Correction: evaluating predictive pharmacogenetic signatures of adverse events in colorectal cancer patients treated with fluoropyrimidines.

[This corrects the article DOI: 10.1371/journal.pone.0078053.].

How folate metabolism affects colorectal cancer development and treatment; a story of heterogeneity and pleiotropy.

Folate was identified as an essential micronutrient early in the twentieth century and anti-folate chemotherapy such as 5-fluorouracil (5-FU) has been central to the medical management of solid tumours including colorectal cancer for more than five decades. In the intervening years, evidence has been gathered which shows that folate deficiency leads to many human diseases throughout the life-course. However, we still do not know all of the mechanisms behind functional folate deficiency, or indeed its rescue through supplementation with natural and particularly synthetic folates. There is growing evidence that one adverse effect of folic acid fortification programmes is an increased risk of colorectal cancer within populations. The complexity of folate-dependent, one-carbon metabolism and the heterogeneity that exists between individuals with respect to the enzymes involved in the anabolic pathways, and the catabolism of 5-FU, are explored in this review. The enzyme products of some genes such as MTHFR exert multiple and perhaps unrelated effects on many phenotypes, including cancer development. We describe this pleiotropy and the common genetic variants that affect folate metabolism; and discuss some of the studies that have investigated their potential as predictive biomarkers.

Evaluating predictive pharmacogenetic signatures of adverse events in colorectal cancer patients treated with fluoropyrimidines.

The potential clinical utility of genetic markers associated with response to fluoropyrimidine treatment in colorectal cancer patients remains controversial despite extensive study. Our aim was to test the clinical validity of both novel and previously identified markers of adverse events in a broad clinical setting. We have conducted an observational pharmacogenetic study of early adverse events in a cohort study of 254 colorectal cancer patients treated with 5-fluorouracil or capecitabine. Sixteen variants of nine key folate (pharmacodynamic) and drug metabolising (pharmacokinetic) enzymes have been analysed as individual markers and/or signatures of markers. We found a significant association between TYMP S471L (rs11479) and early dose modifications and/or severe adverse events (adjusted OR = 2.02 [1.03; 4.00], p = 0.042, adjusted OR = 2.70 [1.23; 5.92], p = 0.01 respectively). There was also a significant association between these phenotypes and a signature of DPYD mutations (Adjusted OR = 3.96 [1.17; 13.33], p = 0.03, adjusted OR = 6.76 [1.99; 22.96], p = 0.002 respectively). We did not identify any significant associations between the individual candidate pharmacodynamic markers and toxicity. If a predictive test for early adverse events analysed the TYMP and DPYD variants as a signature, the sensitivity would be 45.5 %, with a positive predictive value of just 33.9 % and thus poor clinical validity. Most studies to date have been under-powered to consider multiple pharmacokinetic and pharmacodynamic variants simultaneously but this and similar individualised data sets could be pooled in meta-analyses to resolve uncertainties about the potential clinical utility of these markers.

Functional polymorphisms of folate metabolism and response to chemotherapy for colorectal cancer, a systematic review and meta-analysis.

OBJECTIVES: Many primary studies have considered the association of polymorphisms of folate metabolism and response to 5-fluorouracil (5-FU) and capecitabine in patients with colorectal cancer. The conclusions from these studies have been conflicting and few have considered large cohorts of patients. Therefore, we have completed a systematic review and meta-analyses to summarize some of the findings to date. We conducted searches for any studies that had addressed the prognostic value of genotype analysis of thymidylate synthetase (TYMS), Methylenetetrahydrofolate reductase (MTHFR) and dihydrofolate reductase (DHFR). METHODS: We collected data on the study designs, and completed meta-analyses to pool congruent data about treatment effect. A narrative summary is presented for 39 studies that describe three TYMS genotypes and two MTHFR genotypes associated with response to 5-FU-based chemotherapy. RESULTS: Data were synthesized from up to 2402 patients for the most commonly studied markers TYMS 5' UTR repeat polymorphism (rs45445694) and MTHFR 677 C>T (rs1801133). We found that the TYMS genotype associated with the lowest protein expression (2R/2R) was significantly associated with improved clinical benefit; the pooled risk ratio was relative risk=1.36 [1.11, 1.65]; P=0.003. Moreover, the same trend was observed for adverse effects; the pooled risk ratio was 2.04 [1.42, 2.95]; P=0.0001. CONCLUSION: There is a small but statistically significant association between treatment effect (both intended effects and adverse events) and a TYMS genotype associated with low protein expression; however, the effect size is small and therefore indicates limited clinical utility.

Estimation of the 5-methyltetrahydrofolate apparent volume of distribution in humans.

The fractional absorption of a stable isotope-labeled folate dose can be estimated from the subsequent short-term temporal changes in the concentration of labeled L-5-methyltetrahydrofolate (L-5-methyl-THF) in plasma using mathematical modeling. However, the model is dependent on the use of an accurate value for the apparent volume of distribution of L-5-methyl-THF. Previous studies that estimated the apparent volume of distribution of L-5-methyl-THF used large (nonphysiological) doses of unlabeled folates that are not found to any great extent in the circulatory system. The current study estimates the apparent volume of distribution at steady state in 16 healthy humans aged 18-65 y after an i.v. dose (440 nmol) of a stable isotope-labeled version of the naturally circulating plasma folate, L-5-methyl-THF. Blood was collected from 2 min to 2 h postinjection and plasma assayed by specific and sensitive liquid chromatography-tandem MS. The apparent volume of distribution for L-5-methyl-THF was 32.0 ± 11.6 L (mean ± SD; 392 ± 110 mL/kg bodyweight). There was a positive association with volunteer body weight (r = 0.64; P = 0.010), which allowed a simple linear equation to be developed relating apparent volume of distribution to body weight. This has important implications for predicting apparent absorption of labeled folates in future bioavailability studies.

Do novel risk biomarkers reflect the severity of peripheral arterial disease?

The association between novel atherosclerotic risk biomarkers and severity of peripheral arterial disease (PAD) was assessed. Patients (n = 133) with PAD were recruited. Established risk biomarkers including low- and high-density cholesterol, triglycerides, and blood pressure were measured. Novel risk biomarkers including plasma C-reactive protein, von Willebrand factor (vWF), interleukin 6, red cell folate (RCF), vitamin B12, total homocysteine (tHcy), and Hcy genotypes were also determined. The severity of PAD was evaluated, using ankle-brachial pressure index (ABPI), brachial-knee, and brachial-ankle pulse wave velocity (bk- and ba-PWV). Plasma tHcy and systolic blood pressure had a positive independent correlation with bk-PWV (ß = +0.56, P = .02 and ß = +0.38, P < .001, respectively). Red cell folate had an independent inverse correlation with bk-PWV (ß = -0.01, P = .01). Systolic blood pressure showed an independent positive correlation with ba-PWV only after adjustment for other risk biomarkers (ß = +0.1, P = .04). Novel markers, plasma tHcy, and RCF levels correlated with the severity of PAD.

Genetic selection? A study of individual variation in the enzymes of folate metabolism.

BACKGROUND: Genetic variation in folate metabolism has been associated with survival in utero, the success of in vitro fertilisation, multiple pathologies and longevity. METHODS: We have looked at the prevalence of genetic variants of the enzymes MTHFR and TYMS in 2,898 DNA samples derived from five cohorts collected in the United Kingdom. The simultaneous analysis of genetic variants of the MTHFR and TYMS loci was carried out to investigate a putative gene-gene interaction that was first observed in an elderly male population from Norfolk. RESULTS: We have made a consistent observation in five population cohorts; the proportion of individuals who are homozygous for the 2R allele of the 5'UTR TYMS polymorphism is less in individuals who are homozygous for the T allele of MTHFR 677 than in individuals homozygous for the C allele of MTHFR 677 (p = 0.02). CONCLUSIONS: These data may suggest a gene-gene interaction and could be evidence of genetic selection, with some pregnancies more or less viable as a consequence of genetic variation. If these genetic phenomena affect the way folate is handled at the cellular level in utero it is possible that maternal folic acid intake may over-ride such genetic selection.

Peripheral arterial disease and methylenetetrahydrofolate reductase (MTHFR) C677T mutations: A case-control study and meta-analysis.

OBJECTIVE: Hyperhomocysteinaemia is associated with peripheral arterial disease (PAD). There are inter-individual variations in the metabolism of homocysteine because of genetic polymorphisms. This study analyzed the role of one polymorphism that is associated with raised homocysteine, as a risk factor for PAD. METHODS: This study considered the association of methylenetetrahydrofolate reductase (MTHFR) C677T polymorphisms with the incidence of PAD by performing a case-control study and a cross sectional study of homocysteine levels. We recruited 133 patients with PAD in Norfolk and compared the MTHFR allele distribution with 457 healthy individuals. We also carried out a meta-analysis to place our data within the context of other published studies. We searched Medline, Embase, and Cochrane databases up to March 2008 for any studies on the association between MTHFR C677T polymorphism and PAD. RESULTS: The MTHFR C677T allele frequencies in the cases and controls were 0.37 and 0.33, and the odds ratios for the association of the 677 T allele or TT genotype with PAD were 1.18 (95% Confidence Interval [CI] 0.89, 1.58) and 1.99 (95% CI 1.09, 3.63). Homozygotes for the MTHFR C677T mutation had higher concentrations of plasma total homocysteine, odds ratio 2.82 (95% CI 1.03, 7.77) compared to homozygotes for the MTHFR 677 CC genotype. Twelve of 72 articles retrieved from the database search reported the prevalence of mutations in PAD patients. A meta-analysis of 9 appropriate studies, including our own, showed that being homozygous for the C677T allele was associated with an increased risk of PAD, pooled odds ratio 1.36 (95% CI 1.09, 1.68). CONCLUSION: We have found a strong association between raised homocysteine, the TT genotype, and PAD.

Hepatocellular carcinoma and the penetrance of HFE C282Y mutations: a cross sectional study.

BACKGROUND: Although most patients with hereditary haemochromatosis have HFE C282Y mutations, the lifetime risk to HFE C282Y homozygotes of developing fatal diseases such as hepatocellular carcinoma is uncertain. We have carried out a cross-sectional study to determine the proportion of diagnosed hepatocellular carcinoma patients who are homozygous for the HFE C282Y mutation; and to estimate the penetrance of this genotype with respect to hepatocellular carcinoma in East Anglia. METHODS: Tissue biopsies were analysed from 144 cases of hepatocellular carcinoma for HFE C282Y mutations; the data produced were compared with the frequency of HFE mutations in a large sample of the local population. Data were also retrieved from the East Anglian Cancer Intelligence Unit to determine the annual incidence of hepatocellular carcinoma; and from appropriate life tables. RESULTS: Eight out of 144 of the cases were homozygous for the HFE C282Y mutation, all 8 cases were male. 6 of these 8 cases had a previous diagnosis of hereditary haemochromatosis. Male HFE C282Y homozygotes were more likely to be diagnosed with hepatocellular carcinoma (odds ratio [OR] = 14, 95% confidence interval [CI] = 5-37). For this population, we estimate that the penetrance of the HFE C282Y homozygous genotype, with respect to hepatocellular carcinoma, was between 1.31 % and 2.1% for males and was zero for females. CONCLUSION: In this population, we found that only a very small proportion of homozygotes for the HFE C282Y mutation developed hepatocellular carcinoma. However, individuals with this genotype have a significantly increased risk of this rare disease relative to those who do not carry the mutations.

Iron absorption in male C282Y heterozygotes.

BACKGROUND: The suggestion that carriers of the HFE C282Y mutation absorb nonheme iron more efficiently than do carriers of the wild type has public health implications for countries where the C282Y mutation is common and foods are fortified with iron. OBJECTIVE: We investigated the effect of C282Y heterozygosity on nonheme-iron absorption from a diet high in bioavailable iron and from iron-fortified cereals. DESIGN: The subjects were recruited from a parallel study investigating the relation between HFE mutations, habitual diet, and iron status. Iron absorption was measured in 15 wild-type carriers and 15 C282Y heterozygotes aged >/=40 y. Each subject consumed 3 meals of high iron bioavailability (labeled with Fe-57) for 2 d and 2 meals with fortified cereal products (labeled with Fe-54) for the next 3 d. Iron absorption was measured from isotope incorporation into red blood cells 14 d after the last labeled meal and was corrected for utilization of absorbed iron by means of an intravenous infusion of Fe-58. RESULTS: Absorption of Fe-57 with the high-iron-bioavailability diet was 6.8 +/- 6.8% (0.6 +/- 0.6 mg/d) in the wild-type carriers and 7.6 +/- 3.2% (0.7 +/- 0.3 mg/d) in the C282Y heterozygotes. Absorption of Fe-54 with cereal products was 4.9 +/- 2.0% (0.7 +/- 0.3 mg/d) in the wild-type carriers and 5.3 +/- 1.3% (0.8 +/- 0.2 mg/d) in the C282Y heterozygotes. CONCLUSIONS: There was no overall significant difference between C282Y heterozygotes and wild-type men in iron absorption from either dietary nonheme iron or fortified cereal products.

Paraoxonase-1 (PON-1) genotype and activity and in vivo oxidized plasma low-density lipoprotein in Type II diabetes.

The HDL (high-density lipoprotein)-associated enzyme PON (paraoxonase)-1 protects LDL (low-density lipoprotein) from oxidative modification in vitro, although it is unknown if this anti-atherogenic action occurs in vivo. In a cross-sectional study of 58 Type II diabetic subjects and 50 controls, we examined the fasting plasma LDL basal conjugated diene concentration [a direct measurement of circulating oxLDL (oxidatively modified LDL)], lipoprotein particle size by NMR spectroscopy, PON-1 polymorphisms (coding region polymorphisms Q192R and L55M, and gene promoter polymorphisms -108C/T and -162G/A), PON activity (with paraoxon or phenyl acetate as the substrates) and dietary antioxidant intake. Plasma oxLDL concentrations were higher in Type II diabetic patients (males, P = 0.048; females, P = 0.009) and unrelated to NMR lipoprotein size, PON-1 polymorphisms or PON activity (with paraoxon as the substrate) in any group. In men with Type II diabetes, however, there was a direct relationship between oxLDL concentrations and PON activity (with phenyl acetate as the substrate; r = 0.611, P = 0.0001) and an atherogenic NMR lipid profile in those who were PON-1 55LL homozygotes. Circulating oxLDL concentrations in vivo were unrelated to PON-1 genotypes or activity, except in male Type II diabetics where there was a direct association between PON activity (with phenyl acetate as the substrate) and oxLDL levels. These in vivo data contrast with in vitro data, and may be due to confounding by dietary fat intake. Male Type II diabetic subjects with PON-1 55LL homozygosity have an atherogenic NMR lipid profile independent of LDL oxidation. These data do not support an in vivo action of PON on LDL oxidation.

HFE mutations in the elderly.

Most individuals diagnosed with hereditary hemochromatosis have mutations in both copies of the HFE gene, with such mutations being common in populations of north European origin. The number of individuals currently diagnosed and treated for hemochromatosis is small relative to the number carrying two HFE mutations. Studies searching for undiagnosed hemochromatosis cases among disease cohorts have generally failed to find the number of cases that would be expected if disease were the commonest outcome for individuals with two C282Y HFE mutations. Our aim was to test the hypothesis that individuals with two HFE mutations would be under-represented in an elderly population because many would have died from disease caused by hemochromatosis before they reached old age. This is a cross-sectional study of elderly patients referred for full blood counts at the Norfolk and Norwich University Hospital. We screened blood samples from 1,000 elderly men (aged 85 and over) and women (aged 89 and over) for the C282Y, H63D, and S65C mutations of the HFE gene. We also analyzed any recent laboratory data relevant to signs of hemochromatosis. None of the ten possible genotypes was significantly under- or over-represented compared to the expected frequency calculated from the Hardy-Weinberg equation. Four C282Y homozygotes were found. There were few significant differences in the laboratory findings between the genotypes. Our data suggest that most people with HFE mutations survive to old age and do not suffer from signs of iron overload and hemochromatosis.