Validation and evaluation of a gene set classifier for late radiation toxicity in prostate cancer patients

Most patients with prostate cancer can be cured by radiotherapy. Increasing the
radiation dose improves tumour control, but will give more normal tissue
complications. The availability of a test for late radiation toxicity with high
predictive power enables patient-tailored therapy. Recently we obtained
evidence that genetic factors are important for individual radiation toxicity.
Gene set analysis of the radiation-induced transcriptional response in ex vivo
cultured lymphocytes allowed discrimination between patients with and without
severe late toxicity after radiotherapy (Svensson, 2006). A prospective study
with a larger number of patients is required to validate this gene set
classifier. In addition, we will assess *-H2AX foci: This simple assay
determines the rate of disappearance of DNA-DSB associated *-H2AX foci after
ionizing radiation. Combination of the two assays will provide insight in
whether cellular or tissue responses determine late effects. Gene expression
profiles reflect differences in overall tissue response to ionizing radiation;
with the **H2AX test we wish to address a more specific cellular mechanism
involved in radiation response.

Prospective validation of a previously determined gene set expression
classifier that allows assessment of late radiation toxicity risk in irradiated
prostate cancer patients.

Following informed consent, 200 patients with newly diagnosed prostate cancer
who will have curative external beam radiotherapy will be entered in this
study. The following data will be collected:
* 50 ml blood sample for gene expression profiling (according to Svensson,
2006) and determination of gamma-H2AX foci [Kato 2006, 2007].
* Radiation toxicity will be scored using a modified RTOG/SOMA LENT score
before and at the end of radiotherapy, and 6, 12, and 24 months after
radiotherapy [Peeters, 2006; Heemsbergen, 2006],
* Baseline clinical data (a.o. age, tumour stage, histology, PSA, co-morbidity,
smoking),
* Treatment data: diagnostic surgery, hormonal treatment, radiotherapy (dose,
fractionation, and volume data, particularly of prostate, bladder and rectum),
* Tumour follow-up data (PSA, secondary treatment)
Multivariate statistical methods, particularly correcting for the large amount
of genetic data, will be used to validate the predictive value of the earlier
assessed gene set classifier.

-

The combination of clinical and genetic risk factors could yield a screening
test for treatment selection in prostate cancer patients, improving treatments
by reducing side effects. The patient will give a single blood sample (50 ml)
and will answer repeated questionnaires assessing side effects, but will
further not impose anyy burden to the patients.