|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
© 2001 American Society for Clinical Oncology Incidence, Cost, and Outcomes of Bleeding and Chemotherapy Dose Modification Among Solid Tumor Patients With Chemotherapy-Induced ThrombocytopeniaFrom the Department of Health Services Research, The University of Texas M.D. Anderson Cancer Center, Houston, TX. Address reprint requests to Linda S. Elting, DrPH, Department of Health Services Research, The University of Texas M.D. Anderson Cancer Center, 1515 Holcombe Blvd, Box 40, Houston, TX 77030; email: lelting{at}mdanderson.org
PURPOSE: To describe the incidence and outcomes of bleeding and chemotherapy dose modifications associated with chemotherapy-induced thrombocytopenia (platelets < 50,000/µL). PATIENTS AND METHODS: Six hundred nine patients with solid tumors or lymphoma were followed-up during 1,262 chemotherapy cycles complicated by thrombocytopenia for development of bleeding, delay or dose reduction of the subsequent cycle, survival, and resource utilization. The association between survival and bleeding or dose modification was examined using the Cox proportional hazards model. Predisposing factors were identified by logistic regression. RESULTS: Bleeding occurred during 9% of cycles among patients with previous bleeding episodes (P < .0001), baseline platelets less than 75,000/µL (P < .0001), bone marrow metastases (P = .001), poor performance status (P = .03), and cisplatin, carboplatin, carmustine or lomustine administration (P = .0002). Major bleeding episodes resulted in shorter survival and higher resource utilization (P < .0001). Chemotherapy delays occurred during 6% of cycles among patients with more than five previous cycles (P = .003), radiotherapy (P = .03), and disseminated disease (P = .04). They experienced similar clinical outcomes but used significantly more resources. Dose reductions occurred during 15% of cycles but were not associated with poor clinical outcomes or excess resource utilization. Significantly shorter survival and higher resource utilization were observed among the 20% of patients who failed to achieve an adequate response to platelet transfusion. CONCLUSION: The incidence of bleeding is low among solid tumor patients overall but exceeds 20% in some subgroups. These subgroups are easily identifiable using routinely available clinical information. A clinical prediction rule is being developed. Poor response to platelet transfusion is a clinically and financially significant downstream effect of thrombocytopenia and warrants further investigation.
SINCE the relationship between hemorrhage and the depth of thrombocytopenia was first described among patients with acute leukemia in 1960,1 its importance to patients with solid tumors has been debated. Although this relationship was confirmed by Belt et al in 19782 and later by Dutcher et al,3 it was clear that hemorrhage was far less likely among patients with solid tumors than among their counterparts with acute leukemia. It was suggested that this decreased risk of hemorrhage did not justify the costs or risks of platelet transfusion at a threshold of 20,000 platelets/µL among patients with solid tumors.4-9 Nevertheless, practice evolved toward use of a threshold of 20,000 platelets/µL below which prophylactic transfusions were administered.10,11 Two recent developments have added new fuel to this longstanding debate. First, three trials have provided evidence that the 20,000-platelet threshold may be overly conservative for patients with leukemia.12-14 Given the lower rate of bleeding observed among patients with solid tumors, re-examination of the potential benefits of prophylaxis is justified. Furthermore, platelet growth factors are currently being tested in clinical and preclinical trials. One of these, interleukin-11, is commercially available. Studies to date suggest that although these agents may be quite effective,15-17 they are not without side effects and their cost will likely exceed the cost of platelet transfusions. It is unclear whether the potential cost of these agents is justified by the risk of serious clinical outcomes of profound or prolonged thrombocytopenia among solid tumor patients. There are insufficient data from patients who have received modern chemotherapy regimens to address these issues; the last comprehensive study of thrombocytopenia among patients with lymphoma or solid tumors was published in 1984.3 For that reason, we studied episodes of chemotherapy-induced thrombocytopenia, particularly those that resulted in major or minor bleeding, a delay of more than 7 days in the subsequent cycle of chemotherapy, or a reduction in the dose of the subsequent cycle. We focused on three related questions: How frequently are bleeding episodes or chemotherapy dose modification associated with thrombocytopenia among patients with solid tumors? Are the outcomes of these events sufficiently serious and/or costly to justify prophylaxis? If so, which patients are at sufficiently high risk to justify the use of prophylactic platelet transfusions or growth factors?
A retrospective cohort consisting of a random sample of 609 patients with solid tumors or lymphoma, stratified by underlying neoplasm, was selected from among all patients who developed chemotherapy-induced thrombocytopenia between January 1, 1994, and December 31, 1995. This cohort was followed-up through December 31, 1996, for development of clinically significant thrombocytopenia and through December 31, 1997, for survival. For each eligible patient, all cycles of chemotherapy that occurred during the study period were included, provided that thrombocytopenia developed. To insure complete ascertainment of events, only patients whose entire care was provided by our institution were included. Patients with leukemia and bone marrow transplant recipients were excluded, as were those who received stem-cell support. After these exclusions, 1,262 cycles with thrombocytopenia were studied in the 609 patients. Fifty-two percent of patients were female, and the median age was 52 years (range, 17 to 87 years). They had received a median of five cycles of chemotherapy in the past (range, 0 to 19 cycles); 92 patients (15%) were chemotherapy-naive. One half (51%) had disseminated disease, whereas 81 (13%) were receiving adjuvant or neoadjuvant chemotherapy in the absence of clinically measurable disease. Patients with lymphoma, sarcoma, breast, and genitourinary cancers contributed the largest number of cycles to the study ( Table 1). Chemotherapy regimens commonly used for these malignancies (fluorouracil, doxorubicin, and cyclophosphamide; cyclophosphamide, vincristine, doxorubicin, and decadron; doxorubicin or etoposide plus cisplatin, cytarabine, and prednisone; cisplatin, cyclophosphamide, and doxorubicin; cyclophosphamide, doxorubicin, vincristine, and prednisone; and doxorubicin plus either ifosfamide or platinum) were predictably frequent. Platinum-based regimens were used in 41% of cycles. With the exception of patients with lymphoma (who received either doxorubicin or etoposide plus methylprednisolone, high-dose cytarabine, and cisplatin) and sarcoma (who received high doses of ifosfamide and doxorubicin), standard doses of chemotherapy were administered in the majority of cases. Overall, 16% of the cycles involved adjuvant or neoadjuvant therapy. However, there were significant differences in this rate, depending on the underlying malignancy. For example, more than 60% of cycles administered to patients with breast cancer involved adjuvant therapy.
Data Sources and Collection All paper and electronic medical records of eligible patients were reviewed. These included the paper medical record, electronic databases containing all hospital and clinic visits, tumor registry and survival data, diagnostic test results, blood products transfused, and pharmaceutical agents prescribed, as well as electronic records of patients enrolled on clinical research protocols. Information about bleeding obtained from these sources was supplemented by data from The Ambulatory Treatment Center and Emergency Center Database, which contains clinical and resource utilization data pertaining to all outpatient chemotherapy and transfusion episodes and all visits to the emergency center. Information from paper sources was transcribed by physician abstractors and information from databases was transferred electronically. A 10-item set of variables (unknown to the abstractors) was collected both manually and electronically for estimation of the frequency of errors with manual transcription (< 3%). Additionally, a predetermined set of key data items was validated by a separate reviewer in all cases to ensure 100% accuracy of critical items.
Definitions The extent of disease dissemination for each cycle was recorded and dichotomized for regression analysis as either limited (no evidence of disease or local disease) or disseminated (one or more site of metastasis). The duration of thrombocytopenia was computed using the last value carried forward method, as is typical in observational studies and clinical trials. Resource utilization was described only during thrombocytopenia for each cycle. Resources used when platelet counts exceeded 50,000/µL were not considered. Performance status was measured on day 1 of each cycle using the Zubrod score.18 Poor performance status was defined as Zubrod score less than 2. The presence of comorbidities was measured on day 1 of each cycle of chemotherapy using the Charlson score.19
Statistical Considerations Duration of survival after bleeding or chemotherapy dose modification was the primary outcome of interest. The median time to survival was computed from the first day of one randomly chosen cycle for each patient in order to provide a sufficient sample of patients who had experienced bleeding or dose modification. The Kaplan-Meier method was used for this analysis. Factors associated with shorter survival were examined using the Cox proportional hazards regression model. For this analysis, factors were entered in the stepwise regression model using the maximum partial likelihood ratio method.
The secondary goal of this study was to generate hypotheses about high-risk groups for future study. For this analysis, demographic and clinical factors were tested first for association with bleeding, chemotherapy delay, and dose reduction in separate univariate analyses (because they are not mutually exclusive events). Two-tailed
How Frequent Are Episodes of Bleeding and Chemotherapy Dose Modifications? Bleeding occurred during 111 cycles (9%) ( Table 2). Most episodes of bleeding were minor, including mild epistaxis during 31 cycles, petechiae and ecchymoses during 19 cycles, and occasional cases of bleeding gums, mild vaginal bleeding, bloody urine, or sputum. All episodes of minor bleeding resolved. Major hemorrhage occurred during 43 cycles (3%). The most common sites were nasal (13 cycles), gastrointestinal hemorrhage, (eight cycles), bladder hemorrhage, (five cycles), and vaginal or pulmonary hemorrhage (four each). Bleeding was significantly more common during cycles complicated by febrile neutropenia (11% v 7%; P = .02). The relationship between major hemorrhage and febrile neutropenia was particularly striking; major hemorrhage occurred during 5% of cycles complicated by febrile neutropenia but only 2% of those without febrile neutropenia (P = .002). Multiple sites of hemorrhage were involved in six episodes. Although CNS tumors or metastases were present during 129 cycles and prior CNS radiation therapy in 77 cycles, only one CNS hemorrhage occurred in a patient who subsequently recovered.
A delay of more than 7 days in the next cycle of chemotherapy occurred during 99 cycles (8%). Most delays (67%) were between 8 and 14 days; however, a delay of more than 30 days occurred during 15 cycles. A reduction in the dose of the subsequent cycle of chemotherapy was far more common (17%). A delay or dose reduction for the subsequent cycle followed 22% of the 111 cycles with episodes of bleeding. Decisions to delay or reduce the dose during a subsequent cycle were multifactorial. Approximately 30% of delays or dose reductions resulted from infection and prolonged or profound granulocytopenia, despite the presence of thrombocytopenia. However, in the remainder, thrombocytopenia contributed to the decision to delay or reduce subsequent doses, and in approximately 30%, prolonged or profound thrombocytopenia was the sole reason for delay or dose reduction.
How Serious Are the Outcomes of Bleeding Episodes and Chemotherapy Dose Modifications?
When the extent of the underlying malignancy and other prognostic factors are considered, striking differences emerge. For example, the difference in survival after major bleeding episodes was limited to patients with disseminated underlying disease. The few patients with limited disease who developed major bleeding episodes survived a median of 26 months (Table 3). The Cox proportional hazards regression analysis illustrates this phenomenon. The presence of disseminated disease (odds ratio [OR], 3.5; P < .0001), poor performance status (OR, 2.1; P < .0001), and age  65 years (OR, 1.8; P = .0008) were associated with far shorter durations of survival. Minor bleeding episodes and chemotherapy dose delays and reductions did not contribute to shorter survival; only a major bleeding episode was significantly associated (OR, 1.9; P = .02) with shorter survival ( Fig 1). Although the presence of febrile neutropenia did not contribute to shorter median survival times, mortality at 30 days was significantly more common during cycles complicated by febrile neutropenia (6% v 2%; P < .0001).
Episodes of bleeding were significantly associated with a decrement or no increment in platelet count after transfusion compared with cycles with dose modifications or no events (47% v 17%; P < .0001). However, the overall rate of poor response to platelet transfusion was surprisingly high during all cycles (19%) and for both random-donor platelets (23%) and single-donor platelets (15%). Poor response to platelet transfusion was observed most commonly during chemotherapy cycles in males (56%); patients with lymphoma (45%), genitourinary malignancies (15%), or sarcoma (14%); those who had undergone more than 10 cycles of chemotherapy in the past; and those with widely disseminated disease (69%). Bleeding was associated with a failure to achieve adequate response to platelet transfusion in 37% of cycles compared with only 12% of cycles during which an adequate response was achieved (P < .0001). Although this may be due partially to consumption of platelets at bleeding sites, episodes of bleeding during previous cycles were also associated with poor increments (14% v 6%; P = .004). Nineteen percent of cycles during which a poor response to platelet transfusion was observed ended in the death of the patient (including the four patients whose deaths were attributed to hemorrhage) compared with only 3% of patients who responded to platelet transfusion (P < .0001). Forty-one percent of these patients also received single-donor platelets during that cycle and, among these, 23% experienced a decrement in platelet count after single-donor platelets. It is notable that during this time period, platelet transfusions were filtered by the blood bank for patients with hematologic malignancies and for bone marrow transplant recipients only. Therefore, only the lymphoma patients in this study received platelet transfusions filtered in the blood bank. All platelet transfusions were administered through filters.
What Resources Are Used During Cycles Complicated by Bleeding and Dose Modifications?
Failure to achieve an adequate response after platelet transfusion may be a financially significant event. Overall, patients who experienced a decrement in platelets received an average of 31 units of platelets for that cycle. Patients who experienced no increment received 18 units, those with an increment of 1,000 platelets per unit, 17 units, and those who achieved an increment more than 1,000 platelets per unit received only 8 units for that cycle (P < .0001). Patients who failed to achieve an adequate response to platelet transfusion were hospitalized an average of 8 days during thrombocytopenia compared with only 4 days for patients who responded to platelet transfusion (P < .0001). Considering the severity of outcomes and excessive resource utilization, episodes of bleeding were considered of sufficient import for further study. Although not associated with poor outcomes, chemotherapy delays were associated with sufficient increases in resource utilization to justify further study. However, the outcomes and resource utilization associated with dose reductions were virtually indistinguishable from those observed during cycles with no event. Therefore, we studied predisposing factors for bleeding episodes and chemotherapy delays.
What Factors Predispose to Episodes of Bleeding and Chemotherapy Delays?
The risk of bleeding was also unrelated to the specific underlying malignancy ( Table 6). However, it was associated with disseminated disease (P = .0001), bone marrow metastases (P < .0001), prior episodes of bleeding (P < .0001), or a baseline platelet count less than 75,000/µL (P < .0001). Delays of the subsequent cycles were uncommon during adjuvant chemotherapy (P = .0001) but were significantly more common when the baseline platelet count was less than 50,000/µL (P < .0001). The presence of necrotic tumor did not predispose to bleeding. Delays were more common when necrotic tumor was present, although this difference did not reach statistical significance (P = .17). During cycles in which febrile neutropenia accompanied thrombocytopenia, major bleeding episodes were significantly more common (P = .0008).
Numerous previous cycles of chemotherapy did not predispose to bleeding whereas chemotherapy delays were associated with more than five previous cycles (P = .0008) ( Table 7). Previous radiation therapy did not predispose to bleeding but was associated with a high risk of delay in a subsequent cycle of chemotherapy (41% v 22%, P = .001). Concurrent radiation therapy was not associated with an increased risk of bleeding or chemotherapy delay. Regimens containing mitomycin, carmustine, or lomustine predisposed to bleeding, although these agents were used only occasionally. Cisplatin- or carboplatin-based regimens were also associated with higher rates of bleeding. Minor bleeding episodes and chemotherapy delays were more common during cycles in which prophylactic platelet transfusions were administered (P = .01 and .07, respectively). Bleeding and delays were significantly associated with both the depth and duration of thrombocytopenia. Bleeding was particularly common during cycles in which the platelet count fell below 10,000/µL (7% v 21%; P < .0001). In fact, the rate of bleeding doubled (from 5% to 10%) when the platelet count fell below 20,000/µL and doubled again (from 10% to 21%) when the count fell below 10,000/µL ( Fig 2).
Because of the correlation among many of the predictive factors, logistic regression was used to identify the unique contribution of each of these factors to the risk of clinically significant thrombocytopenia. Based on the differing risk profiles suggested by the univariate analysis, separate regression models were developed for bleeding and chemotherapy delay. The single most significant predictor of bleeding was a prior history of bleeding (P < .0001) ( Table 8). Measures of poor bone marrow function, such as a baseline platelet count less than 75,000 and the presence of bone marrow metastases, were also significant predictors of bleeding (P = .0001 and .005, respectively), as was a poor performance status (P = .05). Other factors shown to be predictive in univariate analysis (such as disseminated neoplastic disease and specific chemotherapeutic agents) did not prove to be significant predictors in the multiple-variable model. A delay in the subsequent cycle of chemotherapy was predicted by the presence of disseminated neoplastic disease (P = .04) and measures of previous insults to the bone marrow such as prior radiation therapy (P = .03) and numerous previous cycles of chemotherapy (P = .003) (Table 8).
Major hemorrhage during chemotherapy-induced thrombocytopenia is a serious clinical problem. This is particularly true of CNS or major organ hemorrhage. Numerous units of platelets are administered to patients with cancer each year with the goal of preventing such events. However, the actual risk of these events, particularly among patients with solid tumors, has been disputed. Similar disputes surround the importance of chemotherapy delays and dose reductions. We examined the incidence and outcomes of these events to inform decisions to use prophylactic platelet transfusions and growth factors in these patients. The primary impetus for this study was the lack of data on the incidence of bleeding among thrombocytopenic patients with solid tumors who have received modern chemotherapy regimens. However, despite major changes in chemotherapy regimens, our findings were remarkably similar to those described in the past. We observed bleeding in 9% of patients with thrombocytopenia, Belt et al2 reported bleeding in 10% of similar patients in 1978, and Dutcher et al3 reported bleeding among 15% in 1984. We observed only a single episode of CNS hemorrhage, despite a significant proportion of patients (> 10%) at high risk. The consistency of these findings supports the notion that solid tumor patients with similar depth and duration of thrombocytopenia share a common risk of bleeding, regardless of the specific chemotherapy regimen that is administered. Our data also suggest that in addition to a common risk of bleeding during thrombocytopenia, solid tumor patients share common predisposing factors that are host-specific rather than neoplasm-specific. Most notable among these is a history of bleeding. Also among these are factors suggestive of poor bone marrow function, such as a low baseline platelet count and bone marrow metastasis, as well as poor performance status. Although tested in univariate and multiple variable analyses, the specific neoplasm was not useful in predicting an episode of bleeding. Our results combined with previous data confirm that among thrombocytopenic solid tumor patients, the risk of bleeding is low, overall. However, when episodes of bleeding do occur, they are associated with poor clinical outcomes and significantly increased resource utilization. We have demonstrated that there are easily identifiable subsets of this population in whom the risk of bleeding exceeds 20%. Given these findings, a single strategy (or threshold) for bleeding prophylaxis of all thrombocytopenic solid tumor patients is unlikely to provide the most cost-effective solution; an individualized approach is far more attractive. We are currently developing a clinical prediction rule based on this multiple-variable risk model to guide the use of prophylactic platelet transfusions and growth factors for individual patients in this population. Delays in chemotherapy were uncommon (6%). Compared with cycles with no event, delays were associated with higher resource utilization but not poorer clinical outcomes. Although the specific factors that predisposed to chemotherapy delays were different, they were similar to those predisposing to bleeding in that they were illustrative of poor bone marrow reserve (prior radiation therapy and numerous previous cycles of chemotherapy) and poor performance status (disseminated disease). Thus these patients are also easily identifiable by using information available on day 1 of a cycle of chemotherapy. It may be desirable from a cost standpoint to identify these patients, and, therefore, this may be a fruitful area of research. In contrast to dose delays, dose reductions led to neither poorer outcomes nor increased resource utilization. Although the benefit of high doses has been demonstrated for some malignancies (particularly leukemia, multiple myeloma, some lymphomas, and sarcomas), for many other solid tumors, little or no advantage has been demonstrated to high-dose regimens with substantial toxicity.20 In our Cox proportional hazards model of survival, dose reduction did not prove to be significantly associated with shorter durations of survival, despite an overrepresentation of patients with lymphoma and sarcoma in whom this difference might have been expected to be important. If this benefit does indeed exist, it may be limited to a subset of patients with solid tumors. Finally, we have reported strikingly high rates of inadequate response to both random-donor (23%) and single-donor (15%) platelet transfusion for this population. Although this was not a primary objective of the study, we report these results to stimulate hypotheses and future study, because this may be among the most clinically and financially significant effects of thrombocytopenia. Numerous platelet transfusions (often of the single-donor variety) were required for maintenance of hemostasis by patients who failed to achieve an adequate response to transfusion. Considering the average cost of platelet transfusions in the United States (> $500), this can be an extremely costly undertaking.21 From both a clinical and a statistical standpoint, it is difficult to determine whether poor increments led to bleeding or whether bleeding and its attendant, numerous transfusions led to poor increments. Although tests of statistical significance suggest that the association between these factors is unlikely to have occurred by chance, both may be due to a third factor (such as consumption of platelets at bleeding sites). Whatever the causal relationship, the excess mortality among patients who failed to achieve adequate increments after transfusion (19% v 3%) is troubling. This finding is doubly significant when viewed in the light of the observation that all four patients whose deaths were caused in part by thrombocytopenia-induced hemorrhage experienced inadequate increments after both random-donor and single-donor platelet transfusions. These patients’ deaths were multifactorial, and it is possible that their poor responses to transfusion resulted from consumption of platelets at their extensive bleeding sites. Nevertheless, this association—when considered together with the previously mentioned observations—reinforces a concern that has been voiced by many clinicians. Excessive use of platelet transfusions for prophylaxis and treatment, early in the course of malignancy, may result in patients who are refractory to all platelet transfusions later in the course of their malignancy, when they are at highest risk of bleeding. Our data underscore the importance of this concern. Prospective study of this significant downstream effect of thrombocytopenia should be a high priority. We conclude that some, but not all, solid tumor patients are at sufficiently high risk of serious clinical outcomes to justify aggressive bleeding prophylaxis during chemotherapy-induced thrombocytopenia. Our ongoing research focuses on the development of an easy-to-use clinical prediction rule to guide individualized prescription of prophylactic platelet transfusions and growth factors. We encourage further research on the downstream clinical and financial effects of failure to achieve adequate platelet increments after platelet transfusion.
Supported in part by a grant from Genetics Institute, Inc, Cambridge, MA.
Presented in part at the Thirty-Third Annual Meeting of the American Society of Clinical Oncology, May 17-20, 1997, Denver, CO.
1. Gaydos LA, Freireich EJ, Mantel N: The quantitative relation between platelet count and hemorrhage in patients with acute leukemia. N Engl J Med 266: 905-909, 1962
2.
Belt RJ, Leite C, Haas CD, et al: Incidence of hemorrhagic complications in patients with cancer. JAMA 239: 2571-2574, 1978 3. Dutcher JP, Schiffer CA, Aisner M, et al: Incidence of thrombocytopenia and serious hemorrhage among patients with solid tumors. Cancer 53: 557-562, 1984[Medline]
4.
Goldberg GL, Gibbon DG, Smith HO, et al: Clinical impact of chemotherapy-induced thrombocytopenia in patients with gynecologic cancer. J Clin Oncol 12: 2317-2320, 1994 5. Fanning J, Hilgers RD, Murray KP, et al: Conservative management of chemotherapy-induced thrombocytopenia in women with gynecologic cancers. Gynecol Oncol 59: 191-193, 1995[Medline]
6.
Beutler E: An iconoclastic view of conventional wisdom in hematology. Arch Intern Med 139: 221-223, 1979 7. Baer MR, Bloomfield CD: Controversies in transfusion medicine: Prophylactic platelet transfusion therapy—Pro. Transfusion 32: 377-380, 1992[Medline] 8. Schiffer CA: Prophylactic platelet transfusion. Transfusion 32: 295-298, 1992 (editorial)[Medline]
9.
Beutler E: Platelet transfusions: The 20,000/µL trigger. Blood 81: 1411-1413, 1993 10. Pisciotto PT, Benson K, Hume AB, et al: Prophylactic versus therapeutic platelet transfusion practices in hematology and/or oncology patients. Transfusion 35: 498-502, 1995[Medline] 11. Norfolk DR, Ancliffe PJ, Contreras M, et al: Concensus conference on platelet transfusion: Royal College of Physicians of Edinburgh. Br J Haematol 101: 609-617, 1998[Medline]
12.
Heckman K, Weiner G, Davis C, et al: Randomized study of prophylactic platelet transfusion threshold during induction therapy for adult acute leukemia: 10,000/ml versus 20,000/ml. J Clin Oncol 15: 1143-1149, 1997
13.
Rebulla P, Finazzi G, Marangoni F, et al: The threshold for prophylactic platelet transfusions in adults with acute myeloid leukemia. N Engl J Med 337: 1870-1875, 1997
14.
Wandt H, Frank M, Ehninger G, et al: Safety and cost effectiveness of a 10 x 109/L trigger for prophylactic platelet transfusions compared with 20 x 109/L trigger: A prospective comparative trial in 105 patients with acute myeloid leukemia. Blood 91: 3601-3606, 1998 15. Tepler I, Elias L, Smith II JW, et al: A randomized placebo-controlled trial of recombinant human interleukin-11 in cancer patients with severe thrombocytopenia due to chemotherapy. Blood 87:3607-3614, 1996
16.
Isaacs C, Robert NJ, Bailey FA, et al: Randomized placebo-controlled study of recombinant human interleukin-11 to prevent chemotherapy-induced thrombocytopenia in patients with breast cancer receiving dose-intensive cyclophosphamide and doxorubicin. J Clin Oncol 15: 3368-3377, 1997 17. Basser R: Clinical use of thrombopoietic growth factors. ASCO Educational Book, Spring:266-274, 1999 18. Zubrod CG, Schneiderman M, Frei E, et al: Appraisal of methods for the study of chemotherapy of cancer in man: Comparative trial of nitrogen mustard and trimethylene thiophosphoramide. J Chron Dis 11: 7-33, 1960 19. Charlson ME, Pompei P, Ales KL, et al: A new method of classifying prognostic comorbidity in longitudinal studies: Development and validation. J Chron Dis 40: 373-383, 1987 20. Savarese DM, Hsieh C, Stewart FM: Clinical impact of chemotherapy dose escalation in patients with hematologic malignancies and solid tumors. J Clin Oncol 15: 2981-2995, 1997[Abstract] 21. Cantor SB, Elting LS, Hudson DV, et al: Pharmacoeconomic analysis of rhIL-11 for secondary prophylaxis of thrombocytopenia in solid tumor patients receiving chemotherapy. Proc Am Soc Clin Oncol 18: 419a, 1999 (abstr 1619)
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2001 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 tests were used to test the significance of discrete variables and two-tailed t tests, continuous variables. Factors shown to be significant predictors in univariate analysis (P < .10) were included in logistic regression analyses. For this analysis, only those factors that are routinely available on day 1 of the chemotherapy cycle in all solid tumor patients were used, because the decision to administer prophylactic platelet growth factors is made at this time. Separate regression analyses were conducted for bleeding, chemotherapy delay, and dose reduction. Continuous variables were categorized or dichotomized for the regression analyses. The goodness of fit of the models was estimated using the Hosmer-Lemeshow statistic. Statistical analyses were computed using BMDP Dynamic (Version 7, BMDP Statistical Software, Inc., Los Angeles, CA). 
