Naslov
PREDICTION OF RESPONSE TO CHLORAMBUCIL IN CHRONIC LYMPHOCYTIC LEUKEMIA (CLL). MODEL DERIVED FROM LOGISTIC REGRESSION ANALYSIS ON 291 PATIENTS IN IGCI CLL TRIALS

Autori
Jakšić, Ozren ; Brugiatelli, M. ; Iacopino, P. ; Morabito, F. ; Planinc-Peraica, Ana ; Minigo, Hrvoje ; Lutz, Dieter ; Jakšić, Branimir

Vrsta, podvrsta i kategorija rada
Sažeci sa skupova, sažetak, znanstveni

Izvornik
British Journal of Haematology / - , 1998

Skup
ISH-EHA Combined Haematology Congress

Mjesto i datum
Amsterdam, Nizozemska, 04.07.1998. - 08.07.1998

Vrsta sudjelovanja
Poster

Vrsta recenzije
Međunarodna recenzija

Sažetak
To develop a model for prediction of response, a stepwise logistic regression analysis and discriminant quadrant analysis were performed on 291 CLL patients treated in IGCI CLL trials (Jaksic et al, Cancer 79: 2107-14, 1997) with standard-dose (SD) or high-dose (HD) chlorambucil (CLB). Response was classified complete (CR) if all tumor parameters decreased below diagnostic threshold, partial (PR) if TTM-size decreased for more then 50% of pre-treatment value and all other patients were considered non-responders (NR). Age, Sex, Rai stages, Binet Stages, TTM-size, TTM-distribution, Bone marrow failure, and Dose level of therapy were all entered in analysis at step 0. Backward selection excluded Age, Sex, Binet stages and Bone marrow failure, while Dose level of therapy, Rai stages; TTM-size and TTM-distribution remained in the model (with respective p value of 0.000; 0.0246; 0.0412 and 0.128). Logistic model (with constant of 3.8 and coefficients for Rai stages/0 = 0; I = 2.4; II = 2.2; III = 2.5 and IV = 2.9/, TTM-size: = .0.05, TTM-lymphoma like distribution = 0.5 and Dose level/standard/of therapy = 1.9) was found to be efficient in its prediction of positive and negative responses in over 70% of cases. Thus, a patient in Rai stage II, TTM-size 10, TTM-leukemic distribution has predicted probability of complete response 31%, if treated with SD-CLB, and 75% if treated with HD-CLB. Those results were also corroborated by discriminant quadrant analysis. We conclude that simple clinical parameters can be efficiently used for reliable and reproducible prediction of response.

Izvorni jezik
Engleski

Znanstvena područja
Kliničke medicinske znanosti

POVEZANOST RADA