Dr. Hornberger's recent and current funding includes projects sponsored by county, state, and federal government agencies (National Institutes of Health, Agency for Health Research and Quality, Department of Veterans Affairs, Health Resources and Services Administration, and the U.S. Office of Technology Assessment), professional medical societies (Renal Physicians Association), nonprofit agencies and foundations (Picker Foundation and Salt Lake City Health Access Program), and for-profit companies (Roche Pharmaceuticals, Roche Laboratories, and Genentech, Inc.).
Cost utility in the United States of rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone for the treatment of elderly patients with diffuse large B-cell lymphoma†
Article first published online: 8 MAR 2005
Copyright © 2005 American Cancer Society
Volume 103, Issue 8, pages 1644–1651, 15 April 2005
How to Cite
Hornberger, J. C. and Best, J. H. (2005), Cost utility in the United States of rituximab plus cyclophosphamide, doxorubicin, vincristine, and prednisone for the treatment of elderly patients with diffuse large B-cell lymphoma. Cancer, 103: 1644–1651. doi: 10.1002/cncr.20956
The development of the model and the subsequent analyses were supported by a grant from Genentech, Inc. The sponsors of the study had no role in the model design, cost-utility analysis, or interpretation of the results. All investigators had unlimited access to the model and data. No limitations on publication were imposed. The authors made the final decisions on all analyses presented in the article.
- Issue published online: 4 APR 2005
- Article first published online: 8 MAR 2005
- Manuscript Accepted: 10 DEC 2004
- Manuscript Revised: 9 NOV 2004
- Manuscript Received: 17 AUG 2004
- Genentech, Inc.
- combined cyclophosphamide;
- prednisone (CHOP)
Findings from the Groupe d'Etude des Lymphomes Adultes LNH 98-5 study showed that rituximab added to combined cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) prolonged progression-free survival and overall survival in adults age ≥ 60 years with diffuse large B-cell non-Hodgkin lymphoma (DLBCL). The current study was conducted to investigate the incremental cost utility of the addition of rituximab to CHOP (R-CHOP) compared with CHOP alone.
Clinical prognosis of the time to disease progression and death was estimated using published evidence from the LNH 98-5 study (n = 399 patients) that was linked mathematically to published long-term outcome data on patients with DLBCL. Drug-acquisition costs were based on published data from formulary pricing sources, and the costs of cancer surveillance and end-of-life care were based on published literature sources. The authors assessed cost utility as the difference in costs between R-CHOP and CHOP divided by the increase in expected overall survival adjusted for quality of life.
Over 5 years, it was projected that R-CHOP would prolong overall survival by 1.04 years. The mean cumulative cost of CHOP was $3358, and the mean cost of R-CHOP was $17,225, resulting in a cumulative net increase of $13,867. The posttreatment cancer surveillance cost for CHOP was $3950, compared with $5202 for R-CHOP. It was estimated that R-CHOP would have a cost-utility ratio of $19,297 per year of life gained compared with CHOP when adjusted for quality of life. R-CHOP remained cost effective over wide ranges of variables in sensitivity analyses.
Compared with CHOP alone, it was predicted that R-CHOP would be cost effective in elderly patients with DLBCL. Cancer 2005. © 2005 American Cancer Society.
Diffuse large B-cell lymphoma (DLBCL) is reported to be the most common aggressive non-Hodgkin lymphoma (NHL) seen in the U.S. In 2000, > 70,000 patients in the U.S. received a new diagnosis of NHL, and approximately 39% of those patients had DLBCL.1 When it is left untreated, DLBCL results in expected survival that is measured in months.2
Cyclophosphamide combined with mechlorethamine, vincristine, procarbazine, and prednisone chemotherapy and combined cyclophosphamide, doxorubicin, vincristine, and prednisone (CHOP) chemotherapy, which were introduced in the 1970s, represented significant advances in the treatment of DLBCL with complete response rates of ≈ 50%, extending mean overall survival by > 3 years.2 However, despite extensive research over nearly 3 decades, it has proved difficult to identify new regimens that provide significant improvements over CHOP in terms of disease-free survival or overall survival.2 Moreover, patients age ≥ 60 years, who constitute > 50% of the patients who present with DLBCL, not only experience lower rates of response but are less able to tolerate chemotherapy-associated toxicities, including cytopenias.3 Growth-factor support given along with CHOP, as a consequence, has become a common adjunct to treatment for elderly patients.4
Rituximab emerged in the late 1990s as a novel treatment for the management of patients with indolent or previously untreated, aggressive lymphomas.5 The mechanisms by which rituximab kills lymphoma cells are not understood fully, but they appear to involve antibody-dependent, cell-mediated cytotoxicity and apoptosis. When rituximab was given in addition to the CHOP regimen in a Phase III trial of 399 elderly patients with DLBCL, it increased the complete response rate significantly, from 63% to 76%,6 and prolonged event-free survival and overall survival (i.e., there was an absolute 20% reduction in risk of death at 3 years).
Although rituximab represents a significant clinical advance in the treatment of patients with DLBCL, questions among policymakers may arise regarding its affordability and appropriate use. Specifically, after some success in the middle-to-late 1990s to curtail the rapid growth of health care expenditures in the U.S., there has been a return in recent years to substantially increased growth of health expenditures, similar to that seen in the 1980s and early 1990s.7 Correspondingly, there has been renewed interest in assessing the value of health care spending as part of evidence-based decision making. Some have suggested using cost-utility analyses to evaluate the relative value of health care interventions compared with their next best alternatives.8, 9 To date, there has been no published report on the relative cost and benefit of rituximab given in addition to CHOP (R-CHOP) for any indication, as assessed in a cost-utility analysis specific to the U.S. health care system. The objective of the current study was to estimate the incremental cost utility of a strategy of R-CHOP compared with CHOP alone in patients age ≥ 60 years with previously untreated DLBCL.
MATERIALS AND METHODS
A Markov state-transition model was developed to predict sequence of treatments, outcomes, and costs for patients receiving R-CHOP or CHOP (Fig. 1). From the start of treatment, the model follows a representative patient who has characteristics typical of the population enrolled in LNH 98-5: ages 60–80 years with Ann Arbor Stage II, III, or IV disease and with a performance status of 0–2 according to the criteria of the Eastern Cooperative Oncology Group.10 The model contains five health states. First, patients begin treatment with CHOP or R-CHOP while they are free of events, in which events may include progression of lymphoma, the patient declining further treatment, or intercurrent illness or adverse event.10 Patients may continue in this initial event-free state after the completion of chemotherapy. Second, after an event, salvage treatment is an option that consists of chemotherapy with or without radiation therapy.11 Third, for some patients, transplantation may follow chemotherapy. Fourth, patients who fail to respond to salvage therapy enter a phase of end-of-life care. Death represents the fifth state of the Markov model.
We estimated the incremental cost per year of life saved and per year of quality-adjusted year of life (QALY) saved from a U.S. societal perspective. Quality of life, or utility, associated with a health state typically ranges between 0 (for death) and 1 (for perfect health or best attainable health). The contribution of a health state to QALY is estimated as the time spent in that health state multiplied by the utility of that health state.
The model was designed to mimic the use of CHOP and R-CHOP as it is prescribed commonly in the U.S. Specifically, CHOP is comprised of cyclophosphamide at a dose of 750 mg/m2 on Day 1, doxorubicin at dose of 50 mg/m2 on Day 1, vincristine at a dose of 1.4 mg/m2 up to a dose of 2 mg on Day 1, and prednisone at a dose of 40 mg/m2 per day for 5 days. In the LNH 98-5 study, 8 cycles of CHOP were to be given at 3-week intervals.6 However, in the U.S., CHOP commonly is administered for six cycles with the objective of two cycles of treatment beyond clinical remission; most patients respond after four cycles of CHOP.12 This pattern of early response also was observed in the CHOP arm of LNH 98-5; for a U.S.-based model, we assumed similar efficacy for a 6-cycle standard dose CHOP regimen like that observed in LNH 98-5.
R-CHOP is comprised of CHOP plus rituximab at a dose of 375 mg/m2 on Day 1 of each of the 6 cycles of CHOP. In LNH 98-5, doses of cyclophosphamide and doxorubicin were adjusted for toxicity; there was no reported difference in toxicity that led to dose reductions in the 2 arms.6 In addition, in the trial, the doses of rituximab were not modified, rituximab was discontinued when CHOP was stopped, and treatment was stopped if disease progression occurred.
Event-Free and Overall Survival
Event-free and overall survival data for the first 4.5 years after initiating treatment were derived from the published Kaplan–Meier estimates of the European pivotal Phase III clinical trial: the LNH 98-5 study.6 In LNH 98-5, > 50% of patients in either arm remained alive at 3 years, and survival data were available up to 4.5 years. We extrapolated event-free and overall survival beyond the trial's 4.5 years of follow-up based on transitions rates obtained from other sources. We initially sought information from publicly available, nationally representative registries for our target population. Because no such registry exists, as an alternative information resource, we used published findings of the International Non-Hodgkin's Lymphoma Prognostic Factors Project.13 Findings from this project continue to be the most comprehensive source of information on long-term survival among patients with DLBCL.
After 4.5 years, we assumed that transition rates were identical between treatments (i.e., regardless of whether the patient received CHOP or R-CHOP). For example, based on data from the International Non-Hodgkin's Lymphoma Prognostic Factors Project, the annual mortality rate at 5 years was 14.7%. We validated the use of these data by applying only the 2-year median follow-up trial data from LNH 98-5 and extrapolating event-free and overall survival to 3–4 years, finding that the extrapolations accurately predicted actual survival probabilities reported in subsequent trial updates.
Resource Use and Costs
LNH 98-5 was conducted in France, Switzerland, and Belgium; hence, we deemed it unreasonable to apply medical resource use data to estimate medical resource use or costs in the U.S. Instead, we used U.S.-based sources of data to estimate costs for drugs, cancer surveillance, and end-of-life care. We obtained estimates of the average wholesale price per mg of each chemotherapeutic agent in CHOP from the 2003 Physician's Desk Reference Red Book (see Table 1).14
|Courses and cost of chemotherapy|
|Mean no. courses of CHOP||6||6||NA||NA||Freedman and Friedberg, 200312|
|Mean no. of rituximab infusions||6||NA||NA||Freedman and Friedberg, 200312|
|Drug cost per CHOP cycle||$560||$560||NA||NA||PDR Staff, 200314|
|Drug cost per rituximab infusion||$2871||NA||NA||PDR Staff, 200314|
|Cost of annual posttreatment surveillance||$1800||$1800||Truncated normal||(μ, σ) = (1800,725)||Weeks et al., 199115; Edelman et al., 199735|
|Cost for end-of-life care||$30,000||$30,000||Truncated normal||(μ, σ) = (30000,4500)||Hoover et al., 200218; Hogan et al., 200119|
|Cost of salvage treatment Transplantation||$45,000||$45,000||Truncated normal||(μ,σ) = (45000,45000)||Stockerl-Goldstein et al., 200011;Waters et al., 198816; Rizzo et al., 199917|
|Other salvage regimen||$3358||$3358||Truncated normal||(μ, σ) = (3358,4500)||Freedman and Friedberg, 200412|
|Probability of salvage||100%||100%||β||(α, β) = (4, 0.4)||IBMTR and ABMTR, 200436|
|Probability of transplantation, if receives salvage||10%||10%||β||(α, β) = (0.25, 2.25)||IBMTR and ABMTR, 200436|
|Event-free||0.83||0.83||Uniform||(l,u) = (0.66, 1.0)||Doorduijn et al., 200120|
|End-of-life||0.38||0.38||Uniform||(l,u) = (0.30, 0.46)||Doorduijn et al., 200120|
|Salvage or transplantation||0.83||0.83||Uniform||(l,u) = (0.66, 1.0)||Doorduijn et al., 200120|
|Time discount rate||3%||3%||Uniform||(l,u) = (0, 9%)||Lipscomb et al., 199624|
|Time horizon (yr)||5||5||Uniform||(l,u) = (3, 7)||Weinstein et al., 199637|
We derived per-visit costs of care (complete blood count, chest X-ray, etc.) from a retrospective review of 139 patients with large cell non-Hodgkin lymphoma that was conducted by Weeks et al.15 It was assumed that the estimated cost of this surveillance strategy was $1800 per year. Based on 2000 Surveillance, Epidemiology, and End Results data from the National Cancer Institute, ≈ 10% of patients who receive salvage also undergo transplantation. It was estimated recently that transplantation costs vary between $45,000 and $89,00011 and have been reported as high as $150,000 in some settings, with the ability to mobilize peripheral blood progenitor cells identified as an important predictor of medical resource use and costs.16, 17
A search of the medical literature revealed no published data on the cost of end-of-life care for patients with DLBCL. For a proxy, we used data from elderly Medicare beneficiaries, in whom the cost of the last 6 months of life varies between $23,000 and $37,000.17, 18 In the base-case analysis, we applied the midrange value, $30,000, as the cost for end-of-life care in the last 6 months of life.
