|
Advertisement |
|
|
||
 |
|
|||||
|
|||||
|
 |
||||||
|
||||||
|
||||||
|
Originally published as JCO Early Release 10.1200/JCO.2005.02.9488 on June 12 2006 © 2006 American Society of Clinical Oncology. Does Granulocyte Colony-Stimulating Factor Worsen Anemia in Early Breast Cancer Patients Treated With Epirubicin and Cyclophosphamide?
From the Department of Medical Oncology; Biostatistics Unit; Department of Surgery; Department of Pathology, Regina Elena Cancer Institute; Department of Medical Oncology, S. Andrea Hospital; Department of Medical Oncology, University "La Sapienza" School of Medicine; and the Division of Medical Oncology, S. Eugenio Hospital, Rome, Italy Address reprint requests to Gianluigi Ferretti, MD, PhD, Division of Medical Oncology "A", Regina Elena Cancer Institute, Via Elio Chianesi 53, 00144 Rome, Italy; e-mail: gia.fer{at}flashnet.it
PURPOSE: We report on the effects of granulocyte colony-stimulating factor (G-CSF) on hemoglobin (Hb) value in early breast cancer patients receiving high-dose epirubicin and cyclophosphamide (EC) adjuvant treatment. METHODS: Five hundred and six stage I or stage II female breast cancer patients were treated with E 120 mg/m2 and C 600 mg/m2 with or without G-CSF and randomly assigned to receive in a factorial 2 x 2 design: EC; EC + lonidamine; EC + G-CSF; EC + lonidamine + G-CSF. Five consecutive G-CSF schedules tested 100 randomly assigned patients each: (1) 480 µg subcutaneously on days 8 to 14; (2) 480 µg on days 8, 10, 12, 14; (3) 300 µg on days 8 to 14; (4) 300 µg on days 8, 10, 12, and 14; and (5) 300 µg on days 8 and 12. The mean Hb level of 246 patients receiving EC plus G-CSF was compared with that of 240 patients receiving EC alone. The data presented are derived from an exploratory hypothesis-generating analysis. RESULTS: The EC dose intensity did not statistically differ between the G-CSF and the control arm. From the third cycle onward, the mean Hb value resulted significantly lower in G-CSF arm compared with control at each time point of each cycle (P < .0001). No statistically significant difference in the mean Hb level was observed between schedule 5 and control. Of interest, from the second course onward, the mean Hb level tended to be lower in patients receiving seven or four G-CSF injections compared with those patients who received only two injections. CONCLUSION: Our data suggest that a G-CSF dose-related effect may play a role in worsening anemia in patients receiving adjuvant EC.
Granulocyte colony-stimulating factor (G-CSF) is the major regulator of in vivo granulopoiesis. It is widely administered to patients who recover from chemotherapy to amplify the immature granuloid cells and to shorten the neutropenic period. Different mechanisms of G-CSF action have been suggested: a shortening in granuloid progenitor cell cycle, a reduction of the average transit time of granulocyte production, and a prevention of apoptosis of responsive granuloid cells.1 However, little is known about whether and how the production of one cell lineage is affected when others are stimulated, especially in the clinical setting. The association of epirubicin (E; 120 mg/m2) with cyclophosphamide (C; 600 mg/m2), given on day 1 of each 21-day cycle, appeared to be an effective treatment against advanced breast cancer, but preliminary data reported a dose-limiting neutropenia.2 In order to evaluate the efficacy of a high-dose anthracycline-based regimen and to investigate the clinical value of G-CSF and lonidamine (LND), an energolytic derivative of indazole-carboxylic acid capable of reversing the resistance to anthracyclines, combined with chemotherapy in early breast cancer patients, we tested EC ± LND ± G-CSF in a randomized phase III trial with a 2 x 2 factorial design with the intent to compare the control with LND– and with G-CSF arms. EC was active and well tolerated in early breast cancer patients. The addition of LND or G-CSF did not improve disease-free or overall survival. The final results of this study have been previously described elsewhere.3,4 We hereby report on the effects of G-CSF administration on the mean hemoglobin (Hb) level compared with that observed without G-CSF administration in early breast cancer patients receiving high-dose EC adjuvant treatment.
Female patients with stage I or II breast cancer3 were randomly assigned to receive in a factorial 2 x 2 design: arm A: four cycles of chemotherapy with high-dose EC (120 mg/m2 and 600 mg/m2, respectively) on day 1 every 21 days; arm B: EC plus LND; arm C: EC plus G-CSF; arm D: EC plus LND plus G-CSF. G-CSF (filgrastim) was prophylactically administered from day 8 after each chemotherapy cycle, 24 to 48 hours before the expected neutrophil nadir.4,5 The control arm used in the analyses included both the EC and EC plus LND arms, and the G-CSF arm included both the EC plus G-CSF and EC plus LND plus G-CSF arms. In the control arm, G-CSF was not permitted and dose reduction was performed as per protocol in case of toxicity. The study was conducted in accordance with institutional ethical standards and approved by an independent ethical committee. Oral informed consent was obtained from all patients before random assignment. A total of 506 female patients were randomly assigned between October 1991 and April 1994. Four hundred and ninety seven were fully assessable for treatment efficacy. The G-CSF arm included 254 patients; the control arm consisted of 243 females. The patients' features were comparable across each arm and G-CSF group. At the time of study planning, the standard for G-CSF administration was not yet established, so to assess the optimal G-CSF dose, two doses (480 µg and 300 µg) and five different schedules based on the analysis of neutrophil overshoots,4 were tested in five consecutive cohorts of patients (100 patients enrolled in each group) as follows: 53 patients were included in group 1 and received G-CSF 480 µg SC injection every day (qd) from day 8 to day 14 (schedule 1); 55 patients in group 2 with G-CSF 480 µg every other day (qod) on days 8, 10, 12, and 14 (schedule 2); 43 patients in group 3 with G-CSF 300 µg qd from day 8 to 14 (schedule 3); 52 patients in group 4 with G-CSF 300 µg qod on days 8, 10, 12 and 14 (schedule 4); and 51 patients in group 5 with G-CSF 300 µg on days 8 and 12 (schedule 5). Our aim was to compare the mean Hb level in patients assigned to G-CSF arm versus the control arm. Moreover, we evaluated the mean Hb level in the five groups of patients consecutively assigned to one of five filgrastim schedules described herein. The data presented are derived from an exploratory hypothesis-generating analysis. Concerning the Hb level, data were considered adequate for 246 patients in the G-CSF arm (schedule 1: 53 patients; schedule 2: 52 patients; schedule 3: 39 patients; schedule 4: 51 patients; schedule 5: 51 patients) and 240 patients in the control arm. The chemotherapy dose modifications have been previously reported.3 The CBC count was performed at baseline and weekly, on days 7, 14, and 21 in each group for each cycle. Because at the time of trial start the dose of E was considered high, physicians carefully controlled the hematologic parameters and assessed toxicity, because one of the goals of the study was to investigate the G-CSF effect on the blood cell count. During the accrual, these analyses were performed every 6 months. Moreover, the patients were punctual in performing each CBC count as required by the protocol: approximately 88% of the total theoretical Hb values were registered (5580 of 6318).
Comparisons of treatment toxicity among arms and groups were performed by the
Although the rate of delayed cycles and the frequency of dose reduction were greater in non–G-CSF arm than in G-CSF arm (10% v 3.6%; P = .00001; and 3.6% v 1.4%; P = .002, respectively), the EC dose intensity did not statistically differ between the G-CSF and control arm (98.1% v 95.5% in G-CSF and non–GCSF, respectively; P = .17). Results of the major outcomes and clinical aspects of neutropenia and other hematologic variables have been reported in our previous articles.3,4 All G-CSF schedules significantly reduced hematologic toxicity compared with control (non–G-CSF arm).4 Patients recruited in the G-CSF arm showed early increases in absolute neutrophil count and had patterns of neutrophils overshoot on day 14 related to the G-CSF dose, when the neutrophil nadir occurred in the control arm. No patient received erythropoietin. Five patients were transfused with RBCs. Two of these patients had been assigned to EC alone and received one RBC unit each after the third and fourth chemotherapy cycle, respectively. The other two patients had been randomly assigned to EC plus G-CSF: the patient assigned to group 1 received 3 RBC units after the third cycle, while the patient assigned to group 2 received 2 RBC units after the second course of chemotherapy. The fifth patient had been assigned to EC plus G-CSF (schedule 2) plus LND and received 2 RBC units after the first cycle of treatment.
Comparison between G-CSF arm and control arm.
G-CSF schedule comparisons. No statistically significant difference in the mean Hb level was observed between schedule 5 and control (Table 2). After the second course of chemotherapy, the mean Hb value tended to be lower in patients receiving seven or four G-CSF injections compared with those given only two filgrastim injections (Table 3). These results seem to suggest a possible influence of the G-CSF dose and G-CSF-induced neutrophilia on the reduction of the mean Hb level (Fig 2). No significant difference in the mean Hb level was registered in patients receiving seven G-CSF injections compared with those administered four injections (Table 4).
Platelets. No statistically significant difference in platelet count was observed between the control and G-CSF arms and throughout the G-CSF schedules.
LND.
Delayed cycles.
Repeated measures analysis. Considering the mean Hb value in patients who received G-CSF versus those who did not receive it, the analysis showed that the mean Hb level was significantly correlated with time (P < .0001), G-CSF use (P = .04), and interaction of time by G-CSF (P < .0001). Comparing the Hb level of patients who delayed chemotherapy with that of patients who did not, the repeated measures analysis showed that the mean Hb value was significantly correlated with time (P < .0001), but not with treatment delay (P = .80) or interaction of time by delay (P = .14).
Although the role of G-CSF in erythropoiesis has been widely investigated, several controversial issues remain. In vitro experiments showed that both burst forming unit erythroid (BFU-E) colony growth and Hb synthesis were inhibited by CSF,6 even though these observations were performed with impure preparations of G-CSF6 and purified recombinant human G-CSF was described to enhance BFU-E formation.7 In murine models, G-CSF was reported to stimulate the initial proliferation of erythroid progenitors,8 but it did not appear to be able to sustain continued proliferation of these cells to result in colony formation.9 Long-term G-CSF treatment seems to have opposite effects over murine bone marrow and splenic erythropoiesis, reducing the former and enhancing the latter,10,11 thus causing erythroid depression in splenectomized mice.12 In mice, it has been hypothesized that the granulopoietic and erythropoietic lineages may compete for differentiating stem cells.11 Thus, G-CSF may activate stem cells to differentiate at the expense of self-renewal causing a permanent loss of marrow reserve. G-CSF can damage stem cells directly or indirectly, making them more susceptible to chemotherapy by activation of proliferation or migration. G-CSF induces stem cells to differentiate into more committed hematopoietic cells, resulting in a severe reduction of especially the most primitive stem cells.13 This effect is enhanced in animals with an already compromised hematopoietic stem-cell compartment, as seen with repeated doses of cytotoxic agents.14 According to this hypothesis, one would expect that a stimulation of granulopoiesis by G-CSF administration would lead to a reduction of the stem-cell pool and be followed by a decline of erythropoietic progenitor numbers (lineage steal). However, de Haan et al13 demonstrated that in a murine model the inhibiting effect of G-CSF on erythropoiesis predominantly took place in the erythropoietin-responsive cell compartment (colony forming unit erythroid). The effects on more immature cells were less clear, because red cell progenitor compartment (BFU-E) was stimulated by low doses but inhibited by high doses of G-CSF. The reduction of these immature cell stages was probably caused by the mobilization of these cells from the marrow to the blood (both G-CSF and erythropoietin are known to induce mobilization) or eventually induced by a strongly stimulated differentiation of BFU-E and colony forming unit granulocyte macrophage into later cell stages as well. Thus, the inhibition took place at a late level of differentiation, and these findings provoked the question of whether competition for a common stem cell exists. In humans, G-CSF has been described to positively affect erythropoiesis without involving erythropoietin15 and to improve the effect of erythropoietin when previously administered in HIV patients.16 In contrast, children with aplastic anemia treated with G-CSF showed reduced erythroid/myeloid ratio, indicating a suppressed erythropoiesis,17 though the administration of G-CSF to neonates did not result in down modulation of erythropoiesis.18 In cancer patients, it has been reported that G-CSF does not affect the Hb value when administered before,19 during,20 or after chemotherapy.21 Our results clearly showed that, with a comparable dose intensity, the rate of grade 2 or worse anemia was higher in the G-CSF arm (38.8%) than in control arm (26.2%; P = .005), even though EC caused more than 1 g Hb loss in non–G-CSF arm (Fig 1) at the end of treatment. The significant decrease of the mean Hb level after the third course onward in G-CSF arm compared with control would suggest that anemia could not be solely induced by chemotherapy but also by G-CSF. Moreover, our results suggest a possible relation between the dose of G-CSF, the number of G-CSF injections, G-CSF-induced neutrophilia, and the mean Hb level registered, indicating that not G-CSF per se but its dosage could play a role in worsening anemia in patients treated with adjuvant EC. In fact, the mean Hb level tended to be lower in G-CSF schedules 1 to 4 compared with G-CSF schedule 5 (Fig 2). It must be highlighted that the Hb decrease was minimal with the shortened filgrastim schedule and no significant differences in the mean Hb value was registered between the shortened schedule and control, even though schedule 5 seemed capable of protecting patients from neutropenia, as previously reported.4 The observation that the shortened schedule kept neutrophil values within the normal range and did not induce neutrophilia compared with the prolonged ones4 supports the hypothesis that red cell progenitor compartment (BFU-E) can be inhibited by high doses of G-CSF,11 which could cause the reduction of immature cell stages by the mobilization of these cells from the marrow to the blood, causing competition for a common stem cell. The information provided may have potential relevance to optimize the efficacy of growth factors alone or in combination with other biotherapy and/or chemotherapy in clinical practice, and should be taken into account when designing a dose-dense chemotherapy study. In the study by Citron et al22, 13% of patients on a dose dense (every 2-week) regimen underwent at least one RBC transfusion, while less than 4% of patients on the same 3-week regimen needed transfusions (P = .0002). This dose-dense schedule was accomplished by initiating G-CSF 72 hours after the administration of chemotherapy and continuing the G-CSF daily dosing until day 10. However, in the same trial, seven doses of filgrastim for 12 chemotherapy treatments of sequential dose dense (every 2 weeks) doxorubicin (A) followed by paclitaxel (T) followed by C, totaling 84, resulted in a nonsignificant increase in grade 2 or worse anemia (11%) when this schedule was compared with sequential A followed by T followed by C (3%) or concomitant AC followed by T (8%) administered every 3 weeks.23 By contrast, a significant increase in grade 2 or worse anemia was observed only in the concomitant AC followed by T dose-dense arm (23%). In our study, the greatest number of filgrastim doses was seven injections for four every 3-week courses, totaling 28. At 120 mg/m2 the dose of E is higher than the conventional adjuvant dose range of 90 mg/m2 to 100 mg/m2, even though escalating the dose above 90 mg/m2 might lead to an improved outcome.24-26 The relatively high-dose E (120 mg/m2) concomitantly administered with C could in part explain the high incidence of grade 2 or worse anemia (26.2%) even in non–G-CSF arm. Our findings show that filgrastim alone is not the causative factor producing anemia, but contributes in increasing it when combined with a myelosuppressive agent administered at higher than the conventional dosage and coupled with another antineoplastic drug (C). As mentioned above, the rate of dose delays was significantly greater in patients not receiving filgrastim. Thus, the consequent increase of dose density in the filgrastim groups could be an additional factor in producing higher grade anemia. However, no significant difference in the mean Hb value was registered comparing patients who delayed chemotherapy with those who did not (Table 5). These findings further confirm the possible role of G-CSF in worsening anemia in these patients. To our knowledge, no study so far has reported on the possible correlation between G-CSF and anemia in early breast cancer patients treated with EC. Filgrastim use may cause myalgias, arthralgias, the inconvenience of a certain number of injections per course, and, according to our previous results,4 neutrophilia with eventual worsening of anemia. A cost-effective analysis would be useful to balance the increased use of CSF and recombinant human erythropoietin associated with dose-dense therapy. The aim of using G-CSF in dose-dense schedules is to shorten treatment intervals, increasing the cytotoxicity of chemotherapy and improving breast cancer specific outcomes. Because it is not necessary to repeat a chemotherapy cycle every 2 weeks without reaching leukocytes values much higher than normal ones, it is possible that 4 every-other-day G-CSF injections starting from day 6 can be sufficient to achieve these goals, eventually reducing expenses, the severity of anemia, and other adverse effects, as reported in an our previous experience in advanced gastric cancer.27 Similarly, the prophylactic administration of two filgrastim injections on days 8 and 12 after EC (or regimens with similar neutrophil curves) seems to be an adequate support.4 Our findings raise some concerns about the appropriate use of G-CSF during moderately intensive chemotherapy. We showed that anemia tended to progressively worsen during chemotherapy with increasing G-CSF dosage, and that the Hb decrease was minimal with the shortened G-CSF schedule. These findings could be taken into account in the refinement of a prediction model28 to identify patients at risk of developing anemia during adjuvant chemotherapy for breast cancer.
The authors indicate no potential conflicts of interest.
Authors' disclosures of potential conflicts of interest and author contributions are found at the end of this article.
1. van Os R, Robinson S, Sheridan T, et al: Granulocyte colony-stimulating factor enhances bone marrow stem cell damage caused by repeated administration of cytotoxic agents. Blood 92:1950-1956, 1998 2. Scinto AF, Cercato MC, Botti C, et al: Phase II trial of high-dose epirubicin and cyclophosphamide in advanced breast cancer. Eur J Cancer 30A:1285-1288, 1994[CrossRef] 3. Papaldo P, Lopez M, Cortesi E, et al: Addition of either lonidamine or granulocyte colony-stimulating factor does not improve survival in early breast cancer patients treated with high-dose epirubicin and cyclophosphamide. J Clin Oncol 21:3462-3468, 2003 4. Papaldo P, Lopez M, Marolla P, et al: The impact of five prophylactic filgrastim schedules on hematologic toxicity in early breast cancer patients treated with epirubicin and cyclophosphamide. J Clin Oncol 23:6908-6918, 2005 5. De Vita VT: Cancer: Principles and Practice of Oncology (6th ed). Baltimore, MD, Lippincott, Williams, & Wilkins, 2001, pp 426 6. Van Zant G, Goldwasser E: The simultaneous effects of erythropoietin and colony stimulating factor on bone marrow cells. Science 198:733, 1977 7. Souza LM, Boone TC, Gabrilove J, et al: Recombinant human granulocyte colony-stimulating factor: Effects on normal and leukemic myeloid cells. Science 232:61-65, 1986 8. Li CL, Johnson GR: Stimulation of multipotential, erythroid and other murine haematopoietic progenitor cells by adherent cell lines in the absence of detectable multi-CSF (IL3). Nature 316:633-636, 1985[CrossRef][Medline] 9. Metcalf D, Nicola NA: Proliferative effects of purified granulocyte colony-stimulating factor (G-CSF) on normal mouse hemopoietic cells. J Cell Physiol 116:198-206, 1983[CrossRef][Medline] 10. de Haan G, Loeffler M, Nijhof W: Long-term recombinant human granulocyte colony-stimulating factor (rhG-CSF) treatment severely depresses murine marrow erythropoiesis without causing an anemia. Exp Hematol 20:600-604, 1992[Medline] 11. Bungart B, Loeffler M, Goris H, et al: Differential effects of recombinant human colony stimulating factor (rh G-CSF) on stem cells in marrow, spleen and peripheral blood in mice. Br J Haematol 76:174-179, 1990[Medline] 12. Cronkite EP, Burlington H, Shimosaka A, et al: Anemia induced in splenectomized mice by administration of rhG-CSF. Exp Hematol 21:319-325, 1993[Medline] 13. de Haan G, Engel C, Dontje B, et al: Mutual inhibition of murine erythropoiesis and granulopoiesis during combined erythropoietin, granulocyte colony-stimulating factor, and stem cell factor administration: In vivo interactions and dose-response surfaces. Blood 84:4157-4163, 1994 14. van Os R, Robinson S, Sheridan T, et al: Granulocyte-colony stimulating factor impedes recovery from damage caused by cytotoxic agents through increased differentiation at the expense of self-renewal. Stem Cells 18:120-127, 2000 15. Park K, Im T, Sasaki A, et al: Positive effect of granulocyte-colony stimulating factor on erythropoiesis in humans. Osaka City Med J 37:123-132, 1991[Medline] 16. Miles SA, Mitsuyasu RT, Lee K, et al: Recombinant human granulocyte colony-stimulating factor increases circulating burst forming unit-erythron and red blood cell production in patients with severe human immunodeficiency virus infection. Blood 75:2137-2142, 1990 17. Kojima S, Fukuda M, Miyajima Y, et al: Treatment of aplastic anemia in children with recombinant human granulocyte colony-stimulating factor. Blood 77:937-941, 1991 18. Calhoun DA, Li Y, Christensen RD: Effect of recombinant granulocyte colony-stimulating factor on erythropoiesis in the human fetus and neonate. Pediatr Res 40:872-875, 1996[Medline] 19. Gabrilove JL, Jakubowski A, Fain K, et al: Phase I study of granulocyte colony-stimulating factor in patients with transitional cell carcinoma of the urothelium. J Clin Invest 82:1454-1461, 1988[Medline] 20. Fukutani H, Ogawa M, Horikoshi N, et al: Effect of recombinant human granulocyte colony stimulating factor (rhG-CSF) in patients receiving chemotherapy–phase I study. Gan To Kagaku Ryoho 16:2005-2012, 1989[Medline] 21. de Wit R, Verweij J, Bontenbal M, et al: Adverse effect on bone marrow protection of prechemotherapy granulocyte colony-stimulating factor support. J Natl Cancer Inst 88:1393-1398, 1996 22. Citron ML, Berry DA, Cirrincione C, et al: Randomized trial of dose-dense versus conventionally scheduled and sequential versus concurrent combination chemotherapy as postoperative adjuvant treatment of node-positive primary breast cancer: First report of Intergroup Trial C9741/Cancer and Leukemia Group B Trial 9741. J Clin Oncol 21:1431-1439, 2003 23. Citron ML, Berry DA, Cirrincione C, et al: Dose-dense (DD) AC followed by paclitaxel is associated with moderate, frequent anemia compared to sequential (S) and/or less DD treatment: Update by CALGB on Breast Cancer Intergroup Trial C9741 with ECOG, SWOG, & NCCTG. J Clin Oncol 23:33s, 2005 (abstr 620) 24. French Adjuvant Study Group: Benefit of a high-dose epirubicin regimen in adjuvant chemotherapy for node-positive breast cancer patients with poor prognostic factors: 5-year follow-up results of French Adjuvant Study group 05 randomized trial. J Clin Oncol 19:602-611, 2001 25. Piccart MJ, Di Leo A, Beauduin M, et al.: Phase III trial comparing two dose levels of epirubicin combined with cyclophosphamide with cyclophosphamide, methotrexate, and fluorouracil in node-positive breast cancer. J Clin Oncol 19:3103-3110, 2001 26. Levine MN, Pritchard KI, Bramwell VH, et al: Randomized trial comparing cyclophosphamide, epirubicin, and fluorouracil with cyclophosphamide, methotrexate, and fluorouracil in premenopausal women with node-positive breast cancer: Update of National Cancer Institute of Canada Clinical Trials Group Trial MA5. J Clin Oncol 23:5166-5170, 2005 27. Felici A, Carlini P, Ruggeri EM, et al: A feasibility study of a bi-weekly PELF regimen in advanced gastric cancer. Proc Am Soc Clin Oncol 22:349, 2003 (abstr 1402) 28. Dranitsaris G, Clemons M, Verma S, et al: Chemotherapy-induced anaemia during adjuvant treatment for breast cancer: Development of a prediction model. Lancet Oncol 6:856-863, 2005[CrossRef][Medline] Submitted June 2, 2005; accepted February 10, 2006. Related Editorial
This article has been cited by other articles:
|
|||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
|
|||||||||||
|
|||||||||||
|
|
Copyright © 2006 by the American Society of Clinical Oncology, Online ISSN: 1527-7755. Print ISSN: 0732-183X
|
TOP
ABSTRACT
INTRODUCTION
2 test. Curves of Hb levels were calculated by means per week. Comparisons between the estimates were conducted across various treatment arms and groups by the Student's t test. SPSS for Windows version 13 (SPSS Inc, Chicago, IL) was used for statistical computation. The statistical analysis was supplemented by a repeated measures analysis of variance with the time variable defined by the weekly assessments of Hb level. This analysis was necessary so that the potential effect of a time-by-group interaction could be statistically assessed. To adjust for multiple comparisons concerning the data pertaining to all of the G-CSF schedules, the cited P values were considered significant only if less than .0006 (Bonferroni adjustment). 
