View PDF Version - RePub - Erasmus Universiteit Rotterdam
View PDF Version - RePub - Erasmus Universiteit Rotterdam
View PDF Version - RePub - Erasmus Universiteit Rotterdam
Create successful ePaper yourself
Turn your PDF publications into a flip-book with our unique Google optimized e-Paper software.
Statistical analysis<br />
Dynamic prediction of response to PEG-IFN 65<br />
In a previous study baseline factors infl uencing the SR-rate were described in detail<br />
and a prediction model (PEG-IFN HBV Treatment Index) was developed which provides<br />
a subject specifi c prediction of SR. 9 Baseline factors associated with SR were: HBVgenotype,<br />
age, gender, baseline HBV DNA (copies/ml, log10 ), ALT (loge ) and previous<br />
treatment with IFN.<br />
In the current study the PEG-IFN HBV Treatment Index was fi rst calculated for each<br />
subject and was used throughout the analysis as an offset, i.e. a subject specifi c starting<br />
value with the corresponding regression coeffi cient set to 1. The baseline prediction<br />
of SR, Pbaseline , was hereafter updated with data on HBV DNA and ALT during therapy<br />
applying logistic regression analysis techniques. For comparison purposes two model<br />
approaches were considered: the fi rst was updating the model at each visit with the<br />
new information, resulting in 8 models; one for each visit (week 4, 8, 12, 16, 20, 24, 28<br />
and 32), the second an overall generalized estimating equations model11 using information<br />
of all visits and adding the visit time as a continuous factor. The latter model<br />
approach allows repeated measurement data and hence reduces the 8 models of the<br />
fi rst approach into one overall model. Technically, each visit per patient is treated as<br />
an observation and the model then corrects for the fact that each patient contributes<br />
with 8 visits. This model is referred to as the dynamic logistic regression model and is<br />
sometimes called a pooled logistic regression model. 12-13 The treatment duration (visit<br />
time) was added as a linear variable to the model, a restricted cubic spline was used to<br />
check the linearity assumption. The effect of the crude HBV-DNA (log10 ) as well as the<br />
effect of HBV DNA log10-decline compared to baseline (= HBV DNAlog10 at time t – HBV<br />
DNAlog10 at baseline) was studied. ALT was entered in the models as measured and also<br />
transformed logarithmically. Interactions between HBV DNA and ALT with treatment<br />
duration (visit time) and with HBV-genotype were considered. Changes in HBV DNA and<br />
ALT measurements in prior visits were also studied. Discrimination between the different<br />
models was quantifi ed by the c-statistics, which is the area under the receiver operating<br />
curve. The best model-fi t was assessed comparing these (the higher the better) and the<br />
Akaike’s Information criteria (AIC) or the quasi-likelihood information criteria (QIC) for<br />
the generalized estimating equations method (the lower the better). Cross validation,<br />
leaving one out, was performed to establish overall performances of the models.<br />
Finally a cut-off value of HBV DNA during treatment was sought to fi nd a clinical useful<br />
guiding rule for (dis)continuation of therapy. The optimal cut-off point was established in<br />
a multivariable setting14 including the baseline PEG-IFN HBV Treatment Index: explanatory<br />
plots were evaluated, followed by the maximum chi-square approach. Subsequently<br />
the cut-off point search was repeated in 500 bootstrap samples for validation of the