Risk Assessment in Oncology Clinical Practice
Risk Assessment in Oncology Clinical Practice
Despite the recent introduction of a number of new agents for cancer treatment, chemotherapy remains the principal method of systemic treatment in many of the nearly 1.3 million cancer patients in the United States each year.[1,2] Along with surgery and radiotherapy, chemotherapy plays a major role in the treatment of many common malignancies- including lung, breast, colon, and ovarian cancer-as well as the hematologic malignancies. The importance of the dose-response relation of chemotherapy agents has long been recognized. Long-term studies have shown the importance of maintaining chemotherapy dose to patient survival, especially in responsive and potentially curable malignancies.[4-8] Myelosuppression, specifically neutropenia, is the major dose-limiting toxicity of cancer chemotherapy, constraining the regimens, doses, and schedules that can be used. Consequently, avoiding or reducing the risk of chemotherapy-induced neutropenia continues to be an important consideration in the treatment of patients with cancer. Chemotherapy-Induced Neutropenia Because neutrophils represent a major defense against infection, neutropenia can create a favorable environment for bacterial invasion and multiplication, with the potential for rapidly spreading life-threatening infection.[ 9] The relative absence of neutrophils also reduces the symptoms and signs of infection, with fever often being the only indication of infection in a patient with neutropenia. Because of the risk of life-threatening infection and sepsis, patients with febrile neutropenia must be treated aggressively.[ 9,10] Hospitalization for rapid evaluation and treatment with intravenous antibiotics remains the standard of care for the majority of these patients. Hospitalization for febrile neutropenia represents a considerable financial burden in the care of patients with cancer. The current costs of cancer care are estimated to be more than $150 billion per year, with the financial consequences of hospitalization accounting for the largest single component of these costs. Two economic analyses have placed the mean length of hospitalization for febrile neutropenia at 10 days and the average cost of hospitalization at more than $20,000.[12,13] Substantial indirect and out-of-pocket costs have also been reported in association with neutropenia and its complications. Neutropenia also has negative effects on patients' quality of life. When patients are hospitalized for febrile neutropenia their well-being suffers from being separated from family. Patients hospitalized with febrile neutropenia may be anxious about infection, invasive medical procedures, ineffective treatments, and death. Further, preliminary research suggests an association between the depth of the absolute neutrophil count (ANC) nadir and a decline in quality of life.[16,17] A neutropenia-specific quality-of-life instrument developed to study how neutropenia can affect specific quality-of-life domains such as fatigue and worry is undergoing further validation. Perhaps of greatest importance, the complications of neutropenia can lead to reductions in chemotherapy dose intensity, adversely affecting longterm patient outcomes. The occurrence of neutropenia often leads physicians to reduce the chemotherapy dose or delay subsequent cycles of treatment. The net result is the delivery of lower relative dose intensity.[19-21] In potentially curable cancers, such as earlystage breast cancer and non-Hodgkin's lymphoma, there is considerable evidence that such chemotherapy dose alterations are associated with lower disease- free and overall survival.[4-6,8] The extent to which chemotherapy dose intensity was modified in major randomized clinical trials has been explored in a systematic review of the literature on chemotherapy in patients with early-stage breast cancer and non-Hodgkin's lymphoma. Data on relative dose intensity delivered were reported inconsistently and in various ways in the trials analyzed in this study, with 40% of the trials not reporting dose intensity information at all. Studies of practice patterns have also shown the extent to which chemotherapy dose reductions and delays occur in practice. A survey of practice patterns in more than 20,000 patients treated with chemotherapy in the adjuvant setting for breast cancer found that 56% of patients were treated with a relative dose intensity less than 85% of the reference standard, increasing to 67% of patients aged 65 years and older. Several factors may influence the relative dose intensity delivered, including both planned and unplanned dose reductions and treatment delays (Table 1). The Role of Colony-Stimulating Factor Prophylactic colony-stimulating factor (CSF) is associated with a reduced risk of febrile neutropenia, documented infection, and the need for dose reduction or treatment delay, often enabling the delivery of full dose intensity.[23-25] Clinical practice guidelines for the use of the CSF have been developed by the American Society of Clinical Oncology (ASCO). These guidelines currently support the routine use of prophylactic CSF in patients treated with a chemotherapy regimen with an average risk of febrile neutropenia of 40% or greater. These guidelines were supported by early economic analyses based on a limited assessment of the direct medical costs of hospitalization for febrile neutropenia at a single institution.