Fragmented QRS complex in patients with systemic lupus erythematosus at the time of diagnosis and its relationship with disease activity

Objective Cardiovascular disease is an important contributor to the mortality rate of patients with systemic lupus erythematosus (SLE), which is related to SLE disease activity. Fragmented QRS (fQRS) complexes, defined by additional spikes in the QRS complex, are useful for identifying myocardial scars on electrocardiography and can be an independent predictor of cardiac events. We aimed to assess the relationship between disease activity in patients with SLE and fQRS at the time of diagnosis. Methods Forty-four patients with SLE were included. Patients with cardiac diseases, other rheumatic diseases, and prior treatment at the time of electrocardiography measurement were excluded. The appearance of fQRS represented exposure. The primary outcome was SLE Disease Activity Index 2000 (SLEDAI-2K). Multiple regression analysis was conducted to assess the association between fQRS and SLEDAI-2K adjusted for age, sex, and time from the estimated onset date to the date of diagnosis. Results Among patients with SLE at diagnosis, 26 (59.1%) had fQRS. The median SLEDAI-2K was 18 (interquartile range [IQR], 12–22) and 9 (IQR, 8–15) in the fQRS(+) and fQRS(-) groups, respectively. SLEDAI-2K was significantly higher in the fQRS(+) group than in the fQRS(-) group (regression coefficient, 2.69; 95% confidence interval, 0.76–4.61; p = 0.008). Conclusion Our results suggested that fQRS(+) patients with SLE had high disease activity. fQRS could likely detect subclinical myocardial involvement in patients with SLE and predict long-term occurrence of cardiac events.

Introduction who were aged >15 years, diagnosed with SLE, and underwent ECG at the time of diagnosis. Participants who satisfied at least 4 of the 11 American College of Rheumatology (ACR) criteria from 1997 were included. Patients who underwent treatment prior to ECG measurement and those with ischemic heart disease, severe valvular disease, congenital heart disease, cardiomyopathy, history of arrhythmia, hepatic failure, RA, SSc, and abnormal serum electrolytes were excluded. We used the process of sequential sampling as our sampling method. The study was conducted in accordance with the principles outlined in the Declaration of Helsinki and was approved by the ethical review committee of Showa University School of Medicine (approval numbers 2556). All patient information was anonymized and de-identified prior to analysis.

Data collection
The patients' demographic data including sex, age at the time of diagnosis, blood pressure, smoking status, and comorbid conditions such as hypertension, treatment for diabetes mellitus, and dyslipidemia were collected at the time of diagnosis. Patients were considered to have comorbid hypertension if they were using antihypertensive drugs such as diuretics, beta-blockers, calcium channel blockers, angiotensin-converting enzyme inhibitors, and angiotensin type II receptor blockers. Blood test results and urinalysis data at the time of the SLE diagnosis obtained at the time closest to the ECG measurement and before treatment intervention were included in the analysis. C-reactive protein, low-density lipoprotein cholesterol, high-density lipoprotein cholesterol, triglycerides, uric acid, glycated hemoglobin (HbA1c, measured according to the National Glycohemoglobin Standardization Program), complement (hemolytic complement activity, complement 3, complement 4), C1q-binding immune complexes (IC-C1q), anti-dsDNA antibody, anti-β2-glycoprotein I antibody, anti-SS-A/Ro antibody, and anti-U1-RNP antibody levels were investigated, and urinalysis (proteinuria, casts, hematuria, and pyuria analysis) was performed. The Framingham Risk Score used to estimate the 10-year risk for developing coronary heart disease was evaluated [23]. SLE Disease Activity Index 2000 (SLE-DAI-2K) at the time of the ECG measurement was evaluated by the attending rheumatologist. Morbidities with end-organ involvement including cutaneous manifestations, arthritis, myositis, pericarditis, pleuritis, lupus enteritis, lupus cystitis, vasculitis, renal disorders, neurologic disorders, and hematologic disorders (leukopenia, thrombocytopenia) were documented. Renal disorders were defined as any of the following: renal biopsy indicating lupus nephritis, nephrotic syndrome, increase in serum creatinine level >1.5-times the baseline, proteinuria, urinary casts, hematuria, or pyuria in the absence of other causes. Lupus enteritis was defined according to colonoscopy findings, computed tomography findings, or patient reports of abdominal somatic pain or hematochezia in the absence of other causes. Lupus cystitis was defined by patient reports of bowel or urinary symptoms, or as hydronephrosis in the absence of other causes. Other disease manifestations were defined according to SLEDAI-2K definitions. The estimated date of onset was defined as the day when one or more ACR classification criteria items were reported. Results of TTE examinations performed at the time of the SLE diagnosis were evaluated. Left ventricular (LV) end-diastolic and end-systolic dimensions were measured in the parasternal long-axis view with the M-mode cursor positioned appropriately. The LV ejection fraction (EF) was measured in accordance with the Simpson's method. The right ventricle systolic pressure (RVSP) was calculated using the tricuspid regurgitant velocity.
Model ECG 2550; Nihon Kohden, Tokyo, Japan) were recorded before initiating drug therapy. In fQRS (+) patients, the appearance of fQRS was evaluated again on electrocardiogram after immunosuppressive therapy. All ECGs were evaluated at 400% magnification by 2 experienced cardiologists who were blinded to patient characteristics and outcomes. Differences in ECG readings were discussed until an agreement was reached. The fQRS complex was defined by the presence of an additional R wave (R') or notching in the nadir of the R wave or the S wave or >1 R' (fragmentation) in 2 contiguous leads corresponding to the territory of a major coronary artery during a normal QRS interval (Fig 1, S1 Fig). Complete bundle branch block (BBB) patterns (QRS �120 ms) and incomplete right BBBs were excluded. 11

Outcome measures
The primary outcome was disease activity. SLEDAI-2K, a common method of evaluating disease activity, was used at the time of ECG measurement by an attending rheumatologist who was blinded to the ECG findings. Secondary outcomes included complement and anti-dsDNA antibody levels and end-organ involvement.

Statistical analysis
Categorical data are described as numbers with proportions (%) and were compared using Fisher's exact test. Continuous data are expressed as means with standard deviations (SD) or as medians with interquartile ranges (IQR), as appropriate, and were compared using the Wilcoxon signed-rank test. Inter-observer variabilities were assessed using Cohen's kappa coefficient. During the main analysis, a multiple regression analysis was conducted to assess the association between fQRS and SLE activity after adjusting for age, sex, and period from the estimated date of onset to the date of diagnosis. During secondary analysis, we performed a multilinear regression analysis to examine the correlations between fQRS and serological markers related to SLE activity (complement and anti-dsDNA antibody levels) and a logistic regression analysis to examine the correlations between fQRS and end-organ involvement under the same conditions as described previously. We did not perform multivariate analyses of end-organ involvements that occurred less frequently. Three sensitivity analyses were performed. First, we analyzed all ECG findings interpreted by the 2 cardiovascular physicians (ET and JS) as exposure. Furthermore, inter-observer variabilities between the cardiologists were assessed. Next, the ECG results were evaluated and analyzed as the main outcome at 100% magnification on paper. Additionally, inter-observer variabilities were assessed at 400% and 100% magnifications and compared. Finally, we excluded patients with hypertension, treated diabetes mellitus, and treated dyslipidemia because of their cardiovascular risks and analyzed the main outcome as the result. Missing data were not imputed. All statistical tests were 2-sided, and significance was defined as p<0.05. Analyses were performed using JMP1 Pro, version 14.0.0. (SAS Institute Inc., Cary, NC, USA).
During the main analysis, the regression coefficient of fQRS for SLEDAI-2K was 2.69 (95% confidence interval [CI], 0.76-4.61; p = 0.008) in reference to fQRS(-) ( Table 2). During secondary analysis, the regression coefficient of fQRS for nephritis was 0.94 (95% CI, 0.28-1.74; p = 0.014) in reference to fQRS(-). There were no significant associations between fQRS and the blood tests or evidence of end-organ involvement other than nephritis (Tables 3 and 4). According to the first sensitivity analysis, based on the ECG results determined by each of the cardiologists (ET and JS), SLEDAI-2K was significantly higher for the fQRS(+) group than for the fQRS(-) group. According to the evaluation by ET, 27 patients (61%) had fQRS, and the regression coefficient of fQRS for SLEDAI-2K was 2.80 (95% CI, 0.90-4.70; p = 0.005) in

Discussion
This study investigated the relationship between disease activity in patients with SLE and fQRS at diagnosis after adjustment for age, sex, and time period from the estimated onset date to the date of diagnosis. The fQRS(+) rate among patients with SLE at diagnosis was 59% and 33% of the fQRS(+) patients disappeared fQRS after immunosuppressive therapy. The SLEDAI-2K at diagnosis was significantly higher for the fQRS(+) group than for the fQRS(-) group. Additionally, the fQRS(+) group showed a high rate of nephritis. This study is the first report on an association between ECG abnormalities and disease activity at the time of diagnosis in patients with SLE. Therefore, fQRS should be used to detect subclinical myocardial involvement in patients with SLE. The fQRS can be caused by zigzag conduction around myocardial scarring resulting from a previous myocardial infarction or inflammation [12]. It has been reported that fQRS showed high sensitivity and a high negative predictive value for detecting myocardial scars in patients with CAD identified via myocardial single-photon emission tomography and CMR [11,15]. The fQRS of patients with sarcoidosis was also reported to be associated with cardiac involvement detected using late gadolinium enhancement on CMR to identify myocardial fibrosis [14]. A previous report indicated that myocardial edema, defined as an increased T2 ratio on CMR appearing as myocardial infarction and inflammation, was observed in 63% patients who had not undergone treatment for SLE at the time of diagnosis. This result is similar to ours, which indicates that 59% of untreated patients with SLE had fQRS at the time of diagnosis [24]. A previous study on fQRS in patients with SLE after treatment interventions indicated that the fQRS(+) rate was 41%, which was lower than that in our study [22]. Because 33% of the fQRS(+) patients disappeared fQRS after immunosuppressive therapy, the differences in results may be explained by the disappearance of fQRS with the initiation of therapeutic interventions for SLE. Reversible fQRS has also been reported in several studies on myocardial infarctions [25,26]. Therefore, we hypothesized that fQRS would be expressed by not only irreversible myocardial replacement fibrosis because of myocardial infarction or inflammation but also by ischemia or inflammation themselves in patients with SLE in the absence of typical cardiac symptoms. Our results showed that fQRS and SLE disease activity are related, suggesting that fQRS exists in patients with SLE disease activity high enough to cause myocardial injury. Cardiac involvement in SLE including myocarditis and infarction may be mediated by immunological mechanisms. Immunofluorescence studies indicated that fine granular immune complexes and complement deposits were detected in the walls and perivascular tissues of myocardial blood vessels [27]. Inflammatory cell infiltration in the coronary artery has also been observed in the autopsy results of patients with SLE who died secondary to myocardial infarction [1]. These pathologic findings suggested that lupus myocarditis and infarction are primarily caused by circulating immune complexes that mediate inflammation. Moreover, increasing T2 ratios indicated that myocardial inflammation and infarction detected on CMR were significantly more frequent in patients with SLE who had a high disease activity [8,9]. Furthermore, in the LUMINA study, which included a multi-ethnic United States cohort, it was reported that a high disease activity at the time of diagnosis was related to the development of myocarditis [5]. A cohort of individuals with lupus in Toronto was found to have higher baseline and recent disease activity scores that increased with the occurrence of CAD-related events [6]. Therefore, we believe that higher SLE disease activity results in increased myocardial involvement; therefore, fQRS is relevant to SLEDAI-2K, which can comprehensively evaluate systemic organ damage mediated by immunological mechanisms.
The association between fQRS and lupus nephritis in this study may be explained by an immunological mechanism related to in situ immune complexes (ICs). The cause of lupus nephritis involves not only circulating ICs but also in situ ICs, such as the binding of anti-dsDNA antibody to α-actinin [28,29]. It has been speculated that α-actinin is also present in cardiac muscle and that myocardial damage occurs via a similar mechanism. Although α-actinin is frequently present in skeletal muscle, it was not associated with fQRS and myositis in this study [30]. Since myositis usually presents with atypical and non-specific symptoms such as fatigue, it was not detected in this study and was possibly unrecognized as being related to fQRS [31]. Recent epidemiological studies showed that SLE patients with lupus nephritis have a significantly higher risk of MI and CVD mortality than those without lupus nephritis and that they had carotid atherosclerotic plaques twice as often as non-nephritis SLE patients and population controls [32,33]. Therefore, lupus nephritis may be related to fQRS because of both the kidney and the heart being targets of in situ ICs.
There are 3 main strengths of this study. First, to the best of our knowledge, this is the first study to use a multivariate analysis adjusted for appropriate confounding factors to evaluate fQRS in SLE. In a previous study using univariate analysis, an association was found between fQRS and disease duration [22]. Therefore, we defined the period from the onset of symptoms to the diagnosis as the confounding factor. Second, it was revealed that fQRS disappears after immunosuppressive therapy, suggesting that the mechanism of fQRS by SLE involves reversible inflammation and ischemia. This indicates that fQRS is a marker that not only identifies myocardial involvement at the time of diagnosis but also can evaluate changes due to treatment intervention, and leverage the strength of ECG that can be repeatedly evaluated non-invasively. Third, 3 sensitivity analyses were performed to determine the main outcome, and all results were compatible with each other. There was a significant difference in the results as noted by the 2 blinded cardiologists. Additionally, excellent inter-rater agreement was achieved by each of the cardiologists. A significant relationship was reproduced when fQRS was printed at 100% magnification and read on paper. This was shown to be useful in environments where the ECG could not be read on a monitor. Even after excluding 4 patients with classical cardiovascular risk factors, a significant association between fQRS and SLEDAI-2K was maintained.
There are also 3 main limitations to this study. First, 4 patients were excluded because they had no ECG measurements available at the time of diagnosis. However, despite the small sample size in our study, there was no bias regarding the main outcome of SLEDAI-2K. Second, this was a retrospective evaluation of a series of medical records. However, data deficiencies for the main outcome and the confounders were not admitted. Third, coexistence of myocardial involvement not contributed by SLE at the time of diagnosis cannot be denied. However, there was no significant difference between the fQRS(+) and fQRS(-) groups with respect to the Framingham Risk Score and coronary risk factors (hypertension, diabetes, dyslipidemia, LDL cholesterol, HDL cholesterol, triglyceride, uric acid level, and HbA1c). Furthermore, after excluding patients with classical cardiovascular risk factors, the sensitivity analysis showed a significant association between fQRS and SLEDAI-2K.
There are 2 clinical implications of the study findings. First, fQRS defined by ECG is suitable for screening for myocardial involvement in patients with SLE because it can be measured immediately and in any environment. As most myocardial involvement in patients with SLE is subclinical, it is necessary to evaluate all patients with SLE. Although CMR is highly sensitive for detecting subclinical cardiac involvement in patients with SLE, its use is restricted by its high medical costs, medical infrastructure, and patient condition [7]. Therefore, it is impossible to perform CMR at the time of diagnosis for all patients with SLE. The mechanisms of examination are different for fQRS, which evaluates conduction disturbances electrophysiologically, and CMR, which provides qualitative evaluations of the myocardial tissue. However, there was a good correlation between fQRS and late gadolinium enhancement for various cardiomyopathies identified using CMR [14,34,35]. Therefore, even for patients with SLE, fQRS is useful for routine evaluations of myocardial injury. A mechanized learning approach to detect fQRS has been attempted and is expected to provide more objective and simple indications [36]. The fQRS can be used as a suitable parameter for routinely screening myocardial involvement in patients with SLE. Second, fQRS could be a predictor of long-term cardiac function and arrhythmia in patients with SLE. It represents myocardial replacement fibrosis, which appears at the sites of prior inflammation or infarction and can be associated with ventricular dysfunction and the development of congestive heart failure. Additionally, myocardial scars detected using fQRS are a substrate for re-entrant ventricular tachyarrhythmia [37]. A metaanalysis of patients with reduced EF showed that fQRS was associated with increased all-cause mortality up to 1.63-fold, as well as increased major arrhythmic events up to 1.74-fold [38]. Another meta-analysis of patients with acute coronary syndrome reported that fQRS was an independent predictor of mortality, major adverse cardiovascular events, and deteriorating LV function [15]. In the future, we plan to perform a longitudinal study to clarify whether fQRS can predict cardiovascular disease in patients with SLE.

Conclusion
Our results demonstrated that fQRS(+) is associated with high disease activity in patients with SLE. We believe that fQRS can detect subclinical myocardial involvement in patients with SLE and could be a predictor of long-term cardiac function and arrhythmia.