CIRCULATING DKK3 (DICKKOPF-RELATED PROTEIN 3) IN THE ASSESSMENT OF ATRIAL FIBRILLATION

Abstract

The present invention relates to a method for assessing atrial fibrillation in a subject, said method comprising the steps of determining the amount of DKK3 in a sample from the subject, and comparing the amount of DKK3 to a reference amount, whereby atrial fibrillation is to be assessed. Moreover, the present invention relates to a method for diagnosing heart failure and/or at least one structural or functional abnormality of the heart associated with heart failure.

Claims

1. A method for assessing atrial fibrillation in a subject, comprising the steps of a) determining the amount of DKK3 and optionally of a natriuretic peptide and/or of ESM1 in a sample from the subject, and b) comparing the amount of DKK3 and optionally of the natriuretic peptide and/or of ESM1 to a reference amount (or to reference amounts), whereby atrial fibrillation is to be assessed. c)

2. The method of claim 1, wherein an amount of DKK3 (and optionally an amount of the natriuretic peptide and/or of ESM1) in the sample from a subject which is (are) increased as compared to the reference amount (or to reference amounts) is indicative for a subject suffering from atrial fibrillation and/or wherein an amount of DKK3 (and optionally an amount of the natriuretic peptide and/or of ESM1) in the sample from a subject which is (are) decreased as compared to the reference amount (or to reference amounts) is (are) indicative for a subject not suffering from atrial fibrillation.

3. The method of claim 2, wherein the amounts of DKK3 and a natriuretic peptide and/or ESM1 are determined in step a), and wherein the method comprises the further steps of c) calculating a ratio of the amount of the natriuretic peptide and/or of ESM1 as determined in step a) to the amount of DKK3 as determined in step a), and comparing said calculated ratio to a reference ratio.

4. The method of claim 1, wherein the subject is suffering from atrial fibrillation, and wherein the assessment of atrial fibrillation is the differentiation between paroxysmal and persistent atrial fibrillation.

5. The method of claim 1, wherein the subject suffers from atrial fibrillation and wherein the assessment of atrial fibrillation is the assessment of a therapy for atrial fibrillation.

6. A method for diagnosing heart failure and/or at least one structural or functional abnormality of the heart associated with heart failure in a subject, said method comprising the steps of a) determining the amount of DKK3 in a sample from the subject, and b) comparing the amount of DKK3 to a reference amount, whereby heart failure and/or at least one structural or functional abnormality of the heart associated with heart failure is to be diagnosed.

7. Use of i) the biomarker DKK3 and optionally a natriuretic peptide and/or ESM1, and/or ii) at least one detection agent that specifically binds to DKK3, and optionally at least one detection agent that specifically binds to a natriuretic peptide and/or ESM1, in a sample from a subject for a) assessing atrial fibrillation or b) for diagnosing heart failure and/or at least one structural or functional abnormality of the heart associated with heart failure, or for the prediction of stroke.

8. A method of aiding in the assessment of atrial fibrillation, said method comprising the steps of: a) obtaining a sample from a subject, b) determining the amount of the biomarker DKK3 and optionally the amount of a natriuretic peptide and/or of ESM1 in said sample, and c) providing information on the determined amount of the biomarker DKK3 and optionally on the determined amount of the natriuretic peptide and/or of ESM1 to the attending physician of the subject, thereby aiding in the assessment of atrial fibrillation in said subject.

9. A method for aiding in the assessment of atrial fibrillation, comprising: a) providing a test for the biomarker DKK3 and optionally a test for a natriuretic peptide and/or for ESM 1 and b) providing instructions for using of test results obtained or obtainable by said test(s) in the assessment of atrial fibrillation.

10. A method for predicting the risk of stroke in a subject, comprising the steps of a) determining the amount of DKK3 in a sample from the subject, and b) comparing the amount of DKK3 to a reference amount, whereby the risk of stroke is to be predicted.

11. A method for improving the prediction accuracy of a clinical stroke risk score for a subject, comprising the steps of a) determining the amount of DKK3 in a sample from the subject having a known clinical stroke risk score, and a) combining the amount of DKK3 with the clinical stroke risk score, whereby the prediction accuracy of said clinical stroke risk score is improved.

12. A method for predicting the risk of stroke in a subject, comprising the steps of a) determining the amount of DKK3 in a sample from the subject having a known clinical stroke risk score, and b) combining the amount of DKK3 with the clinical stroke risk score, whereby the risk of stroke of said subject is to be predicted.

13. Use of i) the biomarker DKK3 and/or ii) at least one detection agent that specifically binds to DKK3 in a sample from a subject for improving the prediction accuracy of a clinical stroke risk score.

14. Use of i) the biomarker DKK3 and/or ii) at least one detection agent that specifically binds to DKK3, in a sample from a subject, in combination with a clinical stroke risk score, for predicting the risk of a subject to suffer from stroke.

15. A kit comprising an agent which specifically binds to DKK3 and an agent which specifically binds to a natriuretic peptide and/or to ESM1.

16. The method of claim 4, wherein an amount of DKK3 in the sample from a subject which is increased as compared to the reference amount is indicative for a subject suffering from persistent atrial fibrillation and/or wherein an amount of DKK3 in the sample from a subject which is decreased as compared to the reference amount is indicative for a subject suffering from paroxysmal atrial fibrillation.

Description

[0327] The figures show:

[0328] FIG. 1: Weighted Kaplan-Meier survival estimates for the two groups defined by baseline DKK3 measurement <=28 NPX vs >28 NPX.

[0329] FIG. 2: DKK3 differential expression in pers. AFib vs. SR tissue samples assessed by RNAseq of the MAPPING cohort. DKK3 in perAF is expressed 1.496 fold higher in pers. AF compared to SR (FDR=0.0001).

[0330] FIG. 3: Diagnostic value of DKK3 in PREDICTOR AFib sub panel; AUC=0.66. Diagnosis of AF.

[0331] FIG. 4: Diagnostic value of GISSI AF AFib sub panel; AUC (24 weeks)=0.64, AUC (52 weeks)=0.65. Diagnosis of AF.

EXAMPLES

[0332] The invention will be merely illustrated by the following Examples. The said Examples shall, whatsoever, not be construed in a manner limiting the scope of the invention.

Example 1: Differential Expression of D1(1(3 in Cardiac Tissue of AF Patients

[0333] Differential DKK3 expression levels have been determined in myocardial tissue samples from the right atrial appendage of n=40 samples

RNAseq analyses

[0334] Atrial tissue was sampled during open chest surgery because of CABG or valve surgery. Evidence of AF or SR (controls) was generated during surgery with simultaneous Endo-Epicardial High Density Activation Mapping. Patients with AF and controls were matched with regard to gender, age and comorbidities.

[0335] Atrial tissue samples were prepared for [0336] AF patients; n=11 samples [0337] control patients in SR; n=29 samples

[0338] Differential expression of DKK3 was determined in RNAseq analyses applying the algorithms RSEM and DESEQ2.

[0339] As shown in FIG. 2, DKK3 expression was found to be upregulated in the analyzed atrial tissues of the 11 patients with persistent AF versus the 29 sinus rhythm controls

[0340] The fold change in expression (FC) was 1.496. The FDR (false discovery rate) was 0.00001. The altered expression of DKK3 was determined in the damaged end organ, the atrial tissue. DKK3 mRNA levels were compared to results of high density mapping of the atrial tissue. Elevated DKK3 mRNA levels were detected in atrial tissue samples with conduction disturbances as characterized by electrical mapping. Conductance disturbances may be caused by fat infiltration or by interstitial fibrosis. The observed differential expression of DKK3 in atrial tissue of patients suffering from atrial fibrillation supports, that FABP is released in the circulation from the myocardium, in particular from the right atrial appendage and elevated serum/plasma titers assist the detection of episodes of AF.

[0341] It is concluded, that DKK3 is released from the heart into the blood and may aid the detection of AF episodes and predict the risk of developing AF related stroke.

Example 2: Detection of Atrial Fibrillation in Predictor Cohort—Screen and Identify Patients with Atrial Fibrillation Especially in an Elderly General Population

[0342] The PREDICTOR study was a population-based trial in elderly (±65 years), apparently healthy subjects (n=2001). Participants were referred to cardiology centers for clinical examination and comprehensive Doppler echocardiography and electrocardiogram measures. The patients suffered predominantly from heart failure stage B. However, some patients suffered from heart failure stage A or C. The atrial fibrillation sub-cohort comprises 29 subjects with an ongoing episode of atrial fibrillation during their visit and 83 matched controls.

[0343] DKK3 was determined in the atrial fibrillation sub-cohort selected from the PREDICTOR study. Elevated circulating DKK3 levels were observed in samples from subjects with ongoing atrial fibrillation versus controls.

Example 3: Detection of Paroxysmal Atrial Fibrillation

[0344] In the biomarker sub-study of the GISSI-AF trial, blood samples were collected at study entry, and after 6 and 12 months of follow-up. For more details on the GISSI-AF trial, see the main publication: GISSI-AF Investigators, New Engl J Med 2009; 360:1606-17. For more details on the biomarker substudy, see: Latini R et al., J Intern Med 2011; 269:160-71. For 382 patients DKK3 values from plasma samples were obtained at baseline. After 24 weeks, out of 360 patients 38 developed paroxysmal atrial fibrillation. After 52 weeks, 48 out of 357 have developed atrial fibrillation.

[0345] DKK3 was determined in the atrial fibrillation sub-cohort selected from the GISSI AF study. Elevated circulating DKK3 levels were observed in samples from subjects with ongoing atrial fibrillation versus controls.

Example 4: Prediction of Stroke

Analysis Approach

[0346] The ability of circulating DKK3 to predict the risk for the occurrence of stroke was assessed in a prospective, multicentric registry of patients with documented atrial fibrillation (Conen D., Forum Med Suisse 2012; 12:860-862). DKK3 was measured using a stratified case cohort design as described in Borgan (2000).

[0347] For each of the 70 patients which experienced a stroke during follow up (“events”), 1 matched control was selected. Controls were matched based on the demographic and clinical information of age, sex, history of hypertension, atrial fibrillation type and history of heart failure (CHF history).

[0348] DKK3 results were available for 69 patients with an event and 69 patients without an event.

[0349] DKK3 was measured using the Olink platform therefor no absolute concentration values are available and can be reported. Results will be reported on an arbitrary signal scale (NPX).

[0350] In order to quantify the univariate prognostic value of DKK3 proportional hazard models were used with the outcome stroke.

[0351] The univariate prognostic performance of DKK3 was assessed by two different incorporations of the prognostic information given by DKK3.

[0352] The first proportional hazard model included DKK3 binarized at the median (28 NPX) and therefore comparing the risk of patients with DKK3 below or equal to the median versus patient with DKK3 above the median.

[0353] The second proportional hazard model included the original DKK3 levels but transformed to a log2 scale. The log2 transformation was performed in order to enable a better model calibration.

[0354] Because the estimates from a naïve proportional hazard model on the case control cohort would be biased (due to the altered proportion of cases to controls) a weighted proportional hazard model was used. Weights are based on the inverse probability for each patient to be selected for the case control cohort as described in Mark (2006).

[0355] In order to get estimates for the absolute survival rates in the two groups based on the dichotomized baseline DKK3 measurement (<=28 NPX vs >28 NPX) a weighted version of the Kaplan-Meier plot was created as described in Mark (2006).

[0356] In order to assess if the prognostic value of DKK3 is independent from known clinical and demographic risk factors a weighted proportional cox model including in addition the variables age, sex, CHF history, history of hypertension, Stroke/TIA/Thromboembolism history, vascular disease history and diabetes history was calculated.

[0357] In order to assess the ability of DKK3 to improve existing risk scores for the prognosis of stroke the CHADS.sub.2 the CHA.sub.2DS.sub.2-VASc and the ABC score were extended by DKK3 (log2 transformed). Extension was done by creating a portioned hazard model including DKK3 and the respective risk score as independent variables.

[0358] The c-indices of the CHADS.sub.2, the CHA.sub.2DS.sub.2-VASc and ABC score were compared to the c-indices of these extended models. For the calculation of the c-index in the case-cohort setting a weighted version of the c-index was used as proposed in Ganna (2011).

Results

[0359] Table 1 shows the results of the two univariate weighted proportional hazard models including the binarized or the log2 transformed DKK3.

[0360] The association between the risk for experiencing a stroke with the baseline value of DKK3 is highly significant in both models.

[0361] The hazard ration for the binarized DKK3 implies a 1.8-fold higher risk for a stroke in the patient group with baseline DKK3>28 NPX versus the patient group with baseline DKK3 <=28 NPX. However the confidence interval includes the 1 which implies that either the median is not the optimal cutoff or that binarization is not appropriate for DKK3.

[0362] The results of the proportional hazard model including DKK3 as log2 transformed linear risk predictor suggest the log2 transformed values DKK3 are positively correlated to the risk for experiencing a stroke. The hazard ratio of 2.8 can be interpreted in a way that a 2-fold increase of DKK3 is associated with 2.8 increase of risk for a stroke.

TABLE-US-00001 TABLE 1 Results result of the univariate weighted proportional hazard model including the binarized and log2 transformed DKK3. Hazard Ratio (HR) 95%-CI HR P-Value DKK3 log2 2.755 1.340-5.667 0.006 Baseline 1.826 0.951-3.503 0.070 DKK3 > 28 NPX vs DKK3 <= 28 NPX

[0363] FIG. 1 shows the weighted Kaplan-Meier curves for the two patient groups with baseline DKK3 measurement (<=28 NPX vs >28 NPX).

[0364] Table 2 shows the results of a proportional hazard model including DKK3 (log2 transformed) in the combination with clinical and demographic variables. Although the point estimator of hazard ratio for DKK3 is still notable above 1 the p-value is now above 0.05. However given the still high hazard ratio and that the c-index of the model including only the clinical variables shown in table 2 improves by 0.0055 with the addition of DKK3 it can be expected that the effect of DKK3 would significant on larger cohort with more observed events.

TABLE-US-00002 TABLE 2 Multivariate proportional hazard model including DKK3 and relevant clinical and demographic variables. Hazard Ratio (HR) 95%-CI HR P-Value History hypertension 1.1710 0.5785-2.3703 0.6608 Age 1.0465 0.9917-1.1044 0.0975 History Stroke/TIA/embolism 1.9624 0.7768-4.9576 0.1539 Sex = male 0.8732 0.4378-1.7417 0.7003 History CHF 0.7594 0.3512-1.642  0.4842 History vascular disease 1.3173 0.5516-3.1458 0.5350 DKK3 (log2 transformed) 1.7581 0.5039-6.1345 0.3762

[0365] Table 3 shows the results of the weighted proportional hazard model combining the CHADS.sub.2 score with DKK3 (log2 transformed). Also in this model DKK3 can add prognostic information to the CHADS.sub.2 score. Similar to table 2 the hazard ratio of DKK3 is still above 1 but again with a p-value not reaching 0.05. Also here the relative small number of events has to be considered.

TABLE-US-00003 TABLE 3 Weighted proportional hazard model combining the CHADS.sub.2 score with DKK3 (log2 transformed) Hazard Ratio (HR) 95%-CI HR P-Value CHADS.sub.2 score 1.4401 1.1196-1.8525 0.0045 DKK3 (log2 1.9424 0.8820-4.2777 0.0993 transformed)

[0366] Table 4 shows the results of the weighted proportional hazard model combining the CHA.sub.2DS.sub.2-VASc score with DKK3 (log2 transformed). Similar to table 2 the hazard ratio of DKK3 is still above 1 but again with a p-value not reaching 0.05. Also here the relative small number of events has to be considered.

TABLE-US-00004 TABLE 4 Weighted proportional hazard model combining the CHA.sub.2DS.sub.2-VASc score with DKK3 (log2 transformed) Hazard Ratio (HR) 95%-CI HR P-Value CHA.sub.2DS.sub.2-VASc 1.4361 1.1284-1.8277 0.0033 score DKK3 (log2 1.511 0.6273-3.6399 0.3574 transformed)

[0367] Table 5 shows the results of the weighted proportional hazard model combining the ABC score with DKK3 (log2 transformed). Similar to table 2 the hazard ratio of DKK3 is still above 1 but again with a p-value not reaching 0.05. Also here the relative small number of events has to be considered.

TABLE-US-00005 TABLE 5 Weighted proportional hazard model combining the ABC score with DKK3 (log2 transformed) Hazard Ratio (HR) 95%-CI HR P-Value ABC score 1.1703 1.0453-1.3103 0.0064 DKK3 (log2 1.4801 0.5996-3.6535 0.3950 transformed)

[0368] Table 6 shows the estimated c-indexes of DKK3 alone, of the CHADS.sub.2, the CHA.sub.2DS.sub.2-VASc and the ABC score and of the weighted proportional hazard model combining the CHADS.sub.2, the CHA.sub.2DS.sub.2-VASc and the ABC score with DKK3 (log2).

[0369] The addition of DKK3 to CHA.sub.2DS.sub.2-VASc score improves the c-index by 0.0028 which can be considered still as a clinical meaningful improvement of the risk prediction. For the CHADS.sub.2 score the c-index improvement is higher with 0.0090 and for the ABC score it is the highest with 0.0163.

TABLE-US-00006 TABLE 6 C-indexes of DKK3, the CHADS.sub.2, CHA.sub.2DS.sub.2-VASc and ABC score and their combination with DKK3. C-Index DKK3 univariate 0.6301 CHADS.sub.2 0.6541 CHADS.sub.2 + DKK3 0.6632 CHA.sub.2DS.sub.2-VASc 0.6800 CHA.sub.2DS.sub.2-VASc + DKK3 0.6828 ABC score 0.6531 ABC score + DKK3 0.6694

[0370] The results suggest that DKK3 can be used in several ways to predict the risk for a future stroke for a new patient, either alone, or as a combination to considerably improve the clinical scores in predicting stroke risk (such as CHADS.sub.2 and CHA.sub.2DS.sub.2-VASc).

[0371] For a new patient DKK3 could be measured and compared to a pre-defined cutoff. If the measured value for the new patient is above the predefined cutoff the patient is considered to have a high risk for the experience of stroke and appropriate clinical measures could be initiated.

[0372] It is also possible to define more than two risk groups based on an increasing set of cutoffs. A patient would then be assigned to one of the risk groups based on the value of his DKK3 measurement. The risk for a stroke is expected to increase over the different risk groups.

[0373] Alternatively, it would be also possible to transform the results of DKK3 directly into a continuous risk score based on pre-defined suitable transformation function.

[0374] In addition, it is possible to use the value of DKK3 in a combination with a risk score based on clinical and demographic variables (e.g. CHA.sub.2DS.sub.2-VASc score) and thereby improve the precision of the risk prediction.

[0375] For a new patient the value for risk score would be assessed and combined in an appropriate way with the measured DKK3 values (potentially log2 transformed), e.g. by creating a weighted sum of the risk score results and the DKK3 value with appropriate pre-defined weights (e.g. as shown in table 3).

Example 3: Biomarker Measurements

[0376] DKK3 was measured in a commercially available O-link multi-marker panel for Dickkopf related protein 3 (DKK3); Proximity Extension Assay from O-link, Sweden.