerge. These findings suggest important avenues for future research in both breast cancer and other chemosensitive tumors, and confirmatory studies are encouraged. Conclusions Maintaining the dose of the chemotherapy is important in increasing long-term survival in patients with early-stage breast cancer. Neutropenia, the major dose-limiting toxic effect of myelosuppressive chemotherapy, can be limited with early use of CSF. Evidence- based risk models for predicting which patients are at greatest risk for neutropenia and its complications may be an efficient and cost-effective way of limiting these complications and helping ensure that the chemotherapy is delivered as planned. More prospective research is needed to determine which factors to use in risk models that predict subsequent neutropenia and its complications. Early results with dose-dense chemotherapy in CALGB 9741 are exciting, showing improved disease-free and overall survival and less grade 4 neutropenia when compared to conventionally scheduled chemotherapy. The findings of this study suggest that dose-dense scheduling with appropriate chemotherapy regimens that require CSF support may replace conventional dosing as the new standard of care in early-stage breast cancer.