test About Medicine OnLine Medicine OnLine Home Page Cancer Libraries DoseCalc Online Oncology News
Cancer Forums Medline Search Cancer Links Glossary



Conrad Notes IndexspacerASCO Index

Conrad Notes
a timely medical meeting newsletter
PREDICTING HEMOGLOBIN RESPONSE TO EPOETIN
D. Henry, MD, Graduate Hospital, Philadelphia, PA, proposes using a battery of three factors to predict whether or not an anemic cancer patient receiving chemotherapy will respond to epoetin alpha (EA). The likelihood of an increase in hemoglobin of 2 g/dL or more depends on the hemoglobin change at 4 weeks, blood transfusion status, and observed/predicted EA ratio. Data on 752 anemic cancer patients led to developing this model.

Introduction EA can increase hemoglobin and/or decrease the need for blood transfusion in 53.4% of anemic patients. Several investigators tried to predict who would respond before or soon after treatment with EA (Ludwig, Cazzola, Osterborg, Henry, and Glaspy).
Study Protocol A pool of 2030 cancer patients treated with EA served to provide a random sample of 752 records for analysis. Each patient record gave information on hemoglobin change after 4 weeks' treatment, transfusion dependence, and baseline pretreatment epoetin level. Beguin's method (Brit J Hematol 1992;82:648) allowed calculating the predicted epoetin levels from EA administration. The equation is log predicted epoietin level = 3.420 - (0.056 x % hematocrit). Henry defined a positive response to EA therapy as an increase in hemoglobin of at least 2 g/dL.
Results The following calculations support Henry's proposal:

Response Predictor Percent Response
None 53.4
Hemoglobin Change at 4 Weeks
<1g/dL 29.5
>/=1g/dL 75.1
Transfusion Within Last 4 Weeks
Yes 30.0
No 62.0
Observed/Predicted (O/P) Epoetin Ratio
>/=0.825 39.0
<0.825 61.0
All 3 Combined: 85.0
Hemoglobin Increase >/=1g/dL
Transfusion Independence
O/P Epoetin Ratio <0.825

The likelihood of an increase in hemoglobin of 2 g/dL or more from EA may be predicted at the 85% level by using all three measures.

Comments The investigator's proposed method for predicting response to EA seems simple and direct but requires hemoglobin and blood transfusion data as late as 4 weeks. An earlier decision point would enhance the value of the method. Clinical application of the procedure awaits prospective data collection.

For professional correspondence, please contact Dr. Henry at: dhhenry@juno.com

Eugene A. Conrad

Presented at the Annual Meeting of the American Society of Clinical Oncology (ASCO) on May 17-20, 1997
CONRAD NOTES, © 1997 All Rights Reserved
Eugene A. Conrad, PhD, MPH / ISSN 1078 / June 1997

Return To Top Of PagespacerConrad Notes IndexspacerASCO Index

test About Medicine OnLine Medicine OnLine Home Page Cancer Libraries DoseCalc Online Oncology News
Cancer Forums Medline Search Cancer Links Glossary