Oncologist. The. Symptom Management and Supportive Care

Transcription

1 The Oncologist Symptom Management and Supportive Care Prediction of the Responsiveness to Treatment with Erythropoiesis-Stimulating Factors: A Prospective Clinical Study in Patients with Solid Tumors TILMAN STEINMETZ, a MARTIN HELLMICH, b MICHAEL NEISE, c ALI ALDAUD, d CHRISTIAN LERCHENMÜLLER, e ANTONIS TSAMALOUKAS, f FRANCOISE FANDEL, a CHRISTOPH WEILIGMANN, e UWE TOTZKE, g STEPHAN SCHMITZ a Oncologic Outpatient Clinics in a Cologne, c Krefeld, d Leipzig, e Münster, and f Hilden, Germany; b Institute of Medical Statistics, Informatics and Epidemiology, University of Cologne, Cologne, Germany; g HPM Healthcare Project Management, Geneva, Switzerland Key Words. Anemia Erythropoietic proteins Predictive factors Corresponding receiver operating characteristic curve ABSTRACT Objective. Treatment with erythropoiesis-stimulating factors (ESFs) can ameliorate anemia associated with cancer and chemotherapy. However, half of anemic cancer patients do not respond even to high doses. To determine factors that are predictive of a treatment response, a multicenter, prospective study was performed. Patients and Methods. Investigated factors were baseline erythropoietin, reticulocytes and soluble transferrin receptor (stfr) after 2 weeks, and reticulocytes and hemoglobin after 4 weeks. Anemic patients with solid tumors received 150 g/week of darbepoetin concomitantly with chemotherapy. The dose was doubled if hemoglobin did not increase by >1 g/dl after 4 weeks. Patients were considered responders if hemoglobin increased by >2 g/dl or reached a level >12 g/dl within 8 12 weeks. Results. In total, 196 patients were enrolled; 61% of the intention-to-treat (ITT) and 68% of the per-protocol population were responders. In the ITT population, the hemoglobin increase after 4 weeks indicated an 11-fold higher chance of response (odds ratio, 11.0; 95% confidence interval [CI], ; sensitivity, 88%; specificity, 60%). In a multiple logistic regression model including all factors, the area under the receiver operating characteristic curve was 0.78 (95% CI, ). The combination of stfr after 2 weeks and hemoglobin after 4 weeks was as predictive as the combination of all five tested factors. Conclusion. So far, an early hemoglobin increase remains the single most predictive factor for response to ESF treatment. In contrast to anemic patients with lymphoproliferative malignancies, serum erythropoietin had little predictive value in patients with solid tumors. The Oncologist 2007;12: Disclosure of potential conflicts of interest is found at the end of this article. INTRODUCTION Malignant diseases along with chemotherapy often result in anemia [1, 2]. The most important symptom associated with anemia is fatigue [3], which affects 75% of patients and prevents over 90% of them from maintaining a normal life [4]. Moreover, low hemoglobin (Hb) levels have a negative im- Correspondence: H. Tilman Steinmetz, M.D., Praxis für Hämatologie und Onkologie, Sachsenring 69, Cologne, Germany. Telephone: ; Fax: ; steinmetz@oncokoeln.de Received January 29, 2007; accepted for publication March 29, AlphaMed Press /2007/$30.00/0 doi: /theoncologist The Oncologist 2007;12:

2 Steinmetz, Hellmich, Neise et al. 749 pact on cancer prognosis and therapy outcome [5 9]. In order to improve quality of life and prognosis, the treatment and prevention of anemia is now considered an integral part of the management of malignant diseases. Treatment with transfusions is inconvenient for the patient, acts transiently and insufficiently at peripheral tissues, and is associated with increased risks for infections and adverse immune reactions, which also worsen the prognosis [10, 11]. Erythropoiesis-stimulating factors (ESFs), such as hyperglycosylated darbepoetin alfa (DA), constitute an alternative treatment with proven efficacy and without most of the disadvantages associated with transfusions. However, treatment with ESFs is expensive and, unlike in end-stage renal disease, response rates in cancer amount to only 50% 70%, although they tend to increase with treatment duration [12]. Among the factors known to influence the responsiveness to ESF treatment are the type and state of the underlying malignant disease, the existence of other potential causes of anemia (e.g., hemorrhage), functional iron deficiency, and the type and dose of chemotherapy. Baseline levels of Hb, hematocrit, and endogenous erythropoietin (EPO), as well as increases in Hb, reticulocytes, and soluble transferrin receptor (stfr) during the initial treatment phase with ESF, have been evaluated as predictive factors for treatment response in several studies [13 18]. However, no study has revealed any baseline predictive factor that can be routinely used in clinical practice [19] and only a few small clinical trials have validated predictive factors prospectively [20 22]. A recent Cochrane review [23] concluded that virtually all studies to date assessing predictive factors of response to EPO were small, exploratory in nature, and of low reporting quality. The review stated that the current evidence did not allow for definite conclusions to be made and called for larger, high-quality prospective studies. In line with these recommendations, the current multicenter, prospective study was performed in ambulatory patients with solid tumors who had been attending a regular medical practice. MATERIALS AND METHODS Study Design This was a 12-week prospective trial in outpatients receiving commonly used chemotherapy allowing one dose escalation after 4 weeks. It was approved by the corresponding ethics committee and conducted in accordance with the declaration of Helsinki, good clinical practices, and legal obligations (AMG). All participants gave written informed consent prior to participation. Figure 1. Study populations. Abbreviations: ITT, intention to treat; PP, per protocol; stfr, soluble transferrin receptor. Patients and Treatment Eligible patients had to have an Eastern Cooperative Oncology Group performance status score 2 and anemia (defined as either a total Hb 11g/dl or a Hb decrease 1.5 g/dl over 4 weeks). Patients received DA s.c. for 12 weeks concomitantly with chemotherapy at doses of either 150 g/week or 300 g every other week. The dosage was adjusted to 300 g/week if Hb did not increase by 1 g/dl by week 4. DA treatment was prematurely discontinued when Hb exceeded 14 g/dl in responders or discontinued for no response if the level changed by 1 g/dl after week 8. DA (Aranesp ; Amgen Inc., Thousand Oaks, CA) is an analogue of recombinant human erythropoietin (rhuepo). Its higher number of sialic acid moieties compared with rhuepo results in lower affinity for the erythropoietin receptor but a longer half-life that ultimately makes DA more potent in vivo than rhuepo. Administration can be less frequent and thus more easily synchronized with chemotherapy. In randomized, controlled clinical trials, s.c. DA given at appropriate doses once every 1 3 weeks has been shown to be an effective and well-tolerated erythropoietic agent in anemic cancer patients undergoing chemotherapy [24 27]. Response Criteria Response was defined as an increase in Hb of 2 g/dl from baseline or as a total level of 12 g/dl to be achieved within

3 750 Prediction of Responsiveness to ESF Treatment Table 1. Baseline characteristics of the intention-to-treat population Characteristic Total Nonresponder Responder Age (years) a Weight (kg) a d 14.4 Height (cm) a Sex b Male Female Performance status b according to WHO/ECOG Hematocrit (%) c Erythrocytes (10 12 /l) c Reticulocytes ( ) c Leukocytes (10 9 /l) c Platelets (10 9 /l) c a Mean and standard deviation. Percentage and number of patients. Median, percentiles. d Responders versus nonresponders, Wilcoxon rank sum test p.006. Abbreviations: ECOG, Eastern Cooperative Oncology Group; WHO, World Health Organization. Table 2. Requirement for concomitant anemia therapy or DA dose escalation ( 150 g DA at week 4) Total Nonresponder Responder Intervention % n % n % n Iron therapy No Yes Transfusion therapy No Yes a 9 Dose escalation No Yes b 29 a Responders versus nonresponders, 2 test, p b p Abbreviation: DA, darbepoetin alfa weeks. For patients who were transfused during the first month of the study, the Hb level determined immediately afterward was taken as baseline. Patients who received a transfusion between week 5 and week 8 were considered nonresponders. The following prespecified criteria were investigated for their predictive value of the treatment response: (a) serum EPO 100 U/l at baseline or the ratio of the observed (0.093 hematocrit [%]) to predicted level (O/P) (P 10 [28]) 0.9 at baseline, (b) reticulocytes 40,000/ l after 2 weeks, (3) an increase in stfr 25% after 2 weeks, (d) reticulocytes 40,000/ l after 4 weeks, and (e) an increase in Hb 1 g/dl after 4 weeks. Blood samples were taken every other week and analyzed at the local sites for EPO, Hb, reticulocytes, and stfr. At baseline, vital signs and safety laboratory parameters were also measured.

4 Steinmetz, Hellmich, Neise et al. 751 Table 3. Predictive value (OR), positive/negative LR, and corresponding 95% CI of the five predictive factors (EPO at baseline, reticulocytes and stfr increase after 2 weeks, and reticulocytes and Hb increase after 4 weeks) in the ITT population Factor Day OR (95% CI) LR (95% CI) LR (95% CI) EPO, O/P ratio ( ) 1.1 ( ) 0.6 ( ) EPO 100 U/l ( ) 1.1 ( ) 0.6 ( ) Reticulocytes 40,000/ l ( ) 1.6 ( ) 0.9 ( ) stfr increase 25% ( ) 1.4 ( ) 0.5 ( ) Reticulocytes 40,000/ l ( ) 2.4 ( ) 0.9 ( ) Hb increase 1g/dl ( ) 4.8 ( ) 0.4 ( ) LR values 5 and LR values 0.2 indicate strong diagnostic evidence. Abbreviations: CI, confidence interval; EPO, erythropoietin; Hb, hemoglobin; ITT, intention-to-treat; LR, likelihood ratio; O/P, observed to predicted level; OR, odds ratio; stfr, soluble transferrin receptor. Data Sets Patients who fulfilled all inclusion and none of the exclusion criteria and for whom the response to DA could be evaluated entered the intention-to-treat (ITT) population. All patients without any protocol violation and for whom all scheduled predictive factors were recorded comprised the per-protocol data set (Fig. 1). Statistical Analysis The predictive value of each factor was evaluated by calculating odds ratios (ORs) and likelihood ratios (LRs). In addition, combinations of factors were evaluated based on multiple logistic regression models and corresponding receiver operating characteristic (ROC) curves. The ROC curve depicts the trade-off between sensitivity and 1 specificity for various cutoffs used to classify patients as either a responder or nonresponder. The discriminatory power of a combination of factors is usually summarized by the area under the ROC curve (AUC) (1, perfect discrimination; 0.5, tossing a coin, which corresponds to the random ROC ). Differences between nested models were tested by LR tests. The calibration of models, that is, the agreement of observed and predicted proportions of response, was assessed by the Hosmer Lemeshow statistic. The association between the Hb increase within 8 12 weeks and that after 4 weeks of DA treatment was analyzed by ordinary least-squares regression, 80% prediction limits, and Pearson s correlation coefficient. The sample size required to obtain a reliable classification rule was calculated based on the rule of thumb 10 events per variable of interest [29] and an expected 40% proportion of nonresponders. To account for dropouts and missing values, a total sample size of 200 was targeted. Statistical analyses were performed with SPSS (SPSS Inc., Chicago, IL) and R (R Foundation for Statistical Computing, Vienna, Austria). Figure 2. ROC curves based on ITT data and multiple logistic regression models. (A): Includes all five factors (AUC, 0.78; 95% CI, ). (B): Includes the three factors available after 2 weeks (AUC, 0.63; 95% CI, ). (C): Includes only baseline EPO (AUC, 0.54; 95% CI, ). Abbreviations: AUC, area under the curve; CI, confidence interval; EPO, erythropoietin; ITT, intention to treat; ROC, receiver operating characteristic curve. RESULTS In total, 299 patients were screened; 196 of them, 112 women and 84 men with a mean age of 64 years, were enrolled, treated, and evaluated for their response to DA (Fig. 1). About 61% of the ITT and 68% of the PP population were responders (online supplementary Figure 1). Apart from body weight, responders and nonresponders did not differ significantly in demographics (Table 1), vital signs

5 752 Prediction of Responsiveness to ESF Treatment Table 4. Multiple logistic regression models with one (baseline), three (after 2 weeks of DA treatment), or five factors (after 4 weeks of DA treatment), or two selected factors, based on the ITT population Day B SE (B) Wald test OR Hosmer Lemeshow test LR test p-value Exp (B) 95% CI 2 LL df p-value Statistic df p-value Constant EPO 100 U/l Constant EPO 100 U/l Reticulocytes 40,000/ l stfr increase 25% Constant EPO 100 U/l Reticulocytes 40,000/ l stfr increase 25% Reticulocytes 40,000/ l Hb increase 1 g/dl Constant stfr increase 25% Hb increase 1 g/dl Constant Abbreviations: B, coefficient; CI, confidence interval; DA, darbepoetin alfa; df, degrees of freedom; EPO, erythropoietin; Hb, hemoglobin; ITT, intention to treat; OR, odds ratio; LL, log likelihood; LR, likelihood ratio; SE, standard error; stfr, soluble transferrin receptor.

6 Steinmetz, Hellmich, Neise et al. 753 (data not shown), or laboratory data at baseline (Table 1) (clinical chemistry data not shown). Nonresponders required significantly more frequent transfusions and DA dose escalation (p.001; Table 2). Patients who displayed an increase in Hb after 4 weeks had an 11-fold higher chance of being responders than those who did not. Taken alone, all other investigated factors had rather low predictive values without clinical relevance (Table 3). ROC curves corresponding to multiple logistic regression models including all five factors revealed an AUC of 0.78 (95% confidence interval [CI], ) (Fig. 2) in the ITT population and 0.83 (95% CI, ; data not shown) in the PP population. Blockwise removal of factors, first of those evaluated after 4 weeks and then of those evaluated after 2 weeks, consistently decreased the predictive value of the model (Fig. 2 and Table 4). Following the opposite approach, removal of baseline EPO and of reticulocytes (determined after 2 or 4 weeks) from the model had no impact (AUC, 0.78; 95% CI, ; Table 4). Calibration of all logistic models appeared to be good (Hosmer Lemeshow statistic 2). Multiple linear regression models of the Hb increase within 8 12 weeks with any of the tested factors were not significant, except for the Hb increase measured after 4 weeks (Fig. 3). DISCUSSION This study is the first to prospectively evaluate predictive factors for the responsiveness of ambulatory cancer patients to ESF treatment based on regular medical practice conditions. These conditions were important because the study aimed to provide a valid and readily usable instrument for treatment decisions in oncologists day-to-day practice. In a conservative approach, transfused patients were included in the analysis. Given this, the study revealed relatively high response rates but, like all former studies [23], failed to find a single predictive factor or combinations of factors that would allow a reliable response prediction prior to or during early treatment. However, it did provide: (a) for the first time, a statistical model with Hb and stfr as predominant factors aiding oncologists day-to-day decision of whether or not to continue ESF therapy at least after 4 weeks of treatment, and (b) some new insights into hematopoiesis that may also affect routine practice. In contrast to anemic patients with lymphoproliferative malignancies [17, 30, 31], neither the baseline EPO level nor the O/P ratio predicted response to DA in anemic patients with solid tumors. This confirms isolated observations from retrospective analyses [16, 30]; however, there is no conclusive explanation for such a difference and there is a need for confirmation in further prospective trials enrolling and directly comparing both types Figure 3. Linear regression of the Hb increase within weeks 8 12 with the one measured after 4 weeks of DA treatment (broken lines reflect 80% prediction limits). Patients transfused after 4 weeks were considered nonresponders. Abbreviations: DA, darbepoetin alfa; Hb, hemoglobin. of patients. Contrary to previous studies [14, 18], reticulocytes also had little relevance for response prediction, which might be related to the simultaneous chemotherapy, because this was also observed in another trial with patients who were treated concomitantly [32]. The slightly higher predictive value of the stfr increase was, however, in line with former studies [14] and could be expected, because it is thought to correlate with an increase in the mass of erythropoietic precursors within the bone marrow. The factor with the highest predictive value was the Hb increase after 4 weeks of treatment. Increases in Hb are generally easy to measure and have also been established as the main predictive factor by former studies [13, 18, 22, 32]. A least-squares regression of the Hb increase within 8 12 weeks with that measured after 4 weeks of DA treatment yielded a simple prediction rule that might be helpful in clinical practice, that is, Hb week 8 12 [g/dl] 0.7 Hb week ( 2 for 80% of cases) (Fig. 3). However, the value of such models generally remains uncertain in practice. If after 4 weeks of ESF treatment the probability of an increase in Hb is low, should therapy then be stopped? What level of certainty would be needed to continue? To answer these kinds of questions it must be taken into account that hemoglobin is just a surrogate marker for the real objectives of ESF therapy: (a) the improvement of quality of life and (b) the prevention of transfusions. However, neither of these has yet been built into models predicting ESF treatment response. Thus, further

7 754 Prediction of Responsiveness to ESF Treatment studies are needed to identify factors that allow for the assessment of the efficacy of ESF therapy prior to or during early treatment in order to prevent side effects, reduce costs, and improve quality of life. ACKNOWLEDGMENTS The study was supported by Amgen GmbH, Germany, and was presented as abstract at the American Society of Clinical Oncology 2006 Annual Meeting. DISCLOSURE OF POTENTIAL CONFLICTS OF INTEREST T.S. has acted as a consultant for Amgen, Hoffmann-La Roche, and Ortho Biotech. C.L. has acted as a consultant for Amgen. REFERENCES 1 Khayat D. Is anemia a problem for European cancer patients and treating oncologists? Semin Oncol 2000;27(suppl 4):9 11; discussion Barrett-Lee PJ, Bailey NP, O Brien ME et al. Large-scale UK audit of blood transfusion requirements and anaemia in patients receiving cytotoxic chemotherapy. Br J Cancer 2000;82: Ashbury FD, Findlay H, Reynolds B et al. A Canadian survey of cancer patients experiences: Are their needs being met? J Pain Symptom Manage 1998;16: Cella D. The Functional Assessment of Cancer Therapy-Anemia (FACT- An) Scale: A new tool for the assessment of outcomes in cancer anemia and fatigue. Semin Hematol 1997;34(suppl 2): Fein DA, Lee WR, Hanlon AL et al. Pretreatment hemoglobin level influences local control and survival of T1 T2 squamous cell carcinomas of the glottic larynx. J Clin Oncol 1995;13: Dubray B, Mosseri V, Brunin F et al. Anemia is associated with lower localregional control and survival after radiation therapy for head and neck cancer: A prospective study. Radiology 1996;201: Waters JS, O Brien ME, Ashley S. Management of anemia in patients receiving chemotherapy. J Clin Oncol 2002;20: Littlewood TJ, Bajetta E, Nortier JW et al. Effects of epoetin alfa on hematologic parameters and quality of life in cancer patients receiving nonplatinum chemotherapy: Results of a randomized, double-blind, placebocontrolled trial. J Clin Oncol 2001;19: Glaser CM, Millesi W, Kornek GV et al. Impact of hemoglobin level and use of recombinant erythropoietin on efficacy of preoperative chemoradiation therapy for squamous cell carcinoma of the oral cavity and oropharynx. Int J Radiat Oncol Biol Phys 2001;50: Heiss MM, Mempel W, Delanoff C et al. Blood transfusion-modulated tumor recurrence: First results of a randomized study of autologous versus allogeneic blood transfusion in colorectal cancer surgery. J Clin Oncol 1994;12: Jensen LS, Kissmeyer-Nielsen P, Wolff B et al. Randomised comparison of leucocyte-depleted versus buffy-coat-poor blood transfusion and complications after colorectal surgery. Lancet 1996;348: Malik IA, Khan ZK, Hakimali A et al. The effect of subcutaneous recombinant human erythropoietin (r-huepo) on anemia in cancer patients receiving platinum-based chemotherapy. J Pak Med Assoc 1998;48: Goram AL. Factors and predictors of response with epoetin alfa for chemotherapy-related anemia. J Pharm Technol 2000;16: Cazzola M, Ponchio L, Pedrotti C et al. Prediction of response to recombinant human erythropoietin (rhuepo) in anemia of malignancy. Haematologica 1996;81: Littlewood TJ, Zagari M, Pallister P. Pretreatment erythropoietin levels cannot be used to predict individual response to epoetin alfa in anaemic cancer patients. Curr Med Res Opin 2005;21(suppl 2):S19 S Bokemeyer C, Aapro MS, Courdi A et al. EORTC guidelines for the use of erythropoietic proteins in anaemic patients with cancer. Eur J Cancer 2004; 40: Kolesar JM, Duren BA, Baranski BG. Retrospective evaluation of response to epoetin alfa in patients with hematologic disorders. J Oncol Pharm Pract 2004;10: Littlewood TJ, Zagari M, Pallister C et al. Baseline and early treatment factors are not clinically useful for predicting individual response to erythropoietin in anemic cancer patients. The Oncologist 2003;8: Bokemeyer C, Aapro MS, Courdi A et al. EORTC guidelines for the use of erythropoietic proteins in anaemic patients with cancer: 2006 update. Eur J Cancer 2007;43: Katodritou E, Speletas M, Kapetanos D et al. Evaluation of traditional and novel predictive markers of anemia response to recombinant human erythropoietin (rh-epo) in patients with multiple myeloma and lymphoma: Emerging role of functional iron deficiency. Blood 2004;104:8B. 21 Katodritou E, Speletas M, Zervas K et al. Evaluation of hypochromic erythrocytes in combination with stfr-f index for predicting response to r- HuEPO in anemic patients with multiple myeloma. Lab Hematol 2006;12: Gonzalez-Baron M, Ordonez A, Franquesa R et al. Response predicting factors to recombinant human erythropoietin in cancer patients undergoing platinum-based chemotherapy. Cancer 2002;95: Trelle S, Piper M, Bohlius J et al. Predictive factors of response to recombinant human erythropoietin: Quality and results of studies identified by a systematic review. Onkologie 2006;29(suppl 3): Canon JL, Vansteenkiste J, Bodoky G et al. Randomized, double-blind, active-controlled trial of every-3-week darbepoetin alfa for the treatment of chemotherapy-induced anemia. J Natl Cancer Inst 2006;98: Vansteenkiste J, Pirker R, Massuti B et al. Double-blind, placebo-controlled, randomized phase III trial of darbepoetin alfa in lung cancer patients receiving chemotherapy. J Natl Cancer Inst 2002;94: Hedenus M, Adriansson M, San Miguel J et al. Efficacy and safety of darbepoetin alfa in anaemic patients with lymphoproliferative malignancies: A randomized, double-blind, placebo-controlled study. Br J Haematol 2003; 122: Glaspy J, Henry D, Patel R et al. Effects of chemotherapy on endogenous erythropoietin levels and the pharmacokinetics and erythropoietic response of darbepoetin alfa: A randomised clinical trial of synchronous versus asynchronous dosing of darbepoetin alfa. Eur J Cancer 2005;41: Beguin Y, Yerna M, Loo M et al. Erythropoiesis in multiple myeloma: Defective red cell production due to inappropriate erythropoietin production. Br J Haematol 1992;82:

8 Steinmetz, Hellmich, Neise et al Harrell F Jr. Regression Modeling Strategies. New York: Springer, 2001: Cheer SM, Wagstaff AJ. Epoetin beta: A review of its clinical use in the treatment of anaemia in patients with cancer. Drugs 2004;64: Cazzola M, Beguin Y, Kloczko J et al. Once-weekly epoetin beta is highly effective in treating anaemic patients with lymphoproliferative malignancy and defective endogenous erythropoietin production. Br J Haematol 2003; 122: Lastiri JM, Specterman SR, Rendo P et al. [Predictive response variables to recombinant human erythropoietin treatment in patients with anemia and cancer.] Medicina (B Aires) 2002;62: