To bleed or not to bleed. A prediction based on individual gene profiling combined with dose-volume histogram shapes in prostate cancer patients undergoing three-dimensional conformal radiation therapy.

Purpose: The main purpose of this work was to try to elucidate why, despite excellent rectal dose-volume histograms (DVHs), some patients treated for prostate cancer exhibit late rectal bleeding (LRB) and others with poor DVHs do not. Thirty-five genes involved in DNA repair/radiation response were analyzed in patients accrued in the AIROPROS 0101 trial, which investigated the correlation between LRB and dosimetric parameters.

Methods And Materials: Thirty patients undergoing conformal radiotherapy with prescription doses higher than 70 Gy (minimum follow-up, 48 months) were selected: 10 patients in the low-risk group (rectal DVH with the percent volume of rectum receiving more than 70 Gy [V70Gy] < 20% and the percent volume of rectum receiving more than 50 Gy [V50Gy] < 55%) with Grade 2 or Grade 3 (G2-G3) LRB, 10 patients in the high-risk group (V70Gy > 25% and V50Gy > 60%) with G2-G3 LRB, and 10 patients in the high-risk group with no toxicity. Quantitative reverse-transcriptase polymerase chain reaction was performed on RNA from lymphoblastoid cell lines obtained from Epstein-Barr virus-immortalized peripheral-blood mononucleated cells and on peripheral blood mononucleated cells. Interexpression levels were compared by using the Kruskal-Wallis test.

Results: Intergroup comparison showed many constitutive differences: nine genes were significantly down-regulated in the low-risk bleeder group vs. the high-risk bleeder and high-risk nonbleeder groups: AKR1B1 (p = 0.019), BAZ1B (p = 0.042), LSM7 (p = 0.0016), MRPL23 (p = 0.015), NUDT1 (p = 0.0031), PSMB4 (p = 0.079), PSMD1 (p = 0.062), SEC22L1 (p = 0.040), and UBB (p = 0.018). Four genes were significantly upregulated in the high-risk nonbleeder group than in the other groups: DDX17 (p = 0.048), DRAP1 (p = 0.0025), RAD23 (p = 0.015), and SRF (p = 0.024). For most of these genes, it was possible to establish a cut-off value that correctly classified most patients.

Conclusions: The predictive value of sensitivity and resistance to LRB of the genes identified by the study is promising and should be tested in a larger data set.