Quality of Life/Utilities
R-CHOP was tolerated well and did not contribute to serious adverse events associated with CHOP.6 R-CHOP had a higher incidence of infusion-related events, which resolved rapidly.6 Otherwise, rituximab appeared to have no significant adverse effects on short-term quality of life or psychologic mood state.19 We adjusted survival for quality of life based only on cancer stage using data from the report by Doorduijn et al., in which survival without events was assigned a utility of 0.83, and end-of-life care after recurrence was assigned a utility of 0.38.21
Other Policy Assumptions
We used 5 years as the longest duration of follow-up for 4 reasons. First, a survival advantage was observed for patients who were randomized to receive R-CHOP at the time of last follow-up in the LNH 98-5 study. Second, evidence from the International Non-Hodgkin's Lymphoma Prognostic Factors Project suggests that cure is possible for a large proportion of patients who have survived beyond 2–3 years after initiating therapy.21 Third, greater than half of elderly patients with DLCLB are expected to survive beyond 5 years.21 Fourth, international longitudinal data bases of NHL have followed elderly patients with DLBCL for at least 5 years and often up to 10 years and 15 years.22 In the current sensitivity analyses, we explored the implications of shorter (with 3 years representing the most recently published update of event-free survival and overall survival) and longer (e.g., 10 years) durations of follow-up based on published economic guidelines.23
A common assumption of economic modeling is that policymakers should value therapies that provide benefits to patients in the short term.24 Using standard methodologies, we discounted both costs and survival by a fixed annual rate of 3%.
Planned Sensitivity and Probabilistic Analyses
We conducted extensive, one-way sensitivity and subgroup analyses on the variables in the model to determine which assumptions had the greatest impact on the conclusions in the model. The variables we assessed included costs (CHOP, surveillance, salvage and transplantation, end-of-life care), the probability of salvage therapy, quality of life, time horizon, and the time discount rate. We also conducted a probabilistic analysis of selected variables, in which more than one variable is changed to simulate random error. Distributions and ranges used in these analyses are shown in Table 1.
As noted in the LNH 90-5 study,6 the current model shows that a complete response or an unconfirmed complete response was achieved in 76% of the patients treated with R-CHOP, compared with 63% of patients treated with CHOP alone. The percent of patients with event-free survival at 3 years was 35% with CHOP and 53% with R-CHOP. There was an 11% absolute risk reduction in overall survival at 3 years (51% with CHOP and 62% with R-CHOP).
It is predicted that R-CHOP will increase the mean event-free survival by 0.90 years, from 2.19 years with CHOP to 3.10 years, and will increase overall survival by 1.04 years, from 3.11 years with CHOP to 4.15 years, with R-CHOP. The mean survival adjusted for time discounting and quality of life increased with R-CHOP by 0.66 years (Table 2).
|Mean duration (yrs)|
|Per yr of life gained||$12,304|
|Per quality-adjusted yr of life gained||$19,297|
Costs and Cost Utility
The mean cost of standard-dose CHOP administered for 6 cycles equaled $3358. The mean cost of 6 cycles of rituximab equaled $17,225 (Table 2). The estimated cost of posttreatment cancer surveillance over 5 years for patients who receive CHOP alone is $3950. Patients who receive R-CHOP have a higher rate of response and are projected to survive longer; hence, the cancer surveillance cost increased by $1252 to $5202. By delaying the time to end of life and the requirement for palliative care, these costs declined from $18,228 to $13,725, for a net difference of $4503. The cost of salvage treatment also declined by $1240 with R-CHOP. Over a 5-year time horizon, we project an incremental cost per year of life gained with R-CHOP, compared with CHOP, equal to $12,304 and an incremental cost per QALY equal to $19,297.
Sensitivity and Probabilistic Analyses
We found little impact on the cost-utility ratio of the time discount rates, costs of cancer surveillance, or costs of end-of-life care (Table 3). Using the time horizon of 3 years associated with the last published update increased the cost-utility ratio to $46,942. Assuming a longer time horizon of 10 years, the estimated average total costs were $34,734 and $49,144 for CHOP and R-CHOP, respectively. The net difference in quality-adjusted survival increased to 0.97 years, such that the cost per QALY gained declined to $14,888.
|Variable (base case)a||Cost-effectiveness (QALY method)|
|Cycles (6 cycles each)|
|CHOP (7.1 cycles for CHOP, 7.5 cycles for R-CHOP)|
|Rituximab (7.4 cycles)||$25,727|
|Cost of rituximab ($2871 per cycle)|
|$2584 (10% lower)||$16,687|
|$3158 (10% higher)||$21,907|
|Cost of surveillance ($1800)|
|Cost of end-of-life care ($30,000)|
|Cost of nontransplantation salvage care ($3358)|
|Cost of transplantation ($45,000)|
|Probability of any salvage treatment (100%)|
|Probability of transplantation (15%)|
|Utilities, changing only postevent (0.38)|
|Discount rate (3%)|
|Time horizon (5 yrs)|
We examined the effect of intending to use 8 cycles of CHOP or R-CHOP, as occurred in LNH 98-5. In that trial, patients who were assigned to CHOP received 7.1 cycles, whereas patients who were assigned to R-CHOP received 7.5 cycles of CHOP and 7.4 cycles of rituximab. Assuming no change in effectiveness, the cost per adjusted year of survival increased to $25,727. Varying the cost of CHOP by ± 10% had no effect on cost per QALY gained. Varying the cost of rituximab by 10% changed the cost of quality-adjusted survival by 9.1%.
It is predicted that the higher rate of response and lower rate of events with R-CHOP will reduce expenditures for salvage therapy. For example, at a relatively low cost of $80,833 for autologous transplantation (e.g., patients who are good mobilizers of peripheral blood progenitor cells11), the cost utility of R-CHOP declined to $18,401. At a relatively high cost of $140,224 for autologous transplantation (e.g., patients who are poor mobilizers of peripheral blood progenitor cells11), the cost utility of R-CHOP declined further to $16,917. In probabilistic analyses, the 5th and 95th percentiles of the cost-utility distribution were $13,343 and $25,477, respectively.
Rituximab added to a CHOP regimen significantly increases the percentage of patients who achieve a complete response and prolongs event-free and overall survival for elderly patients with DLBCL.6 R-CHOP and CHOP alone have a similar incidence of serious adverse events.
The economic analyses presented here show that the addition of rituximab to a CHOP regimen costs ≈ $12,000 per year of life gained and < $20,000 per QALY gained over a time horizon of 5 years. In a comprehensive review of 40 studies with 89 cost-utility analyses in oncology, Earle et al. showed that > 30% of those analyses had an estimated cost utility ≥ $50,000.25 Among 6 cited studies of interventions for hematologic malignancies,26–32 cost-utility ratios ranged from a cost saving for autologous bone marrow transplantation in patients ages 15–60 years with intermediate-grade or high-grade NHL30 to > $7,000,000 for intravenous immunoglobulin administration in patients with chronic lymphocytic leukemia and hypo-γ-globulinemia.32
Policymakers sometimes are skeptical about findings from economic analyses. Therefore, we developed the current model intending to use assumptions that favored CHOP over R-CHOP. First, we assumed the same quality-of-life associated with lymphoma regardless of whether the patient received R-CHOP or CHOP alone. This assumption omits the effects of R-CHOP on increasing response rates and prolonging the time to recurrence,6 which should translate into increased time for patients without toxicities and disease-related symptoms. Second, the clinical trial data also showed a trend toward a lower rate of hospitalizations associated with R-CHOP, presumably due to a lower recurrence rate, which was considered in neither the base-case model nor in the sensitivity analyses. Third, the model follows patients for 5 years, and it is expected that > 30% of patients will remain alive at that time. However, published guidelines recommend extending the time horizon of the economic evaluation to the end of life.23, 24 This recommendation is of particular relevance for R-CHOP, because most of the costs are incurred within the first 6 months, but the benefit accrues over a longer time (e.g., as shown in Fig. 2, the improvement in survival persisted at the longest follow-up observed in the trial). Finally, it is well established that cure of lymphoma is achievable in DLBCL. A consensus does not exist about how long patients should be followed before they are considered cured, with times varying between 2 years and 4 years.33 The ability of rituximab to increase response rates and prolong disease-free survival suggests that higher rates of cure may be achievable. Data presented at the 2003 annual meeting of the American Society of Clinical Oncology suggested that survival benefits associated with rituximab persist up to 3 years after the initiation of therapy.34
The clinical evidence used in this model was based on a well designed, randomized clinical trial that was linked mathematically to a large observational data base and has provided important insights to the classification of DLBCL and improved estimation of prognosis to aid treatment decisions. As with most policy models, the assumptions used in the current study were based on the average results for patients enrolled in the trial and on the trial protocol. Hence, the findings of this study should be interpreted with attention to patient characteristics and actual clinical practice. For example, the distribution of International Prognostic Index risk scores among patients in different plans may vary from the scores of patients enrolled in LNH 98-5. In addition, LNH 98-5 was designed to assess the implications of using rituximab with a planned course of 8 cycles of CHOP. The effects of rituximab used in combination with other regimens are under investigation.
Research currently is ongoing to address other key questions regarding the use of rituximab in the management of patients with DLBCL. The analysis presented herein pertains to elderly patients ages 60–80 years. Trials currently are enrolling patients age < 60 years to determine whether the clinical benefits of R-CHOP seen in the LNH 98-5 study also apply to this younger cohort. While waiting for the results of those trials, it may be possible to model the expected long-term costs based on the projected benefits described in the trial protocol.
In the interim, to our knowledge LNH 98-5 is the first study in more than 20 years to demonstrate a statistically significant benefit of a new treatment regimen in response rate, disease-free survival, and overall survival compared with the CHOP regimen. In a short period, R-CHOP has become a widely adopted treatment for elderly patients with DLBCL. Not only has this ushered in a new era in the management of DLBCL, but several other regimens also are showing promise in prolonging survival. We believe it is too soon to say how the various regimens may be used with rituximab or how cost effective they will be. Our analyses show that economics should not be a basis for denying the appropriate use of R-CHOP in the treatment of elderly patients with DLBCL.
The authors appreciate the suggestions of two anonymous reviewers.
- 1American Cancer Society. Cancer facts and figures. Atlanta: American Cancer Society, Inc., 2002.
- 9Cost-effectiveness and evidence evaluation as criteria for coverage policy. Health Aff (Millwood). 2004 (Suppl Web Exclusives): W4-284–W4-296..
- 122005 UpToDate® online 12.3. Available from URL: http://www.uptodate.com [accessed December 13, 2004]., . Treatment of highly aggressive non-Hodgkin's lymphomas.
- 14Physican's Desk Reference Staff. 2003 Drug topics (R) red book. Montvale, NJ: Thomson PDR, 2003.
- 20http://www.asco.org/ac/1,1003,_12-002643-00_18-0016-00_19-002639,00asp [accessed December 13, 2004]., , , et al. Quality-of-life (QoL) in patients treated with rituximab as a single agent [abstract 2693]. Presented at the 2002 Annual Meeting of the American Society of Clinical Oncology. Available from URL:
- 21Quality of life (QOL) in elderly patients with aggressive non-Hodgkin's lymphoma (NHL) treated with CHOP [abstract 1803]. Blood. 2001; 98: 430a., , , , .
- 22Hodgkin's disease and non-Hodgkin's lymphoma. In: BlackR, StocktonD, editors, for the Scottish Executive Health Department. Cancer scenarios: an aid to planning cancer services in Scotland in the next decade. Edinburgh: The Scottish Executive, 2001. Chapter sections in HTML format available from URL: http://www.show.scot.nhs.uk/sehd/publications/csatp/csatp-20.htm [accessed December 13, 2004]..
- 24Time preference. In: GoldMR, SiegellJE, RussellL, WeinsteinM, editors. Cost-effectiveness in health and medicine. New York: Oxford University Press, 1996: 214–246., , .
- 34GELA. GELA study comparing CHOP and R-CHOP in elderly patients with DLCL: 3-year median follow-up with an analysis according to co-morbidity factors. Presented at the 2003 Annual Meeting of the American Society of Clinical Oncology. Abstract and Microsoft® Powerpoint slides available from URL: http://www.asco.org/ac/1,1003,_12-002511-00_18-0023-00_19-001570,00.asp [accessed December 13, 2004]., , et al. for
- 36Report on State of the Art in Blood and Marrow Transplantation: The 2003 BMT summary slides with guide. From the Center for International Blood and Marrow Transplant Research. Newsletter 2004; 10(2). PDF and Microsoft® Powerpoint slides available from URL: http://www.ibmtr.org/newsletter/newsletter_sums.html [accessed December 13, 2004].