IJC Heart & Vasculature 39 (2022) 100977 Contents lists available at ScienceDirect IJC Heart & Vasculature journal homepage: www.sciencedirect.com/journal/ijc-heart-and-vasculature Proteins and pathways in atrial fibrillation and atrial cardiomyopathy underlying cryptogenic stroke Elena Palà a, 1, Jorge Pagola b, 1, Jesus Juega b, 1, Jaume Francisco-Pascual c, d, 1, Anna Penalba a, 1, Maite Rodriguez b, 1, Mercedes De Lera Alfonso e, 1, Juan F. Arenillas e, 1, Juan Antonio Cabezas f, 1, Francisco Moniche f, 1, Reyes de Torres g, 1, Soledad Perez-Sanchez g, 1, Teresa González-Alujas d, h, 1, Carlos A. Molina b, 1, Alejandro Bustamante a, 1, Joan Montaner a, i, *, 1 a Neurovascular Research Laboratory, Vall d’Hebron Institute of Research (VHIR), Hospital Vall d’Hebron, Universitat Autònoma de Barcelona, Spain b Stroke Unit, Vall d’Hebrón Hospital and Autonomous University of Barcelona, Medicine Department, Barcelona, Spain c Arrhythmia Unit-Cardiology Department, Vall d’Hebrón Hospital, Barcelona, Spain d Centro de Investigación Biomédica en Red de Enfermedades Cardiovasculares, CIBERCV, Spain e Stroke Unit, University Hospital of Valladolid, Valladolid, Spain f Stroke Unit, University Hospital Virgen del Rocio, Seville, Spain g Stroke Unit, University Hospital Virgen Macarena, Seville, Spain h Echocardiography Lab Cardiology Department, Vall d’Hebrón Hospital, Barcelona, Spain i Institute de Biomedicine of Seville, IBiS/Hospital Universitario Virgen del Rocío/CSIC/University of Seville & Department of Neurology, Hospital Universitario Virgen Macarena, Seville, Spain A R T I C L E I N F O A B S T R A C T Keywords: Background: Atrial fibrillation (AF) is one of the most prevalent causes of cryptogenic stroke. Also, apart from AF Atrial fibrillation itself, structural and remodelling changes in the atria might be an underlying cause of cryptogenic stroke. We Atrial cardiomyopathy aimed to discover circulating proteins and reveal pathways altered in AF and atrial cardiomyopathy, measured Biomarkers by left atrial volume index (LAVI) and peak atrial longitudinal strain (PALS), in patients with cryptogenic stroke. Cryptogenic stroke Atrial function Methods: An aptamer array (including 1310 proteins) was measured in the blood of 20 cryptogenic stroke pa­ tients monitored during 28 days with a Holter device as a case-control study of the Crypto-AF cohort. Protein levels were compared between patients with (n = 10) and without AF (n = 10) after stroke, and the best can­ didates were tested in 111 patients from the same cohort (44 patients with AF and 67 without AF). In addition, in the first 20 patients, proteins were explored according to PALS and LAVI values. Results: Forty-six proteins were differentially expressed in AF cases. Of those, four proteins were tested in a larger sample size. Only DPP7, presenting lower levels in AF patients, was further validated. Fifty-seven proteins correlated with LAVI, and 270 correlated with PALS. NT-proBNP was common in all the discovery analyses performed. Interestingly, many proteins and pathways were altered in patients with low PALS. Conclusions: Multiple proteins and pathways related to AF and atrial cardiomyopathy have been revealed. The role of DPP7 as a biomarker for stroke aetiology should be further explored. Moreover, the present study may be considered hypothesis-generating. 1. Introduction remains undetected in a high proportion of patients after stroke and is one of the most prevalent causes of cryptogenic stroke [2]. Atrial fibrillation (AF) is a prevalent cardiac rhythm disorder un­ Also, there is increasing recognition that atrial dysfunction itself is derlying up to one-third of all ischemic strokes [1]. Paroxysmal AF associated with an increased risk of thromboembolism, even in patients * Corresponding author at: Neurovascular Research Laboratory, Vall d’Hebron Institute of Research (VHIR), Hospital Vall d’Hebron, Pg. Vall d’Hebron 119-129, 08035 Barcelona, Spain. E-mail address: joan.montaner@vhir.org (J. Montaner). 1 These authors take responsibility for all aspects of the reliability and freedom from bias of the data presented and their discussed interpretation. https://doi.org/10.1016/j.ijcha.2022.100977 Received 14 January 2022; Received in revised form 16 February 2022; Accepted 17 February 2022 2352-9067/© 2022 The Author(s). Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/). E. Palà et al. IJC Heart & Vasculature 39 (2022) 100977 without AF [3]. Therefore, atrial substrate or atrial cardiomyopathy has between independent assay runs. Seventy-nine proteins were marked as been proposed as an important cause of cryptogenic strokes [4,5]. “flags” due to high inter-plate variability and eliminated from the Left atria (LA) enlargement and atrial fibrosis are two structural analysis. Data were reported in relative fluorescent units (RFU) after hallmarks of the atrial substrate [3]. LA size has been associated with normalization and calibration. cardioembolic stroke and AF detection in patients with embolic stroke of undetermined source (ESUS) [6]. Similarly, peak atrial longitudinal 2.3. Elisa strain (PALS), which measures the LA wall deformability and is a sur­ rogate of LA fibrosis, has been associated with AF in cryptogenic stroke Serum coiled-coil domain-containing protein 80 (CCDC80)(Bos­ patients [7]. terBio), and plasma dipeptidyl peptidase 7 (DPP7)(R&D Systems), bone Some blood biomarkers (e.g. natriuretic peptides) have been pro­ morphogenetic protein 1 (BMP-1)(Elabscience), and cystatin-D (Bos­ posed as useful tools to detect paroxysmal AF [8]. In addition, circu­ terBio) were determined by ELISA. All assays were performed blinded to lating markers might allow noninvasive assessment of atrial clinical information and according to the manufacturer’s instructions. cardiomyopathy before AF appears and might guide the selection of All samples were tested in duplicate, and inter-assay variation was patients for more intensive post-stroke monitoring to personalize the determined by a commercial control (Human Serum, male AB, USA secondary prevention treatments [2,5]. origin from clotted, SIGMA, ref number H16914; Human plasma K2 The present study aims to discover circulating proteins and reveal EDTA, Innovative Research, ref number IPLA-N) tested in duplicate in pathways altered in AF and atrial cardiomyopathy, measured by left each plate. When inter-assay variation was > 20%, biomarker levels atrial volume index (LAVI) and PALS, in patients with cryptogenic were standardized by the common control sample. Samples with a CV stroke. (coefficient of variation) > 20% between duplicates were eliminated from the analysis. 2. Methods 2.4. Statistics 2.1. Study population R software version 3.6.1 and SPSS version 20 were used to conduct The study population represented a subpopulation of the Crypto-AF statistical analysis. Categorical variables were expressed as numbers and study [9,10]. Non-lacunar acute ischemic stroke patients over 55 years percentages and continuous variables as mean ± SD, or median (inter­ of age with cryptogenic stroke after standard evaluation were included quartile range) for continuous variables, depending on their distribu­ in the study by four Spanish public Stroke Centers from January 2015 to tion. Student’s t-test, Mann–Whitney or x2 were used to compare July 2017. All patients included had no prior history of AF. Patients were variables between AF cases and controls depending on the type and monitored for 28 days with a wearable Holter device (NuuboTM) distribution of each variable. following a published protocol [9]. The software algorithm classified SOMAscan data were log-transformed as presented a skewed distri­ every episode of irregular ECG rhythm lasting > 120 s as possible AF. An bution. Differential expression analyses were performed using the expert cardiologist blinded to clinical data verified the episodes. From “limma” package (Bioconductor) version 3.42.2, optimized for omics this cohort, a total of ten consecutive patients with AF detected during studies with large amounts of data and few samples [15]. the monitoring period, and ten matched controls (by sex and age) Spearman correlations were calculated between LAVI or PALS and all without AF were selected for the present discovery study. In addition, the analyzed proteins. The R package “Venndiagram” version 1.6.20 was 111 patients with available blood samples were used for the validation used to visualize the common proteins between the different analyses. study (44 patients with AF and 67 matched controls). All p-values were adjusted using Benjamini and Hochberg (BH) false Blood samples were collected into EDTA and serum separator tubes discovery rate (FDR). Group matching by sex and age was used to select within 72 h after symptoms onset. After centrifugation at 1500 g and control samples in the discovery experiment. The validation sample size 4 ◦ C for 15 min, plasma and serum aliquots were frozen at − 80 ◦ C until was estimated based on Somascan results (power of 80%, α = 0.05) (Ene further analysis. 3.0, GlaxoSmithKline, UK). Left atrial size was measured by biplane transthoracic echocardiog­ The addition of DPP7 to a logistic regression model fitted by age, sex, raphy to obtain the left atrial volume adjusted to body surface index echocardiographic markers (LAVI and PALS), and NT-proBNP was (LAVI, ml/m2) following the latest guidelines [11]. Peak atrial longitu­ tested using the Likelihood Ratio Test. Odds ratios (OR) for an increment dinal strain (PALS) was evaluated by speckle tracking software (GE of one unit of concentration were shown. The classification performance EchoPAC®) following expert recommendations [12]. of the models was compared using Reciever Operating Curves. The R Written informed consent was obtained from all participants, and the package “ggeffects” version 1.1.1 was used to plot the average predicted study was approved (PR (AG)49/2014) by the Ethical Committee of Vall probability of the model when varying the variable of interest. d’Hebrón Hospital, Valladolid Hospital, Virgen Macarena Hospital and Virgen del Rocio Hospital, in line with Helsinki guidelines. 2.5. Pathway analysis 2.2. Aptamer array Pathway enrichment analysis was conducted following a published protocol [16]. Gene Set Enrichment Analysis (GSEA) software was Protein levels in plasma were assessed using the SOMAscan® plat­ applied to all the SOMAscan proteins ordered by T-statistic or correla­ form (SomaLogic Inc., Boulder, CO, USA), which is an aptamer-based tion coefficient against Reactome Pathways and Gene Ontology (bio­ proteomic assay that allowed the simultaneous measurement and logical processes) databases. Gene sets with < 15 genes or > 200 genes quantification of 1310 proteins [13]. This approach uses SOMAmers® were excluded. GSEA calculates a normalized enrichment score (NES) reagents, which are short single-stranded DNA sequences with protein for each gene set. Positive and negative NES values represent enrich­ affinity. The platform transforms the proteins present in the biological ment of the corresponding gene set at the top (i.e., upregulated) or sample into a corresponding SOMAmer signal, which then is quantified bottom (i.e., downregulated) of the ranked list. P-values were computed using the microarrays technology. Three different dilutions (depending by gene set permutation for 1000. Then, multiple testing using a false- on each protein abundance) were used. Normalization and calibration discovery rate (FDR) was applied to obtain the Q-values. Results were procedures were performed by SomaLogic according to their protocol visualized via Cytoscape Enrichment Map with a Jaccard Overlap [14]. All samples passed SomaLogic quality controls. A set of control Combined Coefficient > 0.375. Significant pathways were considered at calibrator samples were used to detect and remove systematic variability Q-value < 0.25. 2 E. Palà et al. IJC Heart & Vasculature 39 (2022) 100977 3. Results The descriptive characteristics of the 20 patients included in the discovery study are provided in Table 1. The median age was 71.5, and 55% were women. Clinical variables were similar between the two groups. 3.1. Differential protein expression and altered pathways in AF Among the tested proteins, 46 were differentially expressed in AF cases at a nominal p-value of 0.05 (22 down-regulated and 24 up- regulated). Although no protein remained significant after multiple comparison correction, NT-proBNP showed the strongest association (p- value = 0.001, logFC = 1.86) and BNP was ranked sixth (p-value = 0.008, logFC = 0.39). Both natriuretic peptides were well-known bio­ markers of AF, already validated in this cohort in a published study [8]. Proteins with p-values between NT-proBNP and BNP were selected to evaluate their usefulness in a larger group of patients: CCDC80 (p-value = 0.0013), DPP7 (p-value = 0.0039), BMP-1 (p-value = 0.0051), and Cystatin-D (p-value = 0.0080) (Fig. 1 and Supplemental Table 1). Fig. 1. Volcano plot of differentially expressed proteins between AF cases GSEA analysis revealed five gene sets upregulated in patients with AF and no AF. Black dots above the red line indicate significant proteins, while (Regulation of cellular response to growth factor stimulus, Lymphocyte grey dots below the red line indicate non-significant proteins according to chemotaxis, Calcium ion transmembrane transport, Lymphocyte nominal p-value < 0.05. Labeled proteins are those with a nominal p-value < migration, Chemokine receptors bind chemokines), and five gene sets 0.01 or nominal p-value < 0.05 and |logFC|>1.Proteins with positive logFC had higher levels in the AF group and vice-versa. (For interpretation of the refer­ downregulated (Binding and Uptake of Ligands by Scavenger Receptors, ences to colour in this figure legend, the reader is referred to the web version of Interleukin-12 family signalling, Vesicle-mediated transport, Transport this article.) of small molecules and Interleukin-12 signalling). Most altered path­ ways were related to immune response and intracellular transport 3.1.1. Biomarker validation (Supplemental Fig. 1). The clinical characteristics of the 111 patients included in the vali­ dation study were provided in Supplemental Table 2. This subgroup was older (p = 0.021) and had a lower percentage of obesity (p = 0.041) in comparison to the discovery cohort. Several samples had a CV > 20% between duplicates and were eliminated from the analysis accordingly. Table 1 Clinical characteristics of the patients included in the discovery experiment and Therefore, we had a final sample size of 107 patients for CCDC80, 86 for comparison according to atrial fibrillation detection. DPP7, 95 for BMP1, and 97 for Cystatin-D. Levels from the four bio­ markers were available from 74 patients. Analysis was performed with All (n = AF (n = 10) No AF (n = p- 20) 10) value the patients available for each biomarker. DPP7 was significantly lower in individuals with AF compared with no AF [5.13 ng/ml (IQR Sex (%female) 11 (55%) 6 (60%) 5 (50%) 0.65& 3.42–7.10) vs 6.16 ng/ml (5.04–8.39), p = 0.013], while the remaining Age (years) 71.5 73.5 (69.75–80) 67.5 0.247$ (67–80) (62.25–81.5) biomarkers were not different between the two groups (Fig. 2 and Hypertension 14 (70%) 7 (70%) 7 (70%) 1.00& Supplemental Table 3). Sensitivity analyses were performed only, Diabetes 5 (25%) 3 (30%) 2 (20%) 1.00& including the patients with values from all the biomarkers, obtaining Vasculopathy 1 (5%) 0 (0%) 1 (10%) 1.00& similar results. Renal failure 1 (5.3%) 1 (11.1%) 0 (0%) 0.47& COPD 1 (5.3%) 1 (11.1%) 0 (0%) 0.47& DPP7 alone was a predictor of AF (OR = 0.828 per ng/ml Obesity 8 (40%) 5 (50%) 3 (30%) 0.65& [0.695–0.986], p = 0.034). Then, when adding DPP7 to a logistic Heart disease 2 (10%) 2 (20%) 0 (0%) 0.47& regression model fitted with age, sex, PALS, LAVI, and NT-proBNP, it Basal NIHSS 4 (2–7) 3 (1–7) 5 (3–7) 0.29$ remained an independent predictor (OR = 0.729 per ng/ml LVEF (%) 64.74 ± 63 ± 7.84 66.30 ± 7.51 0.362# [0.539–0.988], p = 0.041). This model, including DPP7, showed better 34.19 PALS (%) 25.76 ± 29.89 ± 14.06 20.59 ± 10.00 0.134# fit according to the likelihood ratio test (χ2 = 6.001, p = 0.014), and the 12.98 AUC increased, although the DeLong test was not significant (0.657 LAVI (ml/m2) 31 30 (27–34) 34 0.815$ [0.484–0.831] vs 0.756 [0.608–0.904], p = 0.155). The models were (27–37) (22.5–37.75) built with 51 complete cases with all the variables available (17 AF and Number of AF 28 (7–42.5) episodes 34 no AF) (Supplemental Fig. 2). Longest AF 780.48 episode (min) (121.25–1510.71) 3.2. Differential protein expression and altered pathways in atrial COPD, chronic obstructive pulmonary disease; NIHSS, National Institutes of myopathy Health Stroke Scale; LVEF, left ventricular ejection fraction; PALS, peak atrial longitudinal strain; LAVI, left atrial volume index. 3.2.1. LAVI The “heart disease” terminology included any cardiopathy that the investigator Fifty-seven proteins correlated significantly with LAVI (40 negative considered of interest, including into this category ischemic cardiopathy, and correlations and 17 positive correlations). From these, only 17 had a hypertensive cardiopathy, between others. The two patients with heart disease in this cohort had a mild mitral and aortic valvulopathy, and a hypertensive correlation coefficient r>|0.6|. No proteins remained significant after cardiomyopathy respectively. multiple corrections in any of the analyses (Supplemental Table 4 and # Student’s t-test. Supplemental Fig. 3). $ Mann–Whitney test. The pathway analysis, considering the protein list ordered by the & 2 x test. correlation coefficient between protein levels and LAVI, revealed only 3 E. Palà et al. IJC Heart & Vasculature 39 (2022) 100977 Fig. 2. Boxplot distribution of CCDC80, DPP7, BMP1 and Cystatin-D between AF and no AF. P-values indicated correspond to the Mann-Whitney test. P-values were not corrected by Bonferroni, which lowered the significance threshold to p < 0.0125. Boxes extend from the 25th to 75th percentiles. The line in the middle is plotted as the median. Whiskers are drawn according to Tukey methodology (±1.5 IQR), and larger values are plotted as individual points. one gene set upregulated with LAVI with a Q-value < 0.25: Netrin-1 (117 negative correlations and 153 positive correlations). From these, signalling (R-HSA-373752). only 64 had a correlation coefficient r>|0.6|, and six remained signifi­ cant after multiple comparison corrections (Supplemental Table 5 and 3.2.2. PALS Supplemental Figure 4). Two-hundred-seventy proteins correlated significantly with PALS When proteins were ordered by the correlation coefficient between Fig. 3. Summary of proteins altered in AF and atrial myopathy.The numbers in the circles indicate proteins related to each variable (AF, PALS, and LAVI). Proteins in the overlap circles are those commonly related to two or three var­ iables. The number of proteins related to PALS and LAVI shown in the figure corresponds to those with significant correlations. The number of proteins related to AF corresponds to those sig­ nificant in the comparison AF vs no AF. In all cases, nominal p-values were considered. 4 E. Palà et al. IJC Heart & Vasculature 39 (2022) 100977 protein levels and PALS, 264 gene sets were enriched in patients with reduced sample size. Therefore, the use of DPP7 as a biomarker should high PALS, and two were enriched in patients with low PALS (R-HSA- be validated in further studies. The fact that the other candidates were 1630316, Glycosaminoglycan metabolism, and R-HSA-3781865, Dis­ not validated indicates that the present discovery approach resulted in eases of Glycosylation). Dysregulated pathways were mostly involved in several false positives. We cannot discard the possibility that the lack of signal transduction, metabolism, immune response, and hemostasis. validation is due to disparity in results between the two techniques used, (Supplemental Table 6 and Supplemental Figure 5). as previous studies presented weak correlations (r < 0.3) in some pro­ teins between SOMAscan platform and conventional immunoassays 3.3. Common protein expression in AF and atrial myopathy [24]. The low reproducibility between techniques may be explained as binding reagents interacting with different epitopes of a specific analyte. NT-proBNP was the only protein common in all the analyses per­ That is why discovery studies aiming to find biomarker candidates formed. Five proteins were related to LAVI and PALS (Leukotriene A-4 should include validations with different techniques. More importantly, hydrolase, Alpha-2-macroglobulin, Histone H2A type 3, Secretin, and the difficulty to find new AF biomarkers with “discovery” strategies may Fibroblast growth factor 12). One protein was associated with AF and be due to the complexity of AF, which makes it challenging to classify LAVI (Cell adhesion molecule 3). Fourteen proteins were associated with the patients in a binary variable. First, we cannot discard there were AF and PALS (Urokinase plasminogen activator surface receptor, false-negative cases in the no AF group as cardiac monitoring might not Lumican, Dermatopontin, Follistatin-related protein 3, C-C motif che­ have been prolonged enough to detect some AF cases [25]. Also, AF mokine 25, Plasminogen, Coagulation factor VII, Natural killer cell re­ could be just a bystander, and atrial dysfunction is the real underlying ceptor 2B4, Angiopoietin-2, Interleukin-23 receptor, Brain natriuretic disease that causes stroke susceptibility [26]. peptide 32, Erythropoietin, Advanced glycosylation end product- Consequently, there is a need to search for biomarkers that identify specific receptor, and Macrophage mannose receptor 1) (Fig. 3). not only AF but also the underlying atrial substrate. Therefore, and taking advantage of the availability of well-phenotyped patients by 4. Discussion echocardiography, we have explored circulating proteins and functional pathways that might be associated with atrial cardiomyopathy. Two In the present study, a wide variety of proteins have been explored in echocardiographic variables have been considered to perform this a cohort of cryptogenic stroke patients monitored for 1 month to detect analysis: LAVI and PALS. Although evaluating different characteristics AF. of the left atria, both variables are used to identify atrial myopathy and Although several studies had previously explored individual bio­ have been shown useful to predict AF [3]. LAVI measures the enlarge­ markers of cardioembolic stroke and/or AF in stroke patients [17], few ment of the atria while PALS provides information on its deformability, studies had used a discovery approach to identify new candidates but both variables are highly related. related to occult AF. In fact, the majority of AF “discovery” studies Interestingly, in our analysis, while many proteins and pathways enrolled asymptomatic patients [18–20] and, to our knowledge, only the were altered in patients with lower PALS, few were affected in patients study of Lambert et al. included stroke patients [21]. This study explored with high LAVI. This may reflect that more pathophysiological changes 184 proteins in patients with previously known causes of stroke. In are behind the remodelling that results in atrial distensibility impair­ contrast, we evaluated a wider range of proteins in patients with ESUS to ment compared to atrial enlargement. Previous studies stated that PALS ascertain new predictors of AF. Although all the patients included in our measurements were more sensitive than volumetric measurements to study had no prior story of AF, we cannot differentiate between new- predict AF and that LA dysfunction may precede and/or be independent onset AF or first-diagnosed AF. However, we aimed to identify AF of anatomical changes like LA enlargement [27]. This may also explain cases in cryptogenic stroke patients independently of the onset of this AF why few protein candidates are associated with both variables in our because, from a clinical perspective, the treatment of the arrhythmia is analysis. the same. The bioinformatics analysis revealed some interesting pathways Regarding the proteins differentially expressed in our analysis when playing a role in AF and atrial cardiomyopathy. The study highlighted comparing patients with and without AF, we should highlight that the importance of the inflammatory response in AF, especially the role of natriuretic peptides, NT-proBNP and BNP, were in the top-ranked po­ lymphocytes and chemokine signalling. This confirms the role of sitions. Both are well-known surrogates of AF and have already been inflammation in the initiation and maintenance of AF, which has validated in the present cohort of cryptogenic patients [8]. Therefore, extensively been described [28]. Also, we found an enrichment of we decided to test other proteins with similar nominal p-values in the growth factors signalling, which can also be linked to the inflammatory comparison. Only DPP7 showed significant differences in the validation response or the fibrosis occurring in the heart [29]. Finally, calcium from the four proteins selected, with higher levels in the patients handling is essential for the correct electrical functioning of the heart, without AF. DPP7 (also called DPP2) is a member of the dipeptidyl and similarly of what we found, abnormal calcium handling can lead to peptidase family, which has been implicated in many immunologic cardiac arrhythmias like AF [30]. processes. Although limited data about DPP7 is available compared to On the other hand, several pathways were downregulated in the AF other family members, there is a substantial overlap in this family group compared to the no AF group, which could be related to other regarding substrate specificity, inhibitors, and functions. stroke etiologies in the second group. This is the case of interleukin-12 Interestingly, members of the dipeptidyl peptidase family have been (IL-12) signalling. Interleukin-12 family members have been related to linked to atherosclerosis and are considered therapeutic targets for various cardiovascular diseases, including atherosclerosis, hyperten­ treating this pathology [22]. Therefore, the increase of DPP7 levels in no sion, aortic dissection, and several cardiac pathologies [31]. Also, in AF individuals might be linked to an underlying atherosclerotic aeti­ patients without AF, we found an upregulation of vesicle-mediated ology in those ESUS cases, but this should be confirmed. intracellular transport and activity of scavenger receptors, which are Another hypothesis would be that DPP7 had a role in the patho­ membrane-bound receptors that bind a variety of ligands, including low- physiology of AF, lowering its risk. DPP7 is a lysosomal protein ubiq­ density-lipoproteins (LDL). These, together with the transport of small uitously distributed in various tissues and organs, including the heart molecules, are pathways that might describe increased levels of LDL, [23]. Its pathophysiological significance requires additional study, involved in atherosclerosis, in patients without AF. including measurements on atrial tissue and/or biological experiments. Although DPP7 adds information to current markers (clinical, 4.1. Study limitations echocardiographic, and blood biomarkers previously described like NT- proBNP), the logistic regression models presented here were built with The present study has several limitations. The main limitation is the 5 E. Palà et al. IJC Heart & Vasculature 39 (2022) 100977 small sample size, limiting our statistical power to correct multiple References comparisons. Second, the selection of patients was performed according to AF diagnosis, and then a posterior exploratory analysis according to [1] B. Freedman, T.S. Potpara, G.Y.H. Lip, Stroke prevention in atrial fibrillation, Lancet 388 (10046) (2016) 806–817, https://doi.org/10.1016/S0140-6736(16) echocardiographic variables was performed. As continuous variables, 31257-0. correlations between the echocardiographic variables and protein levels [2] R.B. Schnabel, K.G. Haeusler, J.S. Healey, et al., Searching for atrial fibrillation were assessed, but we should consider the difficulties of obtaining sig­ poststroke: a white paper of the AF-SCREEN International Collaboration, Circulation 140 (2019) 1834–50. doi:10.1161/CIRCULATIONAHA.119.040267. nificant correlations with a reduced sample size. The same reason pre­ [3] J.J. Goldberger, R. Arora, D. Green, P. Greenland, D.C. Lee, D.M. Lloyd-Jones, cluded sex- and age-based analysis. Therefore, we cannot eliminate the M. Markl, J. Ng, S.J. Shah, Evaluating the atrial myopathy underlying atrial effect of age on the proteins/pathways revealed, especially in the case of fibrillation: Identifying the arrhythmogenic and thrombogenic substrate, Circulation 132 (4) (2015) 278–291. left atrial strain that correlates with age. Nevertheless, we may not be [4] M.S.V. Elkind, Atrial cardiopathy and stroke prevention, Curr. Cardiol. Rep. 20 interested in eliminating this effect as fibrosis of the atrium is a cumu­ (11) (2018), https://doi.org/10.1007/s11886-018-1053-0. lative process that may result from pathways altered with age [32]. [5] H. Kamel, W.T. Longstreth, D.L. Tirschwell, R.A. Kronmal, J.P. Broderick, Y. Y. Palesch, C. Meinzer, C. Dillon, I. Ewing, J.A. Spilker, M.R. Di Tullio, E.A. Hod, E. Similarly, LAVI is larger in women than in men in our cohort, and the Z. Soliman, S. Chaturvedi, C.S. Moy, S. Janis, M.SV. Elkind, The AtRial cardiopathy association of some proteins’ abundance with LAVI may reflect sex and antithrombotic drugs in prevention after cryptogenic stroke randomized trial: differences in our population. Therefore, these results need to be inter­ Rationale and methods, Int. J. Stroke 14 (2) (2019) 207–214, https://doi.org/ preted as hypotheses-generating. Selection of patients according to 10.1177/1747493018799981. [6] K. Jordan, S. Yaghi, A. Poppas, A.D. Chang, B. Mac Grory, S. Cutting, T. Burton, extreme PALS and LAVI matched by sex and age and validation of M. Jayaraman, G. Tsivgoulis, M.K. Sabeh, A.E. Merkler, H. Kamel, M.S.V. Elkind, altered markers in a cohort with a longer follow-up would provide K. Furie, C. Song, Left atrial volume index is associated with cardioembolic stroke further insights. Moreover, we should state that LAVI and PALS might and atrial fibrillation detection after embolic stroke of undetermined source, Stroke 50 (8) (2019) 1997–2001, https://doi.org/10.1161/STROKEAHA.119.025384. not be specific markers of atrial cardiomyopathy. Other parameters and [7] J. Pagola, T. González-Alujas, A. Flores, M. Muchada, D. Rodriguez-Luna, L. Seró, techniques can be used to characterize the disease in other studies (e.g., M. Rubiera, S. Boned, M. Ribó, J. Álvarez-Sabin, A. Evangelista, C.A. Molina, Left left atrial voltage mapping, late gadolinium-enhanced magnetic reso­ Atria Strain Is a Surrogate Marker for Detection of Atrial Fibrillation in Cryptogenic Strokes, Stroke 45 (8) (2014), https://doi.org/10.1161/STROKEAHA.114.005540. nance imaging, and histological analysis). [8] E. Palà, J. Pagola, J. Juega, et al., BNP over NT-proBNP to predict incident atrial Also, the lack of a matched set of similar aged non-stroke patients fibrillation after cryptogenic stroke, Eur. J. Neurol. 2020;ene.14579. doi:10.1111/ limits insight into biomarkers specifically associated with stroke. ene.14579. [9] J. Pagola, J. Juega, J. Francisco-Pascual, et al., Yield of atrial fibrillation detection Finally, we only analyzed a small portion of the full proteome, which with Textile Wearable Holter from the acute phase of stroke: Pilot study of Crypto- may have conditioned our findings. AF registry, Int. J. Cardiol. 251 (2018) 45–50. doi:10.1016/j.ijcard.2017.10.063. [10] J. Pagola, J. Juega, J. Francisco-Pascual, A. Bustamante, A. Penalba, E. Pala, M. Rodriguez, M. De Lera Alfonso, J.F. Arenillas, J.A. Cabezas, F. Moniche, 5. Conclusions R. Torres, J. Montaner, T. González-Alujas, J. Alvarez-Sabin, C.A. Molina, Large vessel occlusion is independently associated with atrial fibrillation detection, Eur. The present study revealed multiple proteins and pathways that may J. Neurol. 27 (8) (2020) 1618–1624, https://doi.org/10.1111/ene.14281. have a role in the development of atrial fibrillation. In particular, the [11] E. Donal, G.Y.H. Lip, M. Galderisi, A. Goette, D. Shah, M. Marwan, M. Lederlin, S. Mondillo, T. Edvardsen, M. Sitges, J. Grapsa, M. Garbi, R. Senior, A. Gimelli, T. role of DPP7 as a biomarker for stroke aetiology should be further S. Potpara, I.C. Van Gelder, B. Gorenek, P. Mabo, P. Lancellotti, K.-H. Kuck, B. explored. Also, we have proposed a strategy considering echocardio­ A. Popescu, G. Hindricks, G. Habib, B. Cosyns, V. Delgado, K.H. Haugaa, graphic parameters to discover new biomarkers and pathways that may D. Muraru, K. Nieman, A. Cohen, EACVI/EHRA Expert Consensus Document on the role of multi-modality imaging for the evaluation of patients with atrial fibrillation, have a role in atrial fibrillation and atrial myopathy. In this regard, the Eur. Heart J. Cardiovasc. Imaging 17 (4) (2016) 355–383, https://doi.org/ present study may be considered hypothesis-generating. 10.1093/ehjci/jev354. [12] F. Pathan, N. D’Elia, M.T. Nolan, T.H. Marwick, K. Negishi, Normal ranges of left atrial strain by speckle-tracking echocardiography: a systematic review and meta- Declaration of Competing Interest analysis, J. Am. Soc. Echocardiogr. 30 (1) (2017) 59–70.e8, https://doi.org/ 10.1016/j.echo.2016.09.007. The authors declare that they have no known competing financial [13] L. Gold, D. Ayers, J. Bertino, et al., Aptamer-based multiplexed proteomic technology for biomarker discovery, PLoS One 5 (2010). doi: 10.1371/journal. interests or personal relationships that could have appeared to influence pone.0015004. the work reported in this paper. [14] J. Candia, F. Cheung, Y. Kotliarov, G. Fantoni, B. Sellers, T. Griesman, J. Huang, S. Stuccio, A. Zingone, B.M. Ryan, J.S. Tsang, A. Biancotto, Assessment of variability in the SOMAscan assay, Sci. Rep. 7 (1) (2017), https://doi.org/ Acknowledgement 10.1038/s41598-017-14755-5. [15] G.K. Smyth, Linear models and empirical bayes methods for assessing differential This work was supported by the Instituto de Salud Carlos III, Spain expression in microarray experiments, Stat. Appl. Genet. Mol. Biol. 3 (2004). doi: [grant number PI15/02265 and PI18/00804]. The Spanish stroke 10.2202/1544-6115.1027. [16] J. Reimand, R. Isserlin, V. Voisin, M. Kucera, C. Tannus-Lopes, A. Rostamianfar, research network INVICTUS+ (RD16/0019/00021) of the Instituto de L. Wadi, M. Meyer, J. Wong, C. Xu, D. Merico, G.D. Bader, Pathway enrichment Salud Carlos III (co-financed by the European Regional Development analysis and visualization of omics data using g:Profiler, GSEA Cytoscape and Fund, FEDER) is also acknowledged. EP has received a predoctoral grant EnrichmentMap, Nat Protoc 14 (2) (2019) 482–517, https://doi.org/10.1038/ s41596-018-0103-9. from Vall D’Hebron Institute of Research. This project is supported by [17] A. Markus, S. Valerie, K. Mira, Promising biomarker candidates for cardioembolic AFFECT-EU, receiving funding from the European Union’s Horizon stroke etiology. A brief narrative review and current opinion, Front. Neurol. 12 2020 research and innovation program under grant agreement (2021) 1–10, https://doi.org/10.3389/fneur.2021.624930. [18] D. Ko, M.D. Benson, D. Ngo, Q. Yang, M.G. Larson, T.J. Wang, L. Trinquart, D. N◦ 847770. Sponsors did not have any role in the study design, analysis, D. McManus, S.A. Lubitz, P.T. Ellinor, R.S. Vasan, R.E. Gerszten, E.J. Benjamin, interpretation of data, report writing, or the decision to submit the H. Lin, Proteomics profiling and risk of new-onset atrial fibrillation: framingham article for publication. heart study, JAHA 8 (6) (2019), https://doi.org/10.1161/JAHA.118.010976. [19] L. Lind, J. Sundström, M. Stenemo, E. Hagström, J. Ärnlöv, Discovery of new biomarkers for atrial fibrillation using a custom-made proteomics chip, Heart 103 Appendix A. Supplementary material (5) (2017) 377–382, https://doi.org/10.1136/heartjnl-2016-309764. [20] W.Y. Zhong, H. Yang, Y.L. Wang, et al., Proteomic profiles of patients with atrial fibrillation provide candidate biomarkers for diagnosis, Int. J. Cardiol. 344 (2021) Supplementary data to this article can be found online at https://doi. 205–12. doi:10.1016/j.ijcard.2021.09.047. org/10.1016/j.ijcha.2022.100977. [21] A. Tancin Lambert, X. Kong, B. Ratajczak-Tretel, D. Atar, D. Russell, M. Skjelland, V. Bjerkeli, K. Skagen, M. Coq, E. Schordan, H. Firat, B. Halvorsen, A. Aamodt, Biomarkers associated with atrial fibrillation in patients with ischemic stroke: a pilot study from the NOR-FIB Study, Cerebrovasc Dis Extra 10 (1) (2020) 11–20, https://doi.org/10.1159/000504529. 6 E. Palà et al. IJC Heart & Vasculature 39 (2022) 100977 [22] Y. Waumans, L. Baerts, K. Kehoe, A.-M. Lambeir, I. De Meester, The dipeptidyl [28] Y.-F. Hu, Y.-J. Chen, Y.-J. Lin, S.-A. Chen, Inflammation and the pathogenesis of peptidase family, prolyl oligopeptidase and prolyl carboxypeptidase in the immune atrial fibrillation, Nat. Rev. Cardiol. 12 (4) (2015) 230–243, https://doi.org/ system and inflammatory disease, including atherosclerosis, Front. Immunol. 6 10.1038/nrcardio.2015.2. (2015), https://doi.org/10.3389/fimmu.2015.00387. [29] S. Nattel, Molecular and cellular mechanisms of atrial fibrosis in atrial fibrillation, [23] M.-B. Maes, S. Scharpé, I. De Meester, Dipeptidyl peptidase II (DPPII), a review, JACC Clin. Electrophysiol. 3 (5) (2017) 425–435, https://doi.org/10.1016/j. Clin. Chim. Acta 380 (1-2) (2007) 31–49, https://doi.org/10.1016/j. jacep.2017.03.002. cca.2007.01.024. [30] G.Y.H. Lip, L. Fauchier, S.B. Freedman, I. Van Gelder, A. Natale, C. Gianni, [24] L.M. Raffield, H. Dang, K.A. Pratte, et al., Comparison of proteomic assessment S. Nattel, T. Potpara, M. Rienstra, H.-F. Tse, D.A. Lane, Atrial fibrillation, Nat. Rev. methods in multiple cohort studies, Proteomics 20 (2020) 1–34. doi:10.1002/ Dis. Prim. 2 (1) (2016), https://doi.org/10.1038/nrdp.2016.16. pmic.201900278. [31] J. Ye, Y. Wang, Z. Wang, L. Liu, Z. Yang, M. Wang, Y. Xu, D.i. Ye, J. Zhang, Y. Lin, [25] T. Sanna, H.-C. Diener, R.S. Passman, V. Di Lazzaro, R.A. Bernstein, C.A. Morillo, Q. Ji, J. Wan, Roles and mechanisms of Interleukin-12 Family members in M.M. Rymer, V. Thijs, T. Rogers, F. Beckers, K. Lindborg, J. Brachmann, cardiovascular diseases: opportunities and challenges, Front. Pharmacol. 11 Cryptogenic stroke and underlying atrial fibrillation, N. Engl. J. Med. 370 (26) (2020), https://doi.org/10.3389/fphar.2020.00129. (2014) 2478–2486. [32] J. Pagola, J. Juega, J. Francisco-Pascual, A. Bustamante, A. Penalba, E. Pala, [26] B. Freedman, H. Kamel, I.C. Van Gelder, R.B. Schnabel, Atrial fibrillation: Villain or M. Rodriguez, M. De Lera-Alfonso, J.F. Arenillas, J.A. Cabezas, F. Moniche, R. de bystander in vascular brain injury, Eur. Hear J.Suppl 22 (Supplement_M) (2020) Torres, J. Montaner, T. González-Alujas, J. Alvarez-Sabin, C.A. Molina, Predicting M51–M59. atrial fibrillation with high risk of embolization with atrial strain and NT- proBNP, [27] G.C.H. Gan, A. Ferkh, A. Boyd, L. Thomas, Left atrial function: evaluation by strain Transl. Stroke Res. 12 (5) (2021) 735–741, https://doi.org/10.1007/s12975-020- analysis, Cardiovasc. Diagn. Ther. 8 (1) (2018) 29–46, https://doi.org/10.21037/ 00873-2. cdt.2017.06.08. 7