[ 27] This model found that CSF use is associated with a reduction in cost when the risk of febrile neutropenia is above the 40% threshold. The risk threshold is the breakeven point at which the added cost of the CSF is offset by the reduction in the costs of hospitalization for febrile neutropenia. The costs of hospitaliza- tion for febrile neutropenia have been further studied both within single-institution and in multi-institution studies. Incorporating updated cost estimates into this economic model provides threshold risk estimates closer to 20% (Figure 1). The ASCO CSF guidelines are currently being revised in consideration of this and other additional clinical and economic information. Risk thresholds for the cost-saving use of CSF have been further evaluated in sensitivity analyses and are seen to depend on the assumed baseline risk of febrile neutropenia. It is clear that some chemotherapy regimens are more likely to cause serious myelosuppression than others, but much about the patterns and incidence of chemotherapy-induced neutropenia with the regimens remains unknown. A survey of the literature for the rates of neutropenia in randomized clinical trials of chemotherapy found that the rates of myelosuppression and its complications often were not reported.[ 22] When they were reported, the rates of hematologic toxicity with the same and similar regimens varied greatly, making it difficult to assess the actual risk. The actual risk has been shown to vary with a large number of treatment-, disease-, and patient-related characteristics. The use of a specified risk threshold for a given regimen to select patients for first-cycle CSF use assumes an average population risk profile. There clearly is a need to "individualize" the risk by considering a specific patient's risk factors. This can be most reliably and reproducibly done by using a risk model to select patients for CSF prophylaxis. Whether to use CSF in accordance with a costminimization model with a specified risk threshold is an economic decision that is based on population risk, and it does not take into account the possible clinical benefits in individual patients in whom neutropenic complications could be avoided. The optimal cost-effective use of CSF hinges on using it in those patients who are most likely to benefit from it, and being able to predict who those patients are. In the absence of reliable values for chemotherapy regimen- specific risks, clinicians often use various patient characteristics to get an idea of a patient's risk for neutropenia and its complications. The ASCO guidelines define "special circumstances" as patient populations in whom the risk is high and primary prophylaxis with CSF is recommended. These include extensive prior treatment, prior radiation therapy to considerable marrow-producing bone, and risk factors for infection, includ- ing open wounds, low performance status, and reduced immune function. However, the list of these special circumstances is not comprehensive, nor is the list routinely used in selecting patients for first-cycle CSF management. The lack of uniform standards can leave oncologists uncertain about which patients are at greatest risk for neutropenic complications. A "watch and wait" approach is often used, in which patients who experienced severe or febrile neutropenia in the first cycle are given CSF after chemotherapy in later cycles. Such "secondary prophylaxis," while reasonable in lower-risk settings, is not a desirable option in those at greater risk, given the life-threatening nature of febrile neutropenia and demonstration with several chemotherapy regimens that the greatest risk of the initial episode of febrile neutropenia occurs early in the treatment course.[29- 31] Initiating CSF later in the treatment course forces most patients to experience the very complication they are intended to prevent. Clearly, a new strategy is needed to better select patients for early CSF management. Risk Models for Neutropenia Considerable research has been directed toward identifying patient-, disease-, and treatment-related factors associated with the risk of neutropenia and its complications. The focus has moved from the uncertainty of reported risk associated with various regimens in highly selected populations to a more focused assessment of risk in the individual patient based on his or her clinical condition and treatment history. The purpose of clinical predictive models is to define the independent contributions of the various risk factors for neutropenic complications and reduced dose intensity. By identifying patients at greater risk, such models can be used to target the CSFs more efficiently and cost-effectively to those most likely to benefit from such supportive care. Predictive risk models more closely reflect the situation in clinical practice, in which patients are selected for treatment on the basis of their individual prognostic and predictive factors, while basing such decisions on explicit, objective, and reproducible measures that can be systematically applied and validated[ 28] (Figure 2). Risk models for neutropenic events have been studied most extensively in chemosensitive, potentially curable malignancies with full-dose chemotherapy, namely early-stage breast cancer and non-Hodgkin's lymphoma.[ 32-35] Those risk factors found to be statistically significant in multivariate models in two or more studies are indicated in Figure 3. Clinical prediction models evaluate relations between risk factors and clinical outcomes, such as hospitalization for febrile neutropenia or reduced chemotherapy dose intensity. Risk models can be based on pretreatment characteristics alone (unconditional) or include the initial results with treatment (conditional). Some obvious pretreatment factors are advanced age, low performance status, comorbidities, and prior treatment. Additional risk factors found to be useful in risk models in patients with non- Hodgkin's lymphoma include serum albumin levels, lactate dehydrogenase levels, bone marrow involvement, and soluble tumor necrosis factor receptor levels.[33-35] The most common type of conditional model for predicting later neutropenic complications incorporates the patient's early hematologic response to chemotherapy, which can identify patients at increased risk for subsequent neutropenic events who are likely to benefit from CSF support.[32,37] Conditional models are most useful with regimens that are associated with a relatively low average risk in the early cycles of treatment, while unconditional models are more useful when the early risk of neutropenic events is relatively high. An example of an unconditional risk model has recently been presented that identified five independent factors associated with an increased risk of hospitalization for febrile neutropenia: age 65 years or greater, serum albumin levels 3.5 g/dL or less, planned average reduced dose intensity 80% or more, pretreatment ANC 1.5 109/L or less, and hepatic disease. This model was then used to classify patients as being at low, medium, or high risk according to how many risk factors were present. Economic evaluations have shown the ability of clinical prediction models to increase cost savings and the cost-effective application of CSF by enabling targeted use in patients at greatest risk and most likely to benefit.[ 29,38] For instance, using a validated conditional model for identifying patients treated with adjuvant chemotherapy for early-stage breast cancer who are at increased risk for future neutropenic events that are likely to compromise dose intensity, Silber and colleagues showed that a strategy of targeting G-CSF in the 50% of patients at greatest risk was associated with an average cost-effectiveness of approximately $34,000 per life-year gained. Evaluating Risk Models Most of the risk models for chemotherapy- induced neutropenia that have been developed to date have been retrospective in design, fail to provide clearly defined hypotheses, and lack important data elements for adequate analysis. Clinical predictive models should be developed from studies that minimize bias by stating the hypothesis in advance and providing clear inclusion and exclusion criteria. Any subgroup analyses that are to be performed should be stated in advance, and any differences in the outcomes between subgroups should be assessed by testing the interaction between the prognostic factor and the variable that defined the subgroups rather than by separate analyses within the subgroups. Studies should be designed to limit the potential for missing data, ideally to less than 10%. The issues of multiple comparisons should be considered when several prognostic factors or cut points are evaluated, and tests of significance should be adjusted accordingly. The pitfalls of stepwise multiple regression models, including model instability, and exaggeration of coefficient estimates (and P values) should be acknowledged. The majority of the studies to date have not been independently validated in a similar but separate patient population. The issue of multiple testing or small sample sizes within subgroups must be properly addressed when apparent differences between subgroups are observed. A risk model should ideally be developed from a prospective, hypothesis- driven study in a large representative patient population. For this purpose, a prospective registry in different tumor types in a large communitybased oncology setting has been created. The registry will make it possible to collect comprehensive patient data, including cycle-by-cycle information on hematologic function. The purpose of the registry is to develop a powerful risk model for neutropenia that is reliable in routine clinical practice.[ 22,39] Conclusions Neutropenia and its complications remain a major challenge in the management of patients with cancer treated with cytotoxic chemotherapy. Colonystimulating factor is effective in reducing neutropenia and its complications across a broad range of disease categories and treatment regimens. The American Society of Clinical Oncology has developed clinical practice guidelines for the use of CSF that were supported by early economic models based on the average risk of febrile neutropenia with various chemotherapy regimens. While updated cost studies support lower cost-savings risk thresholds in the average setting, current efforts have shifted the focus to treatment-, disease-, and patient-specific risk factors for neutropenia and its complications. Early studies of predictive models for identifying patients at increased risk for neutropenic complications show promise for clinical decision making and the effective targeting of CSF use toward those most likely to benefit. Such models have been shown to further reduce the cost associated with CSF support in patients treated with chemotherapy for cancer and to be cost-effective in responsive and potentially curable malignancies by enabling the delivery of full-dose-intensity therapy on time. A multi-institution effort is currently under way to establish a large prospective registry of patients treated with cytotoxic therapy in order to develop accurate and valid risk models to guide clinicians in the optimal use of CSF.[39,40]
2. Yancik R, Ries LA: Aging and cancer in America: Demographic and epidemiologic perspectives. Hematol Oncol Clin North Am 14:17- 23, 2000.
3. Frei E 3rd, Canellos GP: Dose: A critical factor in cancer chemotherapy. Am J Med 69:585-594, 1980.
4. Bonadonna G, Valagussa P, Moliterni A, et al: Adjuvant cyclophosphamide, methotrexate, and fluorouracil in node-positive breast cancer: The results of 20 years of follow-up. N Engl J Med 332:901-906, 1995.
5. Budman DR, Berry DA, Cirrincione CT, et al: Dose and dose intensity as determinants of outcome in the adjuvant treatment of breast cancer. The Cancer and Leukemia Group B. J Natl Cancer Inst 90:1205-1211, 1998.
6. Kwak LW, Halpern J, Olshen RA, et al: Prognostic significance of actual dose intensity in diffuse large-cell lymphoma: Results of a tree-structured survival analysis. J Clin Oncol 8:963-977, 1990.
7. Epelbaum R, Faraggi D, Ben-Arie Y, et al: Survival of diffuse large cell lymphoma. A multivariate analysis including dose intensity variables. Cancer 66:1124-1129, 1990.
8. Lepage E, Gisselbrecht C, Haioun C, et al: Prognostic significance of received dose intensity in non-Hodgkin’s lymphoma patients: Application to LNH-87 protocol. The GELA (Groupe d’Etude des Lympomes de l’Adulte). Ann Oncol 4:651-656, 1993.
9. Bodey GP, Buckley M, Sathe YS, et al: Quantitative relationships between circulating leukocytes and infection in patients with acute leukemia. Ann Intern Med 64:328-340, 1966.
10. Schimpff SC: Empiric antibiotic therapy for granulocytopenic cancer patients. Am J Med 80:13-20, 1986.
11. Hughes WT, Armstrong D, Bodey GP, et al: 2002 guidelines for the use of antimicrobial agents in neutropenic patients with cancer. Clin Infect Dis 34:730-751, 2002.
12. Kuderer NM, Cosler L, Crawford J, et al: Cost and mortality associated with febrile neutropenia in adult cancer patients (abstract 998). Proc Am Soc Clin Oncol 21:250a, 2002.
13. Gandhi SK, Arguelles L, Boyer JG: Economic impact of neutropenia and febrile neutropenia in breast cancer: Estimates from two national databases. Pharmacotherapy 21:684- 690, 2001.
14. Calhoun EA, Chang C-H, Welshman EE, et al: Evaluating the total costs of chemotherapy- induced toxicity: Results from a pilot study with ovarian cancer patients. The Oncologist 6:441-445, 2001.
15. Lyman GH, Kuderer NM: Filgrastim in patients with neutropenia: Potential effects on quality of life. Drugs 62(suppl 1):65-78, 2002.
16. Fortner BV, Stolshek B, Schwartzberg LS, et al: Decline in absolute neutrophil count (ANC) is associated with lower quality of life in cancer patients receiving docetaxel (abstract 2808). Proc Am Soc Clin Oncol 21:247b, 2002.
17. Okon TA, Fortner BV, Schwartzberg L, et al: Quality of life (QOL) in patients with grade IV chemotherapy-induced neutropenia (CIN) (abstract 2920). 21:275b, 2002.
18. Calhoun EA, Chang C-H, Welshman EE, et al: A neutropenia-specific quality of life instrument: Rationale for the development of the FACT-N (abstract 1498). Proc Am Soc Clin Oncol 21:375a, 2002.
19. Link BK, Budd GT, Scott S, et al: Delivering adjuvant chemotherapy to women with early-stage breast carcinoma: Current patterns of care. Cancer 92:1354-1367, 2001.
20. Lyman GH, Dale D, Crawford J: Incidence and predictors of low dose intensity in adjuvant breast cancer chemotherapy: A nationwide study of community practices. J Clin Oncol. In press.
21. Chrischilles E, Link B, Scott S, et al: Factors associated with early termination of CHOP, and its association with overall survival among patients with intermediate-grade non- Hodgkin’s lymphoma (NHL) (abstract 1539). Proc Am Soc Clin Oncol 21:385a, 2002.
22. Dale D, Crawford J, Lyman GH: Myelotoxicity and dose intensity of chemotherapy: Reporting practices from randomized clinical trials. J Natl Compr Cancer Netw 1:440-454, 2003.
23. Lyman GH, Kuderer NM, Djulbegovic B: Prophylactic granulocyte colony-stimulating factor in patients receiving dose intensive cancer chemotherapy: A meta-analysis. Am J Med 112:406-411, 2002.
24. Crawford J, Ozer H, Stoller R, et al: Reduction by granulocyte colony-stimulating factor of fever and neutropenia induced by chemotherapy in patients with small-cell lung cancer. N Engl J Med 325:164-170, 1991.
25. Trillet-Lenoir V, Green J, Manegold C, et al: Recombinant granulocyte colony stimulating factor reduces the infectious complications of cytotoxic chemotherapy. Eur J Cancer 29A:319-324, 1993.
26. Ozer H, Armitage JO, Bennett CL, et al: 2000 update of recommendations for the use of hematopoietic colony- stimulating factors: Evidence-based, clinical practice guidelines. American Society of Clinical Oncology Growth Factors Expert Panel. J Clin Oncol 18:3558- 3885, 2000.
27. Lyman GH, Lyman CG, Sanderson RA, et al: Decision analysis of hematopoietic growth factor use in patients receiving cancer chemotherapy. J Natl Cancer Inst 85:488-493, 1993.
28. Lyman GH: A predictive model for neutropenia associated with cancer chemotherapy. Pharmacotherapy 20(7 pt 2):104S-111S, 2000.
29. Lyman GH, Kuderer NM, Crawford J, et al: Economic impact of pegfilgrastim use based on the risk of febrile neutropenia (FN) in NHL patients treated with CHOP (abstract 2384). Proc Am Soc Clin Oncol 22:593, 2003.
30. Gomez H, Hidalgo M, Casanova L, et al: Risk factors for treatment-related death in elderly patients with aggressive non-Hodgkin’s lymphoma: Results of a multivariate analysis. J Clin Oncol 16:2065-2069, 1998.
31. Caggiano V, Stolshek B, Delgado D, et al: First and all cycle febrile neutropenia hospitalizations (FNH) and costs in intermediate grade non-Hodgkin’s lymphoma (IGL) patients on standard-dose CHOP therapy (abstract 1810). Blood 98:431a, 2001.
32. Silber JH, Fridman M, DiPaola RS, et al: First-cycle blood counts and subsequent neutropenia, dose reduction, or delay in earlystage breast cancer therapy. J Clin Oncol 16:2392-2400, 1998.
33. Intragumtornchai T, Sutheesophon J, Sutcharitchan P, et al: A predictive model for life-threatening neutropenia and febrile neutropenia after the first course of CHOP chemotherapy in patients with aggressive non- Hodgkin’s lymphoma. Leuk Lymphoma 37:351-360, 2000.
34. Tompkins KA, Imrie KR: Neutropenic events in patients receiving CHOP chemotherapy for large cell non-Hodgkin’s lymphoma: Predicting who is at risk (abstract 1427). Blood 98:337a-338a, 2001.
35. Voog E, Bienvenu J, Warzocha K, et al: Factors that predict chemotherapy-induced myelosuppression in lymphoma patients: Role of tumor necrosis factor ligand-receptor system. J Clin Oncol 18:325-331, 2000.
36. Lyman GH, Morrison VA, Dale DC, et al: Risk of febrile neutropenia among patients with intermediate-grade non-Hodgkin’s lymphoma receiving CHOP chemotherapy. Leuk Lymphoma 44(12):2069-2076, 2003.
37. Blay JY, Chauvin F, Le Cesne A, et al: Early lymphopenia after cytotoxic chemotherapy as a risk factor for febrile neutropenia. J Clin Oncol 14:636-643, 1996.
38. Silber JH, Fridman M, Shpilsky A, et al: Modeling the cost-effectiveness of granulocyte colony-stimulating factor use in early-stage breast cancer. J Clin Oncol 16:2435-2444, 1998.
39. Lyman GH, Crawford J, Dale DC, et al: Predicting the risk of chemotherapy-induced neutropenia (CIN) in patients with breast cancer: Rationale for prospective risk model development (abstract 537). Breast Cancer Res Treat 76(1):S126, 2002.
40. Dale DC, Wolff D, Agboola O, et al: Development of a risk model for neutropenic complications based on a prospective nationwide registry (abstract 2229). Proc Am Soc Clin Oncol 22:554, 2003.