Exogenous and endogenous determinants of vitamin K status in cystic fibrosis

Cystic fibrosis (CF) patients are at high risk for vitamin K deficiency. The effects of vitamin K supplementation are very ambiguous. Therefore, we aimed to define the determinants of vitamin K deficiency in a large cohort of supplemented - 146 (86.9%) and non-supplemented - 22 (13.1%) CF patients. Vitamin K status was assessed using prothrombin inducted by vitamin K absence (PIVKA-II) and undercarboxylated osteocalcin (u-OC). The pathological PIVKA-II concentration (≥2 ng/ml) and abnormal percentage of osteocalcin (≥20%) were found in 72 (42.8%) and 60 (35.7%) subjects, respectively. We found that liver involvement, diabetes, and glucocorticoid therapy were potential risk factors for vitamin K deficiency. Pathological concentrations of PIVKA-II occurred more frequently in patients with pancreatic insufficiency and those who have two severe mutations in both alleles of the CFTR gene. Pathological percentage of u-OC was found more frequently in adult CF patients and those not receiving vitamin K. However, it seems that there are no good predictive factors of vitamin K deficiency in CF patients in everyday clinical care. Early vitamin K supplementation in CF patients seems to be warranted. It is impossible to clearly determine the supplementation dose. Therefore, constant monitoring of vitamin K status seems to be justified.

Scientific RepoRts | 5:12000 | DOi: 10.1038/srep12000 Although vitamin K supplementation in CF patients seems to be necessary, the dose required to prevent deficiency is unclear 10 . Evidence from randomized controlled trials on vitamin K supplementation for CF patients is currently weak and limited to two small trials (total of 32 participants). Both trials reported normal levels of undercarboxylated osteocalcin after one month of daily supplementation with 1 mg of vitamin K 9 . Siwamogsatham et al. found that despite nearly doubling fat-soluble vitamin dosage (including vitamin K) in CF patients with pancreatic insufficiency over a 5 year period, there was no parallel increase in blood concentrations of these vitamins 11 .
The aims of the present study were to assess the frequency of vitamin K deficiency in CF patients and to estimate exogenous and endogenous determinants of vitamin K status.

Results
Body weight and height two standard deviations below the mean values were found in 27 (16.1%) and 39 (23.2%) patients, respectively. Hypoalbuminemia was documented in 30 (17.9%) CF patients. Spirometry was performed in 133 (79.2%) patients older than 6. Proper lung function was observed in 46 (34.6%) patients, while significant impairment in spirometry was found in 36 (27.1%) patients. Pathological concentrations of the liver enzymes AST, ALT, and GGT were found in 29.2%, 29.2% and 20.2% of patients, respectively. Comprehensive clinical data in the study group have been presented in Table 1.
A description of the clinical parameters in the subgroups of CF patients with both normal and pathological PIVKA-II concentrations and percentages of u-OC have been presented in Tables 2 and 3.
Vitamin K deficiency defined as PIVKA-II ≥ 2 ng/ml was much more frequent in patients with pancreatic insufficiency, the genotype F508del/F508del or the severe mutations in both alleles of the CFTR gene, and those not receiving vitamin K (Table 2).
Vitamin K deficiency based on abnormal u-OC% was found more frequently in adults CF patients and those not receiving vitamin K (16< 76.2%> vs. 44< 30.8%> , p = 0.000036; Fisher-Freeman-Halton test) (Table 3) A multiple regression analysis showed that none of the independent variables were important for predicting PIVKA-II or u-OC status (independent variables in the regression model: age, Z-score for body weight and height, FEV1, albumin concentration, vitamin K dose mg/week, diabetes, liver disease, pancreatic insufficiency, Pseudomonas aeruginosa colonization, inhaled and oral permanent antibiotic therapy, intravenous and oral antibiotic therapy in the preceding three months, inhaled glucocorticoid therapy, CFTR mutation). In a stepwise multiple regression, the independent variables of liver disease,  Table 2. Clinical parameters in CF patients with normal and pathological PIVKA-II concentration. 1 Severe class of mutations on both alleles of the CFTR gene. 2 Patients not receiving vitamin K or receiving less than the recommended dosage for CF (0.3 mg/d).
inhaled glucocorticoid therapy, diabetes and Pseudomonas aeruginosa colonization were entered according to their statistical contribution in explaining the variance of PIVKA-II. Similarly, vitamin K dose, inhaled glucocorticoid therapy, and intravenous antibiotic therapy in the preceding three months were added to the regression equation according to their statistical contribution in explaining the variance of u-OC. The results of the stepwise multiple regression models have been summarized in Table 4.

Discussion
The main possible causes of vitamin K deficiency in CF patients include fat malabsorption, liver disease, chronic antibiotic treatment, bowel resection, and increased mucous accumulation in the small intestine.
Scientific RepoRts | 5:12000 | DOi: 10.1038/srep12000 Very often, CF patients do not have proper vitamin K status despite vitamin K supplementation 8,10,12 . Therefore, we aimed to define the exogenous and endogenous determinants of vitamin K deficiency in CF patients in a large cohort of supplemented and non-supplemented CF patients. Vitamin K deficiency was characterized in the past as prolonged prothrombin time (PT) and activated partial thromboplastin time (APTT), as well as low concentrations of vitamin K-dependent coagulation proteins such as factors II, VII, IX and X 13 . However, coagulopathy is an insensitive marker of vitamin K deficiency because it may appear when the amount of undercaboxylated coagulation factors  Table 3. Clinical parameters in CF patients with normal and pathological u-OC percentage. 1 Severe class of mutations on both alleles of the CFTR gene. 2 u-OC < 20% vs 20-50% n.s; u-OC < 20 vs. > 50% p < 0.05; u-OC 20-50% vs > 50% n.s. 3 Patients not receiving vitamin K or receiving less than the recommended dosage for CF (0.3 mg/d).
exceeds 50% 4 . PIVKA-II may be detectable before any changes in coagulation tests appear 13 . Therefore, this parameter seems to be a sensitive marker for early vitamin K deficiency 12,14 . In the present study, subjects with abnormal PIVKA-II levels had higher INR values compared to patients with a normal vitamin K status (p = 0.0178). However, only 4 (5.6%) out of the 72 CF patients with pathological PIVKA-II concentrations were found to have abnormal values of PT% (data not shown) and INR.
In the present study, pathological PIVKA-II concentrations were found in 38.4% of CF patients receiving and 72.7% of those not receiving or receiving less than the recommended dosage of vitamin K for CF. Based on the percentage of u-OC, an insufficiency (u-OC 20-50%) and a deficiency (u-OC > 50%) of vitamin K were observed in 13.3% and 17.5% vitamin K supplemented subjects, respectively. In turn, vitamin K insufficiency was observed in 14.3% and deficiency in 61.9% of patients without supplementation. The previous studies have yielded similar results, although the applied doses of vitamin K supplementation were different (≤ 86 μ g/d -10 mg/d) 6,8,[15][16][17] . Therefore, the use of the typical recommended vitamin K dose (2.1-10 mg/week) in CF patients 18 , may not be sufficient to maintain a normal vitamin K status.
The distribution of vitamin K deficiency or insufficiency and normal vitamin K status was different for adults and children. However, a multiple regression analysis showed that age was not a significant predictor of vitamin K status. Dougherty et al. found that the age of males was a significant predictor of u-OC percentage through a multiple regression analysis 6 .
Fat and bile malabsorption were suggested to be the main causes of fat-soluble vitamin deficiency in CF 19 . CF subjects with pancreatic insufficiency were proven to have significantly higher PIVKA-II concentrations than pancreatic sufficient patients or healthy controls 14,20 . Furthermore, Nicolaidou et al. documented higher levels of u-OC in CF subjects with pancreatic insufficiency than in healthy subjects 17 . In the present study, pathological PIVKA-II concentrations appeared more frequently in patients with the genotype F508del/F508del or with two severe mutations in both alleles. Unfortunately, there are no published data concerning the relationship between the CFTR gene mutations and vitamin K deficiency. However, severe mutations such as F508del are generally associated with pancreatic insufficiency and a more severe course of the disease 21 .
Based on a stepwise multiple regression analysis, liver disease, diabetes and glucocorticoid therapy were potentially defined as determinants of PIVKA-II concentrations, and the dose of vitamin K was for u-OC percentage. However, the multiple regression analysis failed to show any statistical significance in any of the regression models. Therefore none of the independent variables seem to be a strong predictive risk factor of vitamin K deficiency in CF.
Generally, vitamin K deficiency is common in patients with cholestatic and noncholestatic liver disease. Liver disease impairs γ -carboxylation of vitamin K-dependent proteins such as PIVKA-II 14,22 . Rashid et al. showed that all CF subjects with liver disease had higher levels of PIVKA-II than patients without liver involvement 14 . Additionally, Wilson et al. found abnormal PIVKA-II concentrations in all CF subjects with liver disease 16 .
There is no data concerning the potential influence of diabetes on vitamin K status in CF patients. However, some of the available evidence shows that vitamin K may decrease the risk of diabetes in healthy people 23,24 . Similarly, there is no evidence that glucocorticoid therapy affects vitamin K status in CF. In the general population, Mokuda et al. showed that oral glucocorticoid dosage is an independent risk factor of lowering u-OC in post-menopausal women 25 .
To summarize, although we documented some statistical relationships and differences, there have been no strong predictive factors of vitamin K deficiency in CF patients in the present study. Overall, the data concerning this issue is limited. Dougherty et al. found that vitamin K supplementation (p = 0.006), 25(OH)D 3 (p = 0.007), age for males (p = 0.045), and vitamin K supplementation only for females (p = 0.001) were significant predictors of % u-OC 6 . Nicolaidou et al. showed that vitamin K supplementation was a predictor of u-OC (p < 0.001), c-OC (p = 0.013), carboxy-terminal propeptide of type I procollagen (p < 0.001), and amino-terminal propeptide of type I procollagen (p < 0.001) concentrations in CF subjects 17  of type I procollagen, amino-terminal propeptide of type I procollagen -markers of early osteoblastic function, carboxy-terminal telopeptide of type I collagen -marker of bone resorption) and bone mineral density (BMD) determined by dual-energy X-ray absorptiometry (DEXA) are utilized in investigating impact on bone health of vitamin K which suggests an increasingly important role of vitamin K in bone health 5,17 . Available evidence also suggests that vitamin K is involved in inflammatory responses associated with bone metabolism. Kim et al. found that serum hs-CRP concentrations were positively correlated with vitamin K deficiency. The authors suggested that vitamin K influences the inflammatory status and bone formation 26 . Shea et al. found that proinflammatory cytokines such as IL-6, intercellular adhesion molecule-1 (ICAM-1), and tumor necrosis factor receptor 2 (TNFR2) were negatively correlated with vitamin K and hs-CRP levels 27,28 . This may be important to consider in further studies.
The major limitation of the present study is the lack of a direct measure of vitamin K status. Both PIVKA-II concentration and u-OC percentage are indirect measures, and may not ideally reflect vitamin K 1 and K 2 concentrations. Another limitation is a lack of vitamin K dietary intake assessment. However, the supplementation doses are many times higher, potentially reducing the significance of differences between individual dietary intakes. The large and well defined clinical study group is the main strength of this study.
In conclusion, vitamin K deficiency in CF patients seems to be very common and persists despite its supplementation. Liver involvement, diabetes, and glucocorticoid therapy were documented to be potential risk factors. Pathological concentrations of PIVKA-II are found more frequently in patients with pancreatic insufficiency and those who have two severe mutations in both alleles of the CFTR gene. However, it seems that there are no good predictive factors of vitamin K deficiency in CF patients in everyday clinical care. Early vitamin K supplementation in CF patients seems to be warranted. It is impossible to clearly determine the supplementation dose. Therefore, constant monitoring of vitamin K status seems to be justified.
The nutritional status (standardized value of the body weight and height 30 , and albumin concentration); clinical expression of the disease (lung function -spirometry, exocrine pancreatic function -fecal elastase-1 [31][32][33] , biochemical markers of liver function -ALT, AST, GGT, colonization of Pseudomonas aeruginosa, diabetes, liver cirrhosis 34 , CFTR genotype); international normalized ratio (INR/ as reflectance) the treatment (supplementation of vitamin K and pancreatic enzymes, inhaled and oral permanent antibiotic therapy, intravenous and oral antibiotic therapy in the preceding three months, and inhaled glucocorticoid therapy) were assessed. Vitamin K status was estimated by PIVKA-II concentration and the percentage of u-OC. PIVKA-II concentration was assessed using the Asserachrom PIVKA-II immunoassay kit (DeCarboxy Prothrombin, Diagnostica Stago, Asnières-sur-Seine, France). The percentage of u-OC was calculated using the following formula: u-OC[ng/ml]*100%/(u-OC[ng/ml] + c-OC[ng/ml]). Concentrations of carboxylated (c-OC) and undercarboxylated (u-OC) osteocalcin were assessed using the Gla-type Osteocalcin EIA and Glu-type Osteocalcin EIA immunoassay kits: (Takara BIO INC. Otsu, Shiga, Japan).
The Shapiro-Wilk test was used to check the normality of the data distribution. The Mann-Whitney test and chi 2 -test were used to assess differences between the subgroups with normal (PIVKA-II < 2 ng/ ml) and abnormal (PIVKA-II ≥ 2 ng/ml) vitamin K statuses. In turn, differences between subgroups of patients with normal (u-OC < 20%), insufficiency (u-OC 20-50%) and deficiency (u-OC > 50%) of vitamin K status were assessed using the ANOVA test (for Gaussian distribution), Kruskal Wallis test (for non-Gausssian distribution), or Fisher-Freeman-Halton test. The potential influence of all studied parameters on the occurrence of vitamin K deficiency were assessed using multiple logistic regression and stepwise multiple regression analysis. The level of significance was set at p < 0.05.
The study was conducted in accordance with the revised Declaration of Helsinki. Informed, written consent was obtained from patients' parents and from patients who were 16 years or older. The study was approved by the Bioethical Committee at Poznan University of Medical Sciences (decision 640/2009).