Author + information
- Received February 17, 2012
- Revision received May 28, 2012
- Accepted June 19, 2012
- Published online October 30, 2012.
- Michael F. Swartz, PhD⁎,
- Gregory W. Fink, MD†,
- Muhammad F. Sarwar, MD‡,
- George L. Hicks, MD⁎,
- Yao Yu, PhD§,
- Rui Hu, PhD§,
- Charles J. Lutz, MD†,
- Steven M. Taffet, PhD∥ and
- José Jalife, MD¶,⁎ ()
- ↵⁎Reprint requests and correspondence:
Dr. José Jalife, Center for Arrhythmia Research, Division of Cardiovascular Medicine, Department of Internal Medicine, University of Michigan, 5025 Venture Drive, Ann Arbor, Michigan 48108
Objectives This study sought to determine if serum markers for collagen I and III synthesis, the carboxyl terminal peptide from pro-collagen I (PICP) and the amino terminal peptide from pro-collagen III (PIIINP), correlate with left atrial (LA) fibrosis and post-operative atrial fibrillation (AF).
Background AF after cardiac surgery is associated with adverse outcomes. We recently demonstrated that LA fibrosis is associated with post-operative AF in patients with no previous history of AF.
Methods Fifty-four patients having cardiac surgery without a history of AF consented to left and right atrial biopsies and a pre-operative peripheral blood draw. Picrosirius red staining quantified the percentage of fibrosis, and reverse transcriptase polymerase chain reaction assessed atrial tissue messenger ribonucleic acid transcripts involved in the fibrosis pathway. PICP and PIIINP levels were measured using an enzyme immunosorbent assay.
Results Eighteen patients developed AF, whereas 36 remained in normal sinus rhythm. LA fibrosis was higher in patients who developed AF versus normal sinus rhythm (6.13 ± 2.9% vs. 2.03 ± 1.9%, p = 0.03). LA messenger ribonucleic acid transcripts for collagen I, III, transforming growth factor, and angiotensin were 1.5- to 2.0-fold higher in AF patients. Serum PICP and PIIINP levels were highest in AF versus normal sinus rhythm (PICP: 451.7 ± 200 ng/ml vs. 293.3 ± 114 ng/ml, p = 0.006; PIIINP: 379 ± 286 pg/ml vs. 191.6 ± 162 pg/ml, p = 0.01). Furthermore, there was a linear correlation between LA fibrosis and serum PICP levels (R2 = 0.2; p = 0.01), and of the markers, only PICP was independently associated with AF.
Conclusions This demonstrates that serum PICP and PIIINP levels correlate with the presence of LA fibrosis and may act as predictors for post-operative AF even in the absence of previous history of AF.
Atrial fibrillation (AF) occurs in ∼30% of patients following cardiac surgery and contributes significantly to the post-operative morbidity, mortality, and costs (1). We recently demonstrated that post-surgical AF was maintained by left atrial (LA) sources and associated with pre-operative fibrosis (2). We hypothesized that left ventricular hypertrophy and diastolic dysfunction contribute to LA dilation and stretch promoting extracellular matrix remodeling, manifested in elevated serum and tissue markers for fibrosis aiding to the prediction of post-operative AF, in patients who have no previous history of AF.
Fibrosis in the atria primarily consists of types I and III collagen, which are synthesized by fibroblasts and myofibroblasts, and is regulated by a cascade of fibro-proliferative signals including transforming growth factor (TGF) beta and angiotensin II (ANG) working in concert to increase collagen secretion (3). Collagen I and III are secreted as pro-collagen precursors containing amino- and carboxyl-terminal pro-peptides (4) that are released into the serum by proteases after collagen deposition. Whereas these pro-peptides correlate with the rate of collagen synthesis (4), type I collagen degradation results in the release of a carboxyl terminal telopeptide of collagen I (CITP) (4–6). Previous investigators have analyzed levels of the carboxyl terminal peptide from pro-collagen I (PICP), CITP, and the amino terminal peptide from pro-collagen type III (PIIINP) to monitor ventricular fibrosis during hypertension and congestive heart failure (4–6). However, the utility of these peptides as markers to predict a collagen substrate necessary for the development of AF is controversial (7,8) and has not been determined in patients undergoing cardiothoracic surgery.
We hypothesized that patients without a history of arrhythmias, who developed AF following cardiac surgery have: 1) left ventricular hypertrophy and diastolic dysfunction resulting in LA stretch reflective of an increased LA diameter; 2) increased LA messenger ribonucleic acid (mRNA) expression of collagen I and III, TGF beta, and ANG; 3) an increased percentage of LA fibrosis; 4) higher serum levels of PICP and PIIINP. We also hypothesize that serum markers for collagen synthesis would correlate with the percentage of LA fibrosis.
After approval from the Institutional Review Board, 54 patients scheduled for cardiac surgery consented to atrial biopsies and evaluation of PICP, PIIINP, and CITP levels. Medical records were reviewed to identify any of the exclusion criteria: <18 years of age, AF history, previous sternotomy, history of hyperthyroidism, pulmonary fibrosis, cirrhosis, or scleroderma. In addition, all patients were interviewed prior to the operation to verify that they had no known history of AF. All patients were monitored by continuous telemetry throughout their hospitalization for the presence of AF and were treated following the Society of Thoracic Surgery guidelines after an open heart procedure (1). AF was defined as an irregularly irregular rhythm without P waves for a period of at least 15 min as observed by a physician (2). All patients who developed AF were given full disclosure and treated following the Society of Thoracic Surgery guidelines (1,2).
Pre-operative electrocardiograms were obtained on the day of surgery, magnified to 6,400× and the duration from the onset of the P-wave to the return of the isoelectric line was determined by 1 blinded investigator. Pre-operative echocardiograms were obtained in all patients. Detailed methods regarding the echocardiogram measurements can be found in the data supplement in the Online Appendix.
We obtained 3- to 5-mm biopsies from the LA and right atrial (RA) appendages. The biopsies were divided and placed in 10% buffered formalin and RNAlater (Ambion, Austin, Texas). RNA was extracted, transcribed, and mRNA levels for the gene of interest were normalized to 18S ribosomal RNA (9,10). Further details regarding the mRNA methods can be found in Online Table 1.
Specimens were embedded in paraffin and sectioned at 6 μm. Slides were stained with picrosirius red as previously described (2,11) and examined with a Bioquant imaging system (Bioquant Image Analysis Corporation, Nashville, Tennessee) to quantify the area of fibrosis and atrial muscle. Epicardial, endocardial, and perivascular fibrosis were excluded from the analysis. The percentage of fibrosis within the atrial muscle from each sample was calculated. All samples were measured in triplicate by a single investigator.
PICP, CITP, and PIIINP
Serum from all 54 patients was obtained pre-operatively as well as from 10 healthy subjects with no known cardiac disease. All samples were obtained from a peripheral vein, the serum extracted, and stored at –80°C. PICP (Takara Bio Inc., Shiga, Japan), CITP (Orion Diagnostica, Espoo, Finland), and PIIINP (Uscn Life Sciences Inc., Houston, Texas) levels were measured by enzyme-linked immunosorbent assay according the manufacturers' specifications and based on previous data (12,13). The sensitivity (lower detection limit) for each assay was 10 ng/ml for PICP, 1.0 μg/l for CITP, and 20 pg/ml for PIIINP, and the intra-assay variations were 6.0% for PICP, 3.3% for CITP, and 1.8% for PIIINP. All samples were run in duplicate and measured at 450λ.
For complete statistical methods see the Online Appendix.
Of the 54 patients enrolled, 36 remained in normal sinus rhythm (NSR) and 18 developed new onset post-operative AF. Patient variables (Table 1) demonstrated increased age as the only factor associated with AF development. From the pre-operative echocardiography data shown in Table 2, increased LA diameter, greater than mild valve disease, the number of patients with class II or III diastolic dysfunction, and increased left ventricular mass index were all associated with post-operative AF. Details regarding the timing and duration of AF episodes can be found in Online Table 2.
Left and right atrial biopsies
Of the 54 patients enrolled, there were 10 patients in whom biopsies were not analyzed. In 3 patients, there were no observable LA appendages, and therefore biopsies were not obtained. In 2 patients, inspection of the LA or RA biopsy demonstrated only fat and an absence of muscle, and these biopsies were not analyzed. Lastly, in 5 patients, the atrial appendage was thought to be excessively friable, thin, and at risk for post-operative bleeding and therefore, a biopsy was not taken.
Reverse transcriptase polymerase chain reaction
Patient characteristics from the 44 patients who had biopsies are presented in Online Table 3. The expression of 18S ribosomal RNA was similar in the left and right atria of all patients (Online Table 4). Comparisons in the expression patterns of the mRNA transcripts from the entire patient population between LA and RA samples are shown in Online Table 5. The only significant difference was a 9.1-fold increase in tissue inhibitor or metalloproteinase 1 expression from LA samples (p = 0.002). The mRNA expression of relevant transcripts by delta threshold cycle (cT) values between those patients who remained in NSR and those who developed AF are provided in Figures 1A and 1B. All of the transcripts analyzed involved in structural remodeling had a mean increase in expression within the LA of patients who developed AF compared to the LA of patients who remained in NSR. To gain a greater appreciation for the relevance in differences between delta cT values that were statistically significant, the delta cT values were converted to a percent-fold change (Fig. 2). Collagen I and III as well as TGF beta and ANG were all significantly increased by nearly 2.0-fold in LA samples from those patients who developed AF.
Fibrosis and AF
Interstitial fibrosis manifested as winding strands within the atrial muscle was observed in all LA and RA biopsies. There were no significant differences between the LA and RA biopsies in the area of tissue available for analysis (LA: 3.7 ± 2.8 mm2 vs. RA: 3.0 ± 2.3 mm2, p = 0.12), or the overall percentage of fibrosis between LA and RA samples (3.1 ± 2.8% vs. 2.9 ± 2.4%, p = 0.63). There was no difference in RA fibrosis between patients who developed post-operative AF and those who remained in NSR (2.83 ± 2.3% vs. 2.94 ± 2.4%, p = 0.4). However, as shown in Figure 3, the percentage of area occupied by fibrosis within each sample was significantly higher in the LA samples of those patients who developed post-operative AF (6.13 ± 2.4% vs. 2.03 ± 1.9%, p = 0.03). These results are consistent with data published previously (2).
Serum markers of collagen synthesis/degradation and AF
Serum PICP, PIIINP, and CITP levels were obtained from 54 patients and 10 control subjects (Fig. 4A). The 10 control subjects had no history of cardiac disease, were annually followed by a healthcare provider, and were currently not taking any medications (Online Table 6). PICP levels in patients who developed post-operative AF (451.7 ± 200 ng/ml) were significantly higher than those who remained in NSR (293.3 ± 114, p= 0.006) and those without a known history of cardiac disease (260.4 ± 66, p = 0.001) (Fig. 4B). Figure 4C provides a histogram in which patient values have been placed in 50 ng/ml bins of PICP and demonstrates that the majority of patients who developed AF had higher PICP values than did those who remained in NSR. Furthermore, Figure 4D presents a receiver-operating characteristic (ROC) curve comparing the development of AF and PICP and demonstrates an overall area under the curve of 0.75.
Similar to PICP levels, CITP levels were lowest in the healthy control group (4.1 ± 2.2 μg/l) compared with patients who underwent cardiac surgery (23.7 ± 9.7 μg/l, p = 0.002). However, there was no significant difference between patients who developed AF (22.8 ± 11.7 μg/l) and those who remained in NSR (24.1 ± 8.7μg/l) (Fig. 5A).
Higher PIIINP levels also correlated with the development of post-operative AF. As shown in Figure 5B, PIIINP levels in patients who developed AF were greater than in those who remained in NSR (AF: 379 ± 286 pg/ml vs. NSR: 191.6 ± 162 pg/ml, p = 0.01). Unlike PICP and CITP levels, there was no significant difference between patients having surgery and the healthy control subjects (238.4 ± 107, p = 0.2). Similar to PICP, the histogram in Figure 5C represents 50 pg/ml bins of PIIINP and demonstrates a lower utility when compared with PICP in differentiating between patients who developed post-operative AF and those who remained in NSR. In Figure 5D, the ROC curve further validates this point demonstrating an area under the curve of 0.64.
Atrial fibrosis correlates with serum peptide levels
To determine whether the systemic synthesis and degradation of collagen I and III correlates with the level of atrial fibrosis, we attempted to match the percentage of LA and RA fibrosis with the serum PICP, CITP, and PIIINP levels. As demonstrated in Figure 6, LA fibrosis did correlate significantly with PICP levels (R2 = 0.2, p = 0.01). However, we found no correlation between the percentage of RA fibrosis and PICP values (R2 = 0.04, p = 0.3). In addition, no significant correlations were demonstrable between the percentage of LA or RA fibrosis and CITP or PIIINP levels.
Serum peptide levels and patient variables
We performed a multivariate analysis to determine whether the serum markers contribute significantly to the development of post-operative AF over and above patient variables such as age, which is an important predictor for the development of AF. Only age and PICP levels provided a significant contribution to the development of post-operative AF over other patient variables (age: odds ratio: 1.13, p = 0.01; PICP: odds ratio: 1.01, p = 0.02).
As age was the most important patient variable in the development of AF, we constructed ROC curves comparing individual PICP and PIIINP levels to age, as well as adjusting PICP and PIIINP levels to age. Online Figures 1 and 2 demonstrate that the areas under the curve for PICP, PIIINP, and age were 0.74, 0.67, and 0.77, respectively. Construction of ROC curves incorporating PICP along with age was 0.84, whereas the area under the curve for PIIINP along with age was 0.81.
The main findings of this study are as follows: Patients without a history of AF who develop post-operative AF have: 1) increased left ventricular mass index and LA diameter measured from pre-operative echocardiograms; 2) LA with increased levels of collagen I and III, TGF beta, and ANG expression demonstrated by mRNA transcripts; 3) increased percentage of LA fibrosis from within the LA appendage; 4) increased serum PICP and PIIINP levels that correlate positively with the development of AF in the post-operative phase of cardiac surgery. Furthermore, in patients who developed post-operative AF, serum PICP levels correlate with the percentage of LA but not RA fibrosis.
Among all patient variables, age was the most important factor in the development of post-operative AF. The accumulation of fibrosis and collagen in the atria resulting from hypertension, valvular disease, and diastolic dysfunction provide a likely mechanism for this correlation (14). Diastolic dysfunction has been shown to correlate with LA stretch and collagen deposition (5,6). Left ventricular mass index quantifies left ventricular hypertrophy and diastolic dysfunction and is increased in our group of patients who developed post-operative AF, which, along with LA diameter, is an indicator of stretch.
Analysis of gene expression patterns in patients who developed AF demonstrated increased pro-fibrotic markers in LA tissue. Within the entire population, there was no difference in collagen I and III expression between LA and RA samples. However, in patients who developed post-operative AF, collagen I and III were increased in the LA and may further validate the hypothesis that increased LA pressures associated with hypertrophy and diastolic dysfunction result in atrial fibrosis and post-operative AF risk (1,2). Of the pro-fibrotic markers analyzed, only TGF beta and ANG were significantly increased in the LA tissue of patients who developed post-surgical AF. These 2 markers are integral to collagen synthesis (3) and provide a molecular mechanism for the increased percentage of LA fibrosis in AF patients. Interestingly, these transcripts were increased only in the LA; however, LA and RA transcript differences have been previously reported (2,15,16). Microarray analysis demonstrated 624 genes, 41% of which were expressed higher within the LA than in the RA myocardium (15). Furthermore, an increased expression of the angiotensin receptors has been found within the LA myocardium of patients who develop AF (16). Our data further supports previous work that demonstrates LA and RA differences (2).
The importance of LA fibrosis as a substrate in the development of lone AF is well documented (2,9,11). The accumulation of fibrosis within the extracellular space prolongs conduction time between myocytes creating conditions favorable for the initiation and maintenance of reentry (11). Yet despite the acceptance of fibrosis as a substrate for the development of lone AF, there is controversy regarding its importance in the development of post-operative AF (2,17). Previous data regarding the percentage of RA fibrosis in the development of post-surgical AF has produced mixed results (17). To date there is a preponderance of evidence (18) to suggest that in the majority of AF cases, the RA is merely a bystander, whereas electrical sources in the LA are integral to the mechanism of AF maintenance (2). Our data corroborate previous work (2), demonstrating that only differences in fibrosis within the LA and not the RA correlate with the development of post-operative AF. Therefore, it is reasonable to conclude that LA fibrosis may have a greater impact on AF initiation and maintenance than RA fibrosis would (2).
Markers of collagen I synthesis (PICP) and degradation (CITP) have been compared in AF patients, typically in those who undergo catheter ablation (19,20). However, differences in serum peptide levels were compared between AF patients and healthy control subjects (20). We demonstrate that there were differences in PICP and CITP levels in all surgical patients versus healthy control subjects, as well as a significant difference between PICP levels in patients who developed and AF and those who remained in NSR, all of whom had cardiovascular disease but no previous history of AF.
PICP levels demonstrated the most significant difference among patients who developed AF, those who remained in NSR, and control patients without a history of cardiac disease. This is in contrast with a previous report that demonstrated no difference in pro-collagen I levels between patients with AF and matched control patients with a similar cardiac history but who were in NSR (8). Matching patients with a similar cardiac history may negate the utility in separating patients with fibrosis. Importantly, the stoichiometric ratio between the number of collagen molecules produced, and the number of PICP molecules released is 1:1 (4). As a result, PICP levels have been shown to correlate in a linear fashion with the percentage of ventricular fibrosis (R2 = 0.4) in patients with hypertensive heart disease (5). In patients having heart surgery, we demonstrate a similar correlation (R2 = 0.2) between the percentage of LA fibrosis and serum PICP levels. Such a correlation serves to further validate the importance of LA fibrosis in the development of post-operative AF.
The use of PIIINP as a marker of collagen III synthesis remains controversial. Unlike PICP levels, which have a stoichiometric ratio of 1:1 with collagen deposition, PIIINP levels are less reliable (4). This may account in part for why when compared with PICP differences, PIIINP levels in patients who remained in NSR and who developed AF did not correlate with the percentage of LA or RA fibrosis.
Although all patients had no known history of AF, some could have had prior episodes of asymptomatic AF, as pre-operative Holter monitoring was not performed. The areas of LA and RA tissues analyzed in our study were relatively small in comparison to the overall size of each chamber, and we have made inferences based on differences only from a small sample. However, each biopsy was taken from the same location on the LA or RA. Therefore, they are useful points of comparison. Ethical considerations prevented us from taking larger samples or samples from multiple locations, which would have given additional information but placed the patient at increased risk for post-operative bleeding. Therefore, tissue for western blots for protein analysis was regrettably not available.
We have demonstrated the importance of fibrosis as a substrate in the development of post-operative AF. It is well known that AF is one of the most frequent complications of cardiac surgery. AF affects more than one-third of patients and is associated with increased morbidity and mortality and longer, more expensive hospital stays. As such, the ability to identify which patients have a pre-existing substrate, and are therefore at high risk for developing AF, could be critical for improved post-operative outcomes.
The authors would like to thank the Clinical Translational Science Institute at the University of Rochester for statistical assistance.
For detailed methods regarding the echocardiogram measurements and supplemental data, please see the online version of this article.
Supported by grants from the National Heart, Lung, and Blood Institute (NHLBI P01-HL39707) and the Leducq Foundation (to Dr. Jalife). Dr. Jalife is a consultant for Topera, Inc.
All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- atrial fibrillation
- angiotensin II
- carboxyl terminal telopeptide from collagen I
- threshold cycle
- left atrial/atrium
- messenger ribonucleic acid
- normal sinus rhythm
- carboxyl terminal peptide from pro-collagen I
- amino terminal peptide from pro-collagen II
- right atrial/atrium
- receiver-operating characteristic
- transforming growth factor
- Received February 17, 2012.
- Revision received May 28, 2012.
- Accepted June 19, 2012.
- American College of Cardiology Foundation
- Burstein B.,
- Nattel S.
- Querejeta R.,
- Varo N.,
- López B.,
- et al.
- López B.,
- González A.,
- Beaumont J.,
- Querejeta R.,
- Larman M.,
- Díez J.
- Tziakas D.N.,
- Chalikias G.K.,
- Papanas N.,
- Stakos D.A.,
- Chatzikyriakou S.V.,
- Maltezos E.
- Cardin S.,
- Libby E.,
- Pelletier P.,
- et al.
- Epperson L.E.,
- Martin S.L.
- Tanaka K.,
- Zlochivier S.,
- Vikstrom K.,
- et al.
- Martos R.,
- Baugh J.,
- Ledwidge M.,
- et al.
- Berruti A.,
- Dogliotti L.,
- Gorzeonon G.,
- et al.
- Anyukhovsky E.P.,
- Sosunov E.A.,
- Plotnikov A.,
- et al.
- Boldt A.,
- Wetzel U.,
- Weigl J.,
- et al.
- Cosgrave J.,
- Foley J.B.,
- Flavin R.,
- et al.
- Okumura Y.,
- Watanabe I.,
- Nakai T.,
- et al.
- Kallergis E.M.,
- Manios E.G.,
- Kanoupakis E.M.,
- et al.