Stull DE, Vernon MK, Legg JC, Viswanathan HN, Fairclough D, Revicki DA. Use of latent growth curve models for assessing the effects of darbepoetin alfa on hemoglobin and fatigue. Contemp Clin Trials. 2010 Mar;31(2):172-9. doi: 10.1016/j.cct.2009.12.006.

BACKGROUND: The relationship between darbepoetin alfa and fatigue in chemotherapy-induced anemia (CIA) patients is complex because of patients receiving transfusions and the mediating effect of hemoglobin. Latent growth models (LGMs) were used to examine simultaneously relationships among drug exposure, fatigue outcomes, covariates, and mediating factors.

METHODS: Data from four CIA studies (AMG 20010145: small cell lung cancer, n=547; AMG 980297: lung cancer, n=288; AMG 20000161: lymphoproliferative malignancies, n=339; AMG 20030232: non-myeloid malignancies, n=320) were analyzed separately. Patients reported fatigue using the FACT-Fatigue. The effect of darbepoetin alfa on FACT-F changes mediated through hemoglobin changes was examined with LGMs controlling for transfusions, age, sex, baseline ECOG performance status, and health status (EQ-5D VAS). Model fit was assessed using multiple indices including the comparative fit index (CFI).

RESULTS: Darbepoetin alfa increased hemoglobin levels which were associated with decreases in fatigue. Increases in hemoglobin were statistically significantly (p<0.05) related to decreases in fatigue in the studies (AMG 20030145: beta=0.28; AMG 980297: beta=0.46; AMG 20000161: beta=0.59; and AMG 20030232: beta=0.39). Darbepoetin alfa statistically significantly increased hemoglobin (AMG 20010145:beta=0.50, AMG 980297:beta=0.53, AMG 20000161:beta=0.47, and AMG 20030232:beta=0.30) while controlling for covariates. Model fit was acceptable (CFI> or =0.89) in all studies.

CONCLUSIONS: Results indicate LGMs may be a valuable statistical method for modeling complex relationships among clinical and patient reported outcomes. A statistically significant effect of darbepoetin alfa on fatigue change through hemoglobin change occurred across four studies, after modeling the effects of transfusions, age, sex, EQ-5D VAS and ECOG. Copyright 2010 Elsevier Inc. All rights reserved.

Share on: