A meta-analysis on the prevalence, associated factors and diagnostic methods of mental stress induced myocardial ischemia

Background The high prevalence of mental stress induced myocardial ischemia (MSIMI) causes double risk of adverse cardiac events in patients with MSIMI. However, multiple types of mental stress, diagnostic techniques, and diagnostic measurements may increase the complexity and heterogeneity in the assessment of MSIMI. Therefore, we performed this meta-analysis to assess the prevalence, associated factors, and diagnostic methods of MSIMI. Methods We systematically searched PubMed, EMBACE, Web of Science, CNKI, Wanfang through 1 Feb 2020 in English and Chinese. Review Manager (RevMan) Version 5.3 and Stata 12.0 were used for data analyses. Results Twenty articles were enrolled. The pooled estimates for the prevalence of MSIMI in CAD patients was 32%. Potential associated factors of MSIMI involved history of post myocardial infarction (MI), or coronary artery bypass graft (CABG) (RR: 1.29, 95% CI 1.00–1.66, P = 0.05; RR: 1.59, 95% CI 1.00–2.52, P = 0.05). Evidence supported that diagnostic methods could influence the prevalence of MSIMI. Significant differences of MSIMI prevalence were found in different types of mental stress (Public Speaking: 22%; Mental arithmetic: 26%; Anger recall: 34%; Two types: 37%; Three or more than three types: 43%, P = 0.02), diagnostic techniques (SPECT: 26%; RNV: 38%; ECG: 16%; Echocardiography: 41%; Two types: 43%, P < 0.0001), and diagnostic measurements (LVEF decrease: 19%; WMA: 51%; ST depression: 16%; MPD: 26%; Two or more than two measurements: 45%, P < 0.00001). Moreover, univariate meta-regression demonstrated that MSIMI was linked with mental stress (exp(b): 1.0508, SE: 0.0201, P: 0.018). Conclusions This meta-analysis implicated that patients with diabetes, post MI or CABG might be more vulnerable to MSIMI. However, the prevalence of MSIMI could be influenced by diagnostic methods, especially the adopted types of mental stress, diagnostic techniques and measurements. Therefore, it is necessary to formulate a standard diagnostic method for MSIMI, which should be adequate, assessable, and affordable worldwide. Registration PROSPERO. Online Protocol: CRD42020162822.

traditional risk factors such as smoke, hypertension, hyperlipidemia, and diabetes [3], increasing evidence identified mental stress as a crucial risk factor in the development and progression of CVD [4]. Researchers discovered that mental stress induced in the laboratory (e.g. mental arithmetic, public speaking, et al.) contributed to myocardial ischemia, which could be assessed by echocardiography, electrocardiogram, or SPECT (single photonemission computed tomography) [5][6][7]. The prevalence of mental stress induced myocardial ischemia (MSIMI) ranges from 50 to 70% in patients with coronary artery diseases (CAD) [8]. Compared with patients without MSIMI, patients with MSIMI have double risk of adverse cardiac events [8].
However, the mechanisms of MSIMI remain uncertain. Previous studies showed that emotional status such as the trait and state of anger [5], anxiety [6], depression [7] could have great impact. Potential mechanisms may involve inflammatory responses, cortisol responses, fibrinogen responses, coagulation system, hypothalamic pituitary adrenal (HPA) [9,10]. Hammadah et al. [11] linked cardiac biomarker with MSIMI, presenting that patients with MSIMI had higher level of resting cTnI. However, various factors such as sex, race, disease history, and drug history, and multiple types of mental stress, diagnostic techniques, and diagnostic measurements increase the complexity and heterogeneity in the assessment of MSIMI. Therefore, we performed this meta-analysis and metaregression in an effort to explore the potential mediators of MSIMI.
Inclusion criteria: (1) Prospective cohort study or cross-sectional study; (2) English or Chinese language; (3) Patients with coronary artery disease; (4) Full articles were able to be found; (5) The data were eligible to be extracted; (6) Articles with high or medium quality.
Exclusion criteria: (1) Articles with repeated data from the same study project; (2) Mental stress tasks followed exercise stress at the same day, which might implicated that myocardial ischemia could be induced by exercise stress rather than mental stress.

Article selection steps
Three authors focused on selecting the proper articles in nearly 1 month. There were four steps in article selection and data extraction. First, the authors read the titles and excluded those unsuitable; Second, they read the abstracts and included those articles in the scope of our research; Third, they downloaded the articles with full text through the internet or our country library; Fourth, they read all articles, extracted necessary data for this study, and excluded articles without qualified data. Agreement must be reached among three authors to process the data.

Quality assessment
The quality of the cross-sectional studies was assessed by Crombie tool [12]. According to the scores, the article was classified into Grade A (6.0-7.0 points), Grade B (4.0-5.5 points), Grade C (< 4 points). Articles with Grade A were regarded as high quality, Grade B as moderate quality, and Grade C as low quality. The quality of the prospective cohort studies was assessed by Newcastle-Ottawa Scale (NOS) [13]. Articles with seven to nine stars were estimated as high quality, five to six stars as medium quality, and zero to four stars as poor quality. Articles with high or medium quality were included in the present study (Table 1).

Data extraction
The data were extracted by two researchers separately and reached agreement after consultation. The following data were extracted: first author; publication date; country; total sample size; the sample of patients with MSIMI; study type; mental stress; diagnostic methods; scores of Crombie/NOS; article quality. All the data were presented in Table 1. More details about diagnostic methods were shown in Table 2.

Statistical analysis
Review Manager (RevMan) Version 5.3 and Stata 12.0 were adopted for data analyses. Cochran's Q-test [14] and I 2 statistic [15] were used for heterogeneity. Pooled effect size was analyzed by random-effects model or fixed-effects model according to the level of heterogeneity. Random-effects model was established for significant heterogeneity (P < 0.10 or I 2 > 50%), while fixed-effects model was used for non-significant heterogeneity (P > 0.10 or I 2 < 50%). Meta-Regression and subgroup analysis were applied for seeking heterogeneity sources. Sensitivity analyses were performed via excluding studies one at a time [16]. Publication bias was estimated by funnel plot and Begg' test [17]. P values were two-sided, and P < 0.05 was considered statistically significant.

Diagnostic methods Types Details
Mental stress Mental arithmetic Participants were required to complete a series of mathematical calculation, for instance, to subtract 7 from a 4-digit number in 5 min as quickly as possible, at the same time, they would receive encouragement or discouragement from the investigators Public speaking Participants were asked to give a speech on a topic given by the investigators, and they had 2 min to prepare and 3 min to deliver the speech. They were told that their speech would be recorded and evaluated by the investigators Mirror trace Participants were instructed to outline the shape of a star from its reflection in a mirror Stroop color word task Participants were showing a series of slides which displaying the written word of a nonmatching color (e.g. the word green in blue color)

Anger recall
Participants were asked to recall a recent annoying event which made them feel angry, upset, irritated, frustrated, then described the situation and feeling to the investigators in details

Reading
Participants were asked to read a passage given by the investigators, such as neutral passage, in front of the investigators Type A structured interview Participants underwent a standard videotaped interview to assess type A behavior which might last 20 min Competitive computer game Participants were asked to play a kind of computer game, which might elicit threat, uncertainty, and avoidance.  (Table 3).

History of post MI
Five articles [19,22,25,31,33] were selected in the subgroup comparison of post MI history, including 585 patients with post MI and 760 without. Difference of MSIMI was found between patients with post MI and patients without (RR: 1.29, 95% CI 1.00-1.66, P = 0.05). This result indicated that patients with post MI history might be at higher risk of MSIMI (Fig. 3a, Table 4).

History of CABG
Three articles [22,25,42] were selected in the subgroup comparison of history of CABG, including 432 patients with CABG history and 696 patients without. Difference of MSIMI was found between patients with CABG and patients without (RR: 1.59, 95% CI 1.00-2.52, P = 0.05), indicating that patients with history of CABG might be at higher risk of developing MSIMI (Fig. 3b, Table 4).

History of diabetes
Eight articles [19,20,22,25,28,29,31,33] were selected in the subgroup comparison of history of diabetes, including 608 patients with diabetes history and 1416 patients without. Although no significant difference was found (RR: 1.26, 95% CI 0.98-1.62, P = 0.07), we still considered the potential risk of diabetes in MSIMI due to its impact in coronary artery disease.

Other characteristics
Other characteristics were also conducted in this metaanalysis including sex  (Table 4).

Diagnostic methods of MSIMI
There were significant differences in the prevalence of MSIMI in different types of mental stress, diagnostic techniques, and diagnostic measurements. The prevalence of MSIMI detected by Public Speaking was 22%, Mental arithmetic was 26%, Anger recall was 34%, Two types was 37%, three or more than three types was 43%, and the result was significant (P = 0.02). The results indicated that two and more than two types of mental stress could be more likely to induce MSIMI. The prevalence of MSIMI detected by different types of diagnostic techniques and diagnostic measurements showed significant difference (Table 3, Figs. 4, 5, 6). Meta-regression was performed to identify the potential moderators in the prevalence of MSIMI, including publication date, sample size, country, different types of mental stress, different types of diagnostic techniques, and different types of myocardial ischemia measurements (Tables 3 and 5).    On univariate meta-regression, types of mental stress    Table 5).
Our results indicated that different types of mental stress might influence the prevalence of MSIMI in CAD patients.

Comparisons of different diagnostic techniques
In all twenty selected articles, there were four articles indicating that different diagnostic techniques might lead to different prevalence of MSIMI in the same population. Two articles compared SPECT and PAT (peripheral arterial tonometry) which was not recognized as a standard criterion. In Burg's study, the prevalence of MSIMI was 32.35% by SPECT, and 42.65% by PAT, while only 19.12% by both. The area under the curve (AUC) was 0.613 (SE, 0.065, onesided P = 0.04). In Hassan's study, when comparing SPECT and PAT, the area under the curve (AUC) was 0.59 (95% CI 0.48-0.69, P = 0.116). In addition, Carels' study showed that the prevalence of MSIMI was 33.09% by RNV, and 44.12% by ambulatory ECG, while only 19.2% by both. Krantz's study showed that the prevalence of MSIMI was 55.7% by RNV, and 57% by echocardiography (Table 6).

Table 4 Subgroup comparisons results of the meta-analysis
ACEI angiotensin converting enzyme inhibitor, ARB angiotensin receptor block, CCB calcium-channel blocker, MI myocardial ischemia

Sensitivity analysis
We performed sensitivity analysis by Stata 12.0, excluding a single study each time to detect the influence of individual dataset on pooled ESs. The results demonstrated that no significant change was found after omitting any of the study (Fig. 7).

Publication bias
Publication bias was detected by funnel plot and modified Begg' test. The funnel plot was symmetric and the Begg' test presented no significant publication bias in this meta-analysis (Z = 1.69, P > 0.05) (Fig. 8).

Discussion
In this meta-analysis, the pooled estimated prevalence of MSIMI in CAD patients was as high as 32%. Consequently, it attracted our attentions to summarize the related factors and diagnostic methods of MSIMI in CAD patients to better understand the MSIMI assessment. To the best of our knowledge, this is the first meta-analysis concentrated on this topic.

Associated factors of MSIMI
Subgroups analyses elucidated that CAD patients with history of diabetes, or post MI, or CABG might be associated with a higher risk of MSIMI, though the statistical analysis was not significant enough. Diabetes is considered as a risk factor of CAD, due to the dysfunction of micro-and macro-vascular damaged by hyperglycemia [38] via inflammation pathway. The sