A French prospective pilot study for identifying dihydropyrimidine dehydrogenase (DPD) deficiency in breast cancer patients (pts) receiving capecitabine (cap)

Background: For fluoropyrimidines, and especially cap, Health Authorities point out that DPD deficiency confers a significant risk of major toxicity (tox). Identification of at-risk pts is thus relevant. This multicentric prospective study of the French GPCO group (Groupe de Pharmacologie Clinique Oncologique, Unicancer) evaluated the sensitivity, specificity and predictive values of DPD phenotyping and genotyping for predicting severe cap-related tox in metastatic breast cancer pts. Methods: 303 pts were included (15 institutions), 88% received cap as monotherapy, 28% were treated as first line (mean dose at 1st cycle 1957 mg/m2/d). Pre-treatment dihydrouracil (UH2) and uracil (U) plasma concentrations were measured in 286 pts (HPLC assay). DPD genotyping (IVS14+1G>A, 2846A>T, 1679T>G, 464T>A) was done on 281 pts. Severe tox (G3-4 CTCAE v3 criteria) was measured over cycles 1-2. Results: Grade 3-4 tox (diarrhea, vomiting, hematoxicity, hand-foot syndrome) was observed in 19.6% of pts (one toxic death). A marked trend for higher U (median 12.7 vs 10.2 ng/ml, p=0.014) and UH2 (median 110 vs 93 ng/ml, p=0.011) concentrations was observed in pts developing severe tox vs those who didn't. However, ROC curves showed that these differences were too small for use as reliable tox predictors. The distribution of UH2/U ratio was similar between pts with or without tox (median 9.1 vs 9.6, respectively, p=0.80). The patient with toxic death had a UH2/U ratio of 6.5 and U concentration of 17 ng/ml. Among the 7 pts with a DPD mutation (3 pts IVS14+1, 3 pts 2846A>T, one 1679T>G, all heterozygous), 5 developed severe tox (including toxic death, 2846A>T), one did not, and the last one was not documented. Relative risk for developing severe tox was 4.60 in mut pts vs wt pts (95%CI 2.95-7.16, p=0.001); positive and negative predictive values were 83.3% and 81.9%, respectively; specificity was 99.5% and sensitivity was 9.8%. Conclusions: These data point out that breast cancer pts harbouring a DPD variant allele are candidate to develop severe, up to lethal, cap-related tox. In contrast, pre-treatment UH2/U ratio and U measurements are not reliable predictors of cap tox. Clinical trial information: Eudract 2008-004136-20.