Suggested Cut-Off Values for Vitamin D as a Risk Marker for Total and Cardiac Death in Patients with Suspected Acute Coronary Syndrome

Background Several studies have demonstrated an association between low vitamin D levels and cardiovascular risk. Vitamin D cut-off levels are still under debate. Objectives To assess two cut-off levels, 40 and 70 nmol/L, respectively, for vitamin D measured as 25-hydroxyvitamin D in chest pain patients with suspected acute coronary syndrome. Methods We investigated 1853 patients from coastal-Norway and inland Northern-Argentina. A similar database was used for pooling of data. Two-year follow-up data including all-cause mortality, cardiac death, and sudden cardiac death in the total patient population were analyzed, applying univariate and multivariable analysis. Results Two hundred fifty-five patients with known vitamin D concentrations died. In the multivariable analysis, there was a decrease in total mortality above a cut-off level of 40 nmol/L and a decrease in cardiac death above a cut-off level of 70 nmol/L [HRs of 0.66 (95% CI, 0.50–0.88), p = 0.004 and 0.46 (95% CI, 0.22–0.94), p = 0.034, respectively]. Conclusion Vitamin D cut-off levels of 40 and 70 nmol/L were related to total mortality and cardiac death, respectively.

inTrODUcTiOn Vitamin D is synthesized in the skin on exposure to UV light, or obtained from the diet, mainly from fatty fish (1). Vitamin D deficiency in humans is prevalent and cardiovascular disease (CVD) is the primary cause of mortality worldwide (2). Several observational and epidemiological studies have demonstrated an association between low levels of vitamin D and (CVD) and mortality (3)(4)(5)(6)(7)(8). Vitamin D, a Predictor of Mortality in ACS Frontiers in Cardiovascular Medicine | www.frontiersin.org We have previously studied separately two populations, one from coastal Norway and one from inland Northern-Argentina, with respect to the prognostic significance of vitamin D, measured as 25-hydroxyvitamin D [25(OH)D] in acute coronary syndrome (ACS). Fish is frequently consumed by the costal population of Norway, and is less preferred by the inland beef-consuming population in Northern Argentina. At the time of this study, vitamin D fortification in food had not been introduced in that country. Furthermore, the subtropical location of Salta, Argentina, is associated with a higher exposure to sun in comparison to that of the temperate location of Norway. Our two patient populations were divided into 25(OH)D quartiles, comparing patients with high and low levels of vitamin D. We found that vitamin D was an independent prognostic biomarker in relation to total mortality and cardiac death in the Argentinean population, whereas in the Norwegian population its prognostic utility was less evident, but with a clear trend supporting the findings obtained in the Argentinean study (9,10).
Optimal and exact cut-off levels of vitamin D are still under debate (8,(11)(12)(13). Additionally, studies have shown that high levels of vitamin D are also associated with increased mortality (3,14).
In the present study, we have pooled our two populations and grouped them according to gender, in order to better assess the prognostic significance of 25(OH)D according to (1) 25(OH)D groups and (2) two cut-off values. We have added information based on a healthy blood donor population from both countries in order to obtain a reference value for vitamin D.

MaTerials anD MeThODs study Design and Patient Population
Participants in the present analysis were pooled from two studies, both designed to evaluate the prognostic utility of serum 25(OH) D status in patients with chest pain and ACS.
The RACS study (Risk Markers in the Acute Coronary Syndrome) (ClinicalTrials.gov NCT00521976) was a singlecenter prospective observational study in which 871 patients with chest pain and suspected ACS were consecutively admitted to the Stavanger University Hospital from November 2002 to October 2003.
The ARgentinean Risk Assessment Registry in the Acute Coronary Syndrome, the "ARRA-RACS Study" (Ref. Clinical Trial.gov identifier: NCT01377402), was performed at nine hospitals in the Province of Salta, Northern Argentina, in which 982 patients with chest pain and suspected ACS were hospitalized from December 2005 to January 2009. The details, including blood sampling procedures, laboratory measurements, and exclusion criteria of the RACS and ARRA-RACS studies, have been published previously (9,10,15).
The primary outcome in both studies was 2-year all-cause mortality, and the secondary outcomes included cardiac death and sudden cardiac death (SCD). The population was divided into three groups according to the following 25(OH)D: ≤40 nmol/L, between >40 and ≤70 nmol/L, and >70 nmol/L, and two cutoff levels; ≤40 nmol/L and ≤70 nmol/L. The prognostic impact within the suggested 25(OH)D levels was analyzed applying univariate and multivariable analysis.
The blood bank study included blood samples from 86 healthy individuals, with an equal number of males and females, recruited from the blood bank in Stavanger, Norway, and 104 healthy individuals (53 females and 49 males) recruited from two blood banks in Salta, Argentina. Their information was treated anonymously. 25(OH)D measurements were carried out at the Department of Medical Biochemistry at Stavanger University Hospital, as previously described (9).

sTaTisTical analYsis
Patients were divided into three categories according to their 25(OH)D levels. Results are presented for the complete 2-year follow-up period and after censoring the events in the first 3 months (91 days). Approximately, normally distributed variables were given as mean and standard deviation (SD) within vitamin D groups. The Chi-square test for association was applied between the 25(OH)D and categorical variables at baseline. The one-way ANOVA was used to test for the equality of means of scale variables (e.g., age) among vitamin D categories and the two-sample t-test was used for comparing means of two samples.
Stepwise Cox multivariable proportional hazards regression models with total death, cardiac death, and SCD, as the dependent variable, and 25(OH)D categories and other variables as potential independent predictors (listed below) were fitted. To examine the differences in prognosis between subjects in the higher vs. the lowest 25(OH)D group, we adjusted for gender, age, smoking, hypertension, index diagnosis, diabetes mellitus, congestive heart failure (CHF) [defined by Killip-Kimball class (16) at admission; patients in classes 2-4 were classified as CHF patients and patients in class 1 were classified as non-CHF], history of previous coronary heart disease (i.e., history of either angina pectoris, MI, coronary artery bypass graft, or percutaneous coronary intervention), hypercholesterolemia/use of statins, TnT > 0.01 ng/mL, estimated glomerular filtration rate, high sensitivity C-reactive protein, B-type natriuretic peptide, body mass index (BMI) (kg/m 2 ), month of sampling, country, and use of betablockers prior to enrollment. The hazard ratios (HRs) are presented with 95% confidence interval (CI). The Kaplan-Meier product limits were used for plotting time-to-event and the log-rank test was used to compare survival curves across 25(OH)D groups. The added prognostic value by including 25(OH)D levels to the significant confounders in the Cox regression model was assessed by using the log-likelihood test. The statistical analyses were performed using the statistical package SPSS version 19.0. All tests were two sided with a significance level of 5%. Highly significant differences were based on a two-tailed significance level of 1%.

eThics sTaTeMenT
The RACS study was approved by the Regional Board of Research Ethics and the Norwegian Health Authorities and conducted in accordance with the Helsinki Declaration of 1971, as revised in 1983. The ARRA-RACS study was approved by the Ethics Committee at the Board of Medical School of Salta and conducted in accordance with the Helsinki Declaration of 1971, as revised in 1983. At San Bernardo Hospital and Sanatorio El Carmen, the study was also approved by the local Ethics Committee and Institutional Review Board of San Bernardo Hospital and the Institutional Review Board of Sanatorio El Carmen, respectively. The Norwegian biobank containing Argentinean blood samples was approved by the Regional Board of Research Ethics and the Norwegian health authorities. This study was monitored by Stavanger Health Research, Stavanger, Norway.
The blood bank study was approved by the Regional Board of Research Ethics.
Written informed consent was obtained from all patients in all three studies.  Table 1.

Prognosis estimates related to Different Vitamin D concentrations
After a 2-year follow-up, 255 patients with known 25(OH)D values had died. Of these, 150 deaths were caused by a cardiac event, of whom, 75 patients died suddenly. Kaplan-Meier survival curves for all-cause mortality in three 25(OH)D groups are presented in Figure 1. Receiver-operated characteristics (ROC) curve for 25(OH)D for all-cause mortality is shown in Figure 2. The area under the ROC for 25(OH)D was 0.619 (p < 0.001). Using 40 nmol/L as threshold, the sensitivity was 0.435 and specificity was 0.744. Using 70 nmol/L as threshold, the sensitivity was 0.898 and specificity was 0.150.
In the univariate Cox regression analysis, there was a highly statistically significant difference between the upper (>70 nmol/L) and middle (40-≤70 nmol/L) as compared to the lowest group of 25(OH)D with respect to all-cause mortality, cardiac death, and SCD. These findings were maintained for total death in the multivariable Cox regression analysis, with HRs of 0.56 (95% CI, 0.34-0.92), p = 0.021 and 0.69 (95% CI, 0.51-0.92), p = 0.013 in the upper groups, respectively. However, for cardiac death and SCD, the difference in prognosis was significant only when comparing the upper with the lowest group, with HRs of 0.41 (95% CI, 0.19-0.87), p = 0.021 and HR of 0.23 (95% CI, 0.05-0.96), p = 0.044, respectively.
Similar results were obtained for total mortality and cardiac mortality after censoring the first 3 months. The univariate and multivariable HRs for 25(OH)D groups for the complete followup and after censoring the first 3 months are presented in Tables 2 and 3, respectively.

Prognosis estimates Based on set cut-Off Values of Vitamin D
In the multivariable analysis, there was a statistically significant decrease in all-cause mortality in the population with 25(OH)D level above 40 nmol/L as compared to those with a level below or equal to 40 nmol/L, with a HR of 0.66 (95% CI, 0.50-0.88), p = 0.004, whereas no significant difference was found in relation to cardiac and SCD. Furthermore, these differences were maintained when assessing the total population according to gender.
In the multivariate analysis, we also noted a statistically significant decrease in cardiac death in patients with 25(OH)D levels above 70 nmol/L as compared to those with levels equal to or below 70 nmol/L, with a HR of 0.46 (95% CI, 0.22 -0.94), p = 0.034, whereas no significant difference was found in relation to total mortality and SCD. Furthermore, no significant differences were found when assessing the total population according to gender.  Table 4.

DiscUssiOn
As optimal and exact cut-off levels of vitamin D are still under debate, we have addressed this issue in the present analysis and studied the prognostic value of vitamin D cut-off values of 40 and 70 nmol/L, respectively, in a pooled ACS suspected patient population from inland Argentina and coastal-Norway, applying univariate and multivariate analyses. Furthermore, survival was compared for patients with vitamin D levels ≤40 nmol/L, >40 to ≤70 nmol/L, and >70 nmol/L, respectively. The proportion of total-, cardiac-and SCD decreased significantly with increasing vitamin D levels in the univariate analysis. This pattern was maintained for total death in the multivariable analysis, whereas cardiac death and SCD differed only when comparing the highest to the lowest group. Adding 25(OH)D levels to the significant confounders improved the prognostic value of the Cox regression model. We chose a cut-off level of 40 nmol/L for vitamin D, based on a report by Ross et al (12), claiming that a vitamin D concentration above this level will meet the needs of 50% of the population. As demonstrated previously (10), patients in the lower vitamin D group had a significant lower survival rate than those with levels above 40 nmol/L. These differences were maintained when assessing the total population according to gender. In our Vitamin D, a Predictor of Mortality in ACS Frontiers in Cardiovascular Medicine | www.frontiersin.org reFerences population, 28% of the patients had vitamin D levels below or equal to 40 nmol/L. Patients in the lower level vitamin D group were older, had higher BMI, and were sicker as judged by a higher rate of heart insufficiency and diabetes as compared to patients in the higher vitamin D group. As suggested by the US Endocrine Society, levels below 50 nmol/L are defined as deficiency/insufficiency and levels above 75 nmol/L are regarded as sufficient (13). Accordingly, we chose a second cut-off level of 70 nmol/L and above this level we demonstrated a statistically significant reduced HR of cardiac death. It has previously been reported that also high concentrations of 25(OH)D are associated with elevated risks of overall mortality (3,14), but this relationship was not noted in the present study.
Vitamin D measurements were performed in blood samples obtained at admission, as close as possible to the index event, decreasing the possibility of a fall of vitamin D during hospitalization and reconvalescence. Mean vitamin D levels in our study patient population were similar to that of a healthy blood donor population, suggesting that low vitamin D levels are not necessarily secondary to patients' health status. The similarity in vitamin D status between patients and healthy individuals may support the utility of vitamin D as a primary predictor of future outcome in our ACS patient population.
With the applied cut-off levels of 40 and 70 nmol/L, the sensitivity and specificity favors the use of the former cut-off level, demonstrating a sensitivity of 0.435 and specificity of 0.744, whereas a cut-off level of 70 nmol/L is associated with a sensitivity of 0.898 and a low specificity of 0.150.
Strengths consist of an unselected chest pain population with suspected ACS and a relatively short follow-up time of 2 years.
Limitations have previously been described (9, 10) and consist mainly of one set of blood samples at admission and no measurements of parathyroid hormone. As this is an observational study, unknown and residual confounders may have been missed. The "parent" studies RACS and ARRA-RACS were performed in two countries in different hemispheres with different sun exposure and dietary habits. Lifestyle confounders, including physical activity, exposure to sunlight, diet, and alcohol consumption are difficult to measure and could potentially influence the results. Furthermore, reverse causation is inherent in an observational study; thus, vitamin D concentrations could simply be an indicator of chronic disease or a marker of a lifestyle. Although blood samples were drawn in each country, vitamin D analysis was performed by one biochemist at the Department of Medical Biochemistry at Stavanger University Hospital.
In conclusion, we noted an increased rate of total mortality with a level of vitamin D below 40 nmol/L and increased cardiac mortality was found to be related to vitamin D levels below 70 nmol/L. The results might have practical implications in terms of prophylactic vitamin D treatment in subjects at high risk of coronary heart disease, but we need randomized intervention studies to assess whether vitamin D supplementation reduces mortality and CVDs. aUThOr cOnTriBUTiOns PN: contributed to study design, data collection, vitamin D analysis, interpretation of results, and preparation of the manuscript. RF: contributed to study design, data collection, clinical followup, interpretation of results and preparation of the manuscript. STN: contributed to study design, interpretation of results, and preparation of the manuscript. VP: contributed to study design, data collection, clinical follow-up, interpretation of results, and preparation of the manuscript. TB-A: contributed to study design, data collection, clinical follow-up, interpretation of results, and preparation of the manuscript. HG: contributed to study design, data collection, interpretation of results, and preparation of the manuscript. HS: contributed to data collection, interpretation of results, and preparation of the manuscript. DN: conceived the idea of the study, supervised the study, including interpretation of results, and preparation of the manuscript. All authors have read and approved the final manuscript.

FUnDing
This study was funded by Helse Stavanger HF.