Annals of Oncology Advance Access originally published online on December 1, 2005
Annals of Oncology 2006 17(3):507-514; doi:10.1093/annonc/mdj092
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© 2005 European Society for Medical Oncology
Can we predict the duration of chemotherapy-induced neutropenia in febrile neutropenic patients, focusing on regimen-specific risk factors? A retrospective analysis
1 Department of Médecine Interne and Laboratoire d'Investigations Cliniques H-J Tagnon, Centre des Tumeurs de l'Université Libre de Bruxelles; 2 Jules Bordet Institut, Data Centre, Rue Héger Bordet 1, 1000 Brussels, Belgium
* Correspondance to: Dr Y. Lalami, Institut Jules Bordet, Rue Héger Bordet 1, 1000 Bruxelles, Belgium. Tel: +322/541-32-35; Fax: +322/541-33-12; E-mail: yassine.lalami{at}bordet.be
 |
Abstract |
|---|
 Top Abstract introductionpatients and methodsresultsdiscussionReferences |
|---|
Background: The aim of the study was to elaborate a predictive model for the duration of chemotherapy-induced neutropenia (CIN) allowing the identification of patients with a higher risk of complications, especially complicated febrile neutropenia, who might benefit from preventive measures.
Patients and methods: A score ranging from 0 to 4 on the basis of expected CIN was attributed to each cytotoxic agent given as part of chemotherapy treatment in solid tumours for patients with febrile neutropenia (FN). The individual scores were combined into several overall scores.
Results: A total of 203 patients with FN were eligible for this retrospective analysis. We were able to identify two groups of patients with statistically different neutropenia durations with median durations until hematological recovery of ANC 
0.5 and 1.0 x 109/l, being respectively 6 versus 4 days (P = 0.03) and 8 versus 6 days (P = 0.01).
Conclusions: The duration of neutropenia is directly influenced by the aggressiveness of the chemotherapy regimen. In this retrospective study, we were able to identify a group of patients who needed two more additional days to recover from grade 3 and grade 4 neutropenia, based on the degree of aggressiveness of the cytotoxic agents used.
Key words: chemotherapy, neutropenia, febrile neutropenia, duration of neutropenia
|
introduction |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionReferences |
|---|
Chemotherapy-induced neutropenia (CIN) is the most common toxicity linked to the administration of anticancer drugs and continues to be associated with substantial morbidity and mortality, which is directly associated with its duration and severity [1
].
The most commonly used scale for grading the severity of CIN is derived from the Common Toxicity Criteria of the National Cancer Institute. It delineates neutropenia into four grades following the absolute neutrophil count (ANC): grade 1, ANC 1.5 to <2 x 109/l; grade 2, 1.0 to <1.5 x 109/l; grade 3, 0.5 to <1.0 x 109/l; grade 4, <0.5 x 109/l [2].
The clinical impact of CIN has been studied on the basis of complications during neutropenia, as well as on the frequency of reduced dose intensity. During a neutropenic episode, all the patients treated with chemotherapy have a risk of developing febrile neutropenia, potentially followed by serious medical complications, but it is difficult to predict which patients have the greatest risk. It is now widely accepted that febrile neutropenia (FN) represents a heterogeneous condition and that all the patients suffering from it will not have an identical outcome in terms of morbidity and mortality. As a result, various decision-making models have been established for the management of patients at the onset of FN. Moreover, colony-stimulating factors (CSF) have been evaluated in decision-making models to reduce the risk of FN; the use of CSF has been codified on the basis of large clinical trials [3]. Neutropenia duration would be very important in assessing the risk at the onset of a febrile neutropenic episode; however, it is impossible to integrate this duration in a risk–prediction model because it is, a priori, unknown and presently unpredictable when febrile neutropenia occurs.
For this reason, the MASCC (Multinational Association of Supportive Care in Cancer) Study Section for Infectious Diseases is currently conducting a muticentric survey of febrile neutropenia in order to develop a multivariate prediction model for further neutropenia duration using data on chemotherapy regimen as well as data on underlying cancer and patient's characteristics. The final aim is to improve our ability to predict the development of a serious medical complication following a febrile neutropenic episode.
When this survey was designed, it was already clear that the type of chemotherapy regimen should be one of the major determinants for the risk of neutropenia and for its severity and duration. Indeed, in a multivariate analysis, the chemotherapy schedule was considered as being a significant predictor for the reduction of dose intensity, in association with age 65 and body surface area >2 m2 [4]. However, we are not aware of any validated tool for categorizing chemotherapy regimens according to their toxicity on the neutrophil count. Therefore, we tested a way of classifying the effect on neutrophils of chemotherapy regimens, on a subgroup of episodes registered in one of the participating institutions (Institut Jules Bordet), with the aim of using it for the analysis of the overall MASCC observational study [5, 6].
The objective of the present analysis was to study the univariate association between a score allocated to each chemotherapy regimen and the duration of CIN. To our knowledge, no study has attempted to derive such an association. Only a few trials have been conducted in the past for the elaboration of clinical models using characteristics of patients and chemotherapy for the prediction of FN after chemotherapy administration [7–12].
|
patients and methods |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionReferences |
|---|
Our analysis was done using a subgroup of patients included in the MASCC survey for whom the eligibility criteria were as follows: (1) diagnosis of malignancy; (2) over the age of 16 years; (3) fever defined as an oral temperature above 38°C with a presumed infection in the absence of other non-infectious causes of fever such as blood transfusion; (4) neutropenic (granulocyte count <500/mm3 including polymorphonuclear leucocytes and bands) with non-chronic neutropenia induced by recent chemotherapy administration; and finally, (5) treatment with an appropriate initial empirical antibiotic regimen (acceptable regimens include: an antipseudomonal ß-lactam in combination with an aminoglycoside; two ß-lactams in combination or monotherapy with a third generation cephalosporin, piperacillin/tazobactam or a carbapenem).
The primary objective of the survey was to integrate an estimation of further neutropenia duration in a model for predicting the occurrence of serious medical complication. We therefore selected patients included in the multicentric survey by our institution, in order to retrospectively collect data on hematological counts since the beginning of chemotherapy administration and not only since the onset of febrile neutropenia. We excluded patients with hematological malignancy, in order to focus on tumors that are not themselves inducing neutropenia, and those without a full description of the chemotherapy regimen (including doses) and/or availability of hematological counts for assessing the total neutropenic duration. Patients included in the present analysis were registered in the survey between April 1998 and March 2002.
We allocated a score for chemotherapy aggressiveness, ranging from 0 to 4, which was supposed to reflect the expected severity of neutropenia induced by each cytotoxic agent (score 0, no neutropenia; score 1, mild neutropenia; score 2, moderate neutropenia; score 3, intermediate neutropenia; score 4, severe neutropenia). This score for individual agents is shown in Table 1 and was constructed using data on single drug therapy. We also tried to translate the degree of neutropenia into duration of CIN. High doses of cytotoxic agents were defined as proposed by Blay et al. [7]: doxorubicin >90 mg/m2, cisplatin >100 mg/m2, cyclophosphamide >1000 mg/m2, ifosfamide >9000 mg/m2, etoposide >500 mg/m2 and cytarabin >1000 mg/m2. When the dose had an impact on the score, we used the actual dose and not the planned dose.
|
For combining the scores attributed to each individual cytotoxic drug and to have a global classification score of a regimen, we tested three models: score S1, maximum of the individual scores; score S2, addition of the individual scores; score S3, addition of individual scores completed by interaction terms (weighted products of two individual scores) (Tables 2–4
).
|
|
|
The end point of the analysis was the total duration of neutropenia following chemotherapy administration. The beginning of neutropenia duration was defined as the first day with the documentation of an absolute neutrophil count (ANC) <0. 5x 109/l (all the patients included had grade 4 neutropenia as they developed a febrile neutropenic episode). Recovery from neutropenia was defined using three different thresholds: ANC
0.5 x 109/l, ANC 1.0 x 109/l and ANC 2.0 x 109/l. All available blood counts were taken into account for calculating the observed neutropenia duration. The main objective of our analysis was to test a potential association between the chemotherapy global classification score and the duration of CIN.
statistical considerations
We used non-parametric Kaplan–Meier analysis in order to estimate the time to neutropenia recovery, as well as the Cox regression model. In order to have a simple and easy to use categorization of the chemotherapy regimens, we chose to test first a three-level classification (expected short, intermediate and long duration of neutropenia). These three scores were categorized without looking at CIN duration but only by looking at the distributions of the scores in order to have groups of meaningful size. Comparing the Kaplan–Meier estimates using log-rank tests assessed the impact of the categorized chemotherapy aggressiveness scores. The impact of the original scores was assessed using Cox regression models. All reported significance probabilities are two-tailed. Multiple tests were done and we were aware of an increase of the probability of a false result. This methodology is justified because our analysis was aimed at generating a classification of a chemotherapy regimen that will be tested further on the overall study database. For this reason, as well as for sample size reasons, we did not attempt to conduct multivariate analysis although we looked at the association between the chemotherapy score and possible confounding factors such as underlying disease, treatment setting and administration of growth factors.
|
results |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionReferences |
|---|
Two hundred and five patients, registered in the MASCC survey were eligible for the present analysis but two were excluded because there was a lack of data about the chemotherapy regimens. Moreover, 26 of these patients presented a subsequent episode of FN that was not considered for our analysis. Among the 203 analyzed patients there were 57 men (28%) and 146 women (72%). The median age was 55 years, with a range of 18–83 years. The majority of underlying tumors were breast (49.3%) and lung cancer (20.2%). Table 5 summarizes the principal characteristics of the population. The chemotherapy treatment was adjuvant in 73 patients (36%), first-line for advanced disease in 62 patients (31%) and second-line in 57 patients (28%). Eleven patients (5%) received their chemotherapy in association with autologous bone marrow transplantation. Fifty-four patients (27%) were treated with a monotherapy, while 79 (39%) received a combination of two drugs, 63 (31%) received a tri-therapy and seven patients (3%) received a combination of four anticancer drugs. Fourteen patients (7%) had a previous history of FN before inclusion. A delay in chemotherapy administration was necessary in 12 patients (6%) because of hematological toxicity and reductions of doses associated with a delay of treatment.
|
Hematological recovery to ANC
0.5 x 109/l was observed in 195 patients (96%). Among the eight patients (4%) for whom recovery from neutropenia was not documented, six patients died without recovering from grade 4 neutropenia and two patients were lost to follow-up. Hematological recovery to ANC 1.0 x 109/l and 2.0 x 109/l was documented in 163 (80%) and 75 patients (37%), respectively. This absence of documentation is mainly due to an early discharge from hospital, without further blood sampling available for differential counts assessment.
The median time to recovery was 5 days for ANC 0.5 x 109/l; 25% of the patients recovered after 3 days and 25% had not recovered after 7 days. When we considered the level of ANC 1.0 x 109/l, the median duration of neutropenia was 6 days; 25% of patients recovered after 4 days and 25% remained neutropenic (
1000 x 109/l) after 7 days. Finally, the median time to recovery for ANC 2.0 x 109/l was 9 days (25% of the patients recovered after 6 days and 25% after 15 days or more).
The original three scores were redefined using three categories, described as follows: S1, 2 versus 3 versus 4; S2, <5 versus 5–7 versus >7; S3, 4 versus 5–10 versus >10.
impact of S1 on neutropenia duration
No impact of S1 was observed, whichever time to recovery was considered. For hematological recuperation with an ANC 0.5 x 109/l, the median durations of neutropenia were not different in the three groups considered. In the category with the lowest score and the shortest expected duration (score 2), the median duration for recovery was 5 days, whereas the respective medians were 4 days and 5 days for the groups having a higher score (3 or 4) (P = 0.64). Looking at recovery above 1.0 x 109/l, the same figures were 7, 7 and 6 days (P = 0.28) and 7, 8 and 10 days for the last definition of recovery (P = 0.34). Treating S1 as a continuous variable did not allow the identification of any predictive value to the score.
impact of S2 on neutropenia duration
Borderline significance levels were reached for each of the three outcomes: median durations were 5, 4 and 6 days for reaching an ANC 0.5 x 109/l (P = 0.06); 6, 6 and 8 days for obtaining a count 1.0 x 109/l (P = 0.04); 8, 9 and 10 days were required for a count above 2.0 x 109/l (P = 0.04).
A categorization with three classes did not appear to be discriminant enough for the first two levels and led us to consider a binary evaluation using a threshold at 7. Estimates of neutropenia duration were 4 versus 6 days (P = 0.03) (Figure 1), 6 versus 8 days (P = 0.01) (Figure 2), respectively, for a score <8 (n = 169) and a score 8 (n = 34) and for thresholds for recovery set at 0.5 x 109/l and 1.0 x 109/l.
|
|
The analysis for ANC
2.0 x 109/l level led to median durations of 9 versus 10 days (P = 0.02) but cannot be considered as reliable due to the small number of complete observations in the group with the high score (26 out of 34 observations were censored). Treating S2 as a continuous variable did not lead to improved significance of the score.
impact of S3 on neutropenia duration
As for S2, the association between S3 and neutropenia duration was of borderline significance for each of the three outcomes: for the threshold at 0.5 x 109/l, medians were 5 versus 4 versus 7 days (P = 0.08), 5 versus 6 versus 8 days (P = 0.05) and 8 versus 9 versus 10 days (P = 0.07). Results were similar to those obtained with S2. We therefore preferred to choose S2 as the best way to classify the chemotherapy regimens because of easier calculation (Table 6). Treating S3 as a continuous variable did not lead to improved significance of the score.
|
Finally, an analysis of various factors that could be considered as confounding with the neutropenic score was also realised, although the purpose of the present analysis was not to develop a predictive model of neutropenia duration. As far as the underlying tumour is concerned, a score >7 was more frequent among patients with lung cancer (32%), compared with breast (9%) and the other types of tumor (19%) (P = 0.04).
The presence of bone metastases was not associated with a significant difference; there was a high score in 15% of the patients in both groups (with and without bone metastases; P = 1.00). Concomitant radiotherapy (RT) had no significant impact with 18% of high scores in the absence of RT versus 8% for the patients receiving concomitant RT (P = 0.38).
Regarding the treatment setting, a statistical significant association was established. High scores were identified in 11% of the adjuvant treatments, in 24% of advanced diseases (first line), in 9% of second-line treatment and in 55% in the case of high-dose treatment with autologous BMT (P < 0.001). For the administration of colony stimulating factors (CSF), we found a high score in 14% of patients in the absence of CSF and 27% in the presence of CSF administration (P = 0.09). Finally, documented infection did not appear to be a confounding factor in our analysis (P = 0.67).
|
discussion |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionReferences |
|---|
Myelosuppression is a major dose-limiting toxicity associated with cancer chemotherapy. The type of chemotherapy agents, the dosage and number of drugs are clearly related to the risk of neutropenia and infections [1
]. Neutropenia is the most common cause of chemotherapy dose reduction and delay. Reduction of standard dose intensity has been shown to increase the risk of disease recurrence and death among patients treated for potentially curable malignancies, such as early-stage breast cancer and non-Hodgkin's lymphoma [13, 14]. Despite its clinical importance as a major side-effect of chemotherapy, neutropenia is often under-reported in various clinical trials. Many investigators still report myelotoxicity in terms of leucopenia rather than neutropenia, and blood counts are often not taken enough, especially in outpatients [4]. This may be the cause of difficulty in assessing, with accuracy, the neutropenia-related infection risk in many patients, and the reason for the possible underestimation of grade 4 neutropenia.
Crawford et al. [15] found that 96% of the 102 patients with small-cell lung cancer, treated with higher doses of CAE (cyclophosphamide, doxorubicin and etoposide) experienced grade 4 neutropenia. Blay et al. [7] reported an overall incidence of grade 4 neutropenia in 51% of patients treated for lymphoma and solid tissue malignancies.
In current practice, the incidence of febrile neutropenia (FN) with standard chemotherapy regimens rarely exceeds 25%. It is associated with a significant mortality rate, ranging from 9% in solid tumors and lymphoma to 14% in patients treated for hematological malignancies. Moreover, the morbidity and economic costs associated with FN are substantial [2].
All patients who are treated with chemotherapy are at risk of developing FN and complicated febrile neutropenia. It appears very important for clinicians to identify, based on predictive models, patients at the greatest risk of complications and those who might benefit from preventive measures such as CSF administration [5, 16]. Risk-stratification prediction models, identifying specific patient populations at increased risk of neutropenia, have been developed. A number of independent risk factors, including first-cycle ANC levels or more advanced age, have been identified and may facilitate effective targeting of prophylactic G-CSF to the patients at greatest risk of severe neutropenia [7–11]. Once febrile neutropenia has been developed, one of the validated models for predicting complications in an adult patient population is that developed by MASCC (Multinational Association of Supportive Care in Cancer) [17, 18] but it does not take into account the duration or the severity of CIN, because of its inaccessibility at the time of the therapeutic decision. The objective of the current, ongoing second survey of febrile neutropenia by MASCC is to incorporate an estimate of further neutropenia duration into the risk–prediction model.
Our main objective in this retrospective analysis was to look at the association between the duration of CIN and chemotherapy by taking into consideration only specific regimen–risk factors. As described earlier, the severity of CIN is directly related to the type and doses of the cytotoxic agent administered. Other significant co-factors include age, performance status, stage of the disease and number of previous chemotherapy-lines of treatment. In trying to confirm our hypothesis, we conducted the retrospective analysis presented here, to identify a potential correlation between the degree of chemotherapy-aggressivity and the duration of CIN.
We started with a simple score for each individual drug, ranging from 0 to 4; each score being attributed to a considered cytotoxic drug and its potential degree of neutropenia (score 0, no CIN; 1, mild; 2, intermediate; 3, moderate and score 4; high degree of CIN). This initial classification was based on general data of CIN emerging from the literature, even if there is a lack of consistency and homogeneity in the data reported in various clinical trials. This methodology leads, in our opinion, to a widely applicable method of classification, and is not specific to particular chemotherapy regimens. In the case of the use of new drugs into a combination, it would be possible to apply the method just by providing a score for the individual new drugs.
True duration of neutropenia is a continuous outcome; however, for the ease of use of a patient's classification, we decided to develop a discrete categorization. Our first aim was to develop three categories, in order to make possible the classification of a chemotherapy regimen when only considering its neutropenic impact.
Our initial choice to consider three groups from the initial scores was arbitrary but it theoretically provided more accurate information than a binary classification. The attribution of the various level scores for each variable (S1, S2 and S3) was done in a way to obtain three equivalent groups without previous analysis of neutropenia duration, so our results could remain easily reproducible. However, our initial results were disappointing; the two first categories being identical in terms of CIN. Therefore, we based our further analysis on a binary categorization. We chose the score S2, because it was easier to determine and also, because the score S3 did not add any statistical significant advantage over S2. This may be due to the sample size not being large enough or to a lack of accuracy in our measurement of neutropenia duration. The decision to use a two-category classification rather than three groups deserves further investigation. S1 was even simpler but it has no predictive value on CIN, as the combination appeared logically to have a cumulative impact on neutropenia duration and severity.
Due to its design with registration of patients who had all developed grade 4 neutropenia, our study did not allow for the estimation of neutropenia incidence but was completely dedicated to total neutropenia duration in association with the chemotherapy regimen that was administered. This means that our study results cannot necessarily be generalized to a population of patients undergoing chemotherapy but not developing fever. We should stress that our end point may lack accurate assessment, especially before fever development. Indeed, the documentation of onset of neutropenia often coincided with the diagnosis of fever in 150 patients (74%). Lack of documentation at the onset of neutropenia was more frequent in non-hospitalized patients compared with the inpatient situation (87% versus 44%, respectively; P < 0.001). Documentation of recovery at a threshold of 0.5 x 109/l was not a problem as the patients were followed-up closely because of the febrile neutropenic episode, but was more difficult when higher thresholds were considered.
Our observed neutropenia duration is, therefore, probably underestimated in many patients but we believe that our categorization remains pertinent, as we do not expect the length of the non-documented neutropenia to be dependent on the chemotherapy regimen. Our work is, therefore, only a first step to looking at the association between chemotherapy and neutropenia duration, and further studies are certainly required to obtain a tool that might be of true practical importance. Ideally, we should have prospectively registered chemotherapy cycles and not febrile neutropenic episodes and should have required blood samples to be taken every 2 days even in absence of fever development. In practice, such an ideal situation cannot be targeted for obvious ethical reasons. Better documentation could be reached for hematological patients who are more frequently hospitalized but we did not choose this patient population for our study since we preferred to have data on neutropenia induced by chemotherapy without interaction with the underlying disease.
A significant impact of the chemotherapy regimen on the CIN appears to be adequately described by the dichotomized S2 score; it allows a simple categorization of the score into two groups: low score (<8) and high score (8). Such a discriminant score allowed the selection of a patient group with a high score for whom the median duration of CIN lasted two more additional days prior to hematological recovery to ANC 0.5 x 109/l (4 days versus 6 days, P = 0.03) and 1.0 x 109/l (6 days versus 8 days, P = 0.01) (Figures 1, 2). Distribution and details of the chemotherapy regimens following the chemotherapy score is shown in Tables 7 and 8. Moreover, we could identify factors related to a high score, including underlying tumour (P = 0.04), treatment setting (P < 0.001) and administration of CSF (P = 0.09).
|
|
We studied a population of patients treated exclusively for solid tumors, so that we could have a homogeneous group. Moreover, these patients represented a population for whom there are still many controversies about the potential indications of preventive measures for the management of FN, after a prolonged period of CIN, compared with hematological malignancies. As far as complications from a febrile neutropenic episode are concerned, further investigations and, in particular, the MASCC ongoing survey, will inform us about the pertinence of this classification in a more general and multifactorial context and also when applied to patients with hematological tumors for whom larger differences in time to neutropenia recovery can be expected.
In conclusion, the duration of neutropenia is directly influenced by the aggressiveness of the chemotherapy regimen. In this retrospective study, we were able to identify a group of patients who needed two more additional days to recover from grade 3 and grade 4 neutropenia, based on the degree of aggressiveness of the cytotoxic agents used.
Received for publication December 8, 2004. Revision received June 7, 2005. Accepted for publication October 31, 2005.
|
References |
|---|
|
TopAbstractintroductionpatients and methodsresultsdiscussionReferences |
|---|
1. Bodey GP, Buckley M, Sathe Y et al. Quantitative relationships between circulating leucocytes and infection in patients with acute leukemia. Ann Intern Med 1966; 64: 328–339.[ISI][Medline]
2. Crawford J, Dale DC, Lyman GH. Chemotherapy-induced neutropenia. Risks, consequences and new directions for its management. Cancer 2004; 100: 228–237.[CrossRef][Medline]
3. 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. J Clin Oncol 2000; 18: 3558–3585.
4. Dale DC, Crawford J, Lyman GH. Chemotherapy-induced neutropenia and associated complications in randomised clinical trials: an evidence-based review. Proc Am Soc Clin Oncol 2001; 20: 410a (Abstr 1638).
5. Lyman GH, Kuderer NM. The economics of the colony-stimulating factors in the prevention and treatment of febrile neutropenia. Crit Rev Oncol Hematol 2004; 50: 129–146.[Medline]
6. Rolston K. Prediction of neutropenia. Int J Antimicrob Agents 2000; 16: 113–115.[Medline]
7. Blay JY, Chauvin F, Le Cesne B et al. Early lymphopenia after cytotoxic chemotherapy as risk factor for febrile neutropenia. J Clin Oncol 1996; 14: 636–643.
8. Silber JH, Fridman M, Robert S et al. First-cycle blood counts and subsequent neutropenia, dose reduction or delay in early-stage breast cancer therapy. J Clin Oncol 1998; 16: 2392–2400.[Abstract]
9. Ray-Coquard I, Borg C, Bachelot T et al. Baseline and early lymphopenia predict for the risk of febrile neutropenia after chemotherapy. Br J Cancer 2003; 88: 181–186.[Medline]
10. Kondo M, Oshita F, Kato F et al. Early monocytopenia after chemotherapy as a risk factor for neutropenia. Am J Clin Oncol 1999; 22: 103–105.[CrossRef][ISI][Medline]
11. Voog E, Bienvenu J, Waezocha K et al. Factors that predict chemotherapy-induced myelosuppression in lymphoma patients: role of the tumor necrosis factor ligand-receptor system. J Clin Oncol 2000; 18: 47–56.
12. Wilson-Royalty M, Lawless G, Palmer G et al. Predictors for chemotherapy-related severe or febrile neutropenia: a review of the literature. J Oncol Pharm Pract 2002; 7: 141–147.
13. Lyman GH, Dale DC, Crawford J et al. Incidence practice patterns, and predictors of low dose-intensity in adjuvant breast cancer chemotherapy: results of a nationwide survey of community oncology practices. J Clin Oncol 2003; 21: 4524–4531.
14. Wood WC, Budman DR, Korzun AH et al. Dose and dose-intensity of adjuvant chemotherapy for stage II node-positive breast carcinoma. N Engl J Med 1994; 330: 1253–1259.
15. 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 1991; 325: 164–170.[Abstract]
16. Lyman GH. A predictive model for neutropenia associated with cancer chemotherapy. Pharmacotherapy 2000; 20: 104S–111S.[CrossRef][ISI][Medline]
17. Paesmans M. Risk factors assessment in febrile neutropenia. Int J Antimicrob Agents 2000; 16: 107–111.[CrossRef][ISI][Medline]
18. Klastersky J, Paesmans M, Rubenstein EB et al. The Multinational Association for Supportive Care in Cancer risk index: a multinational scoring system for identifying low-risk febrile neutropenic cancer patients. J Clin Oncol 2000; 18: 3038–3051.
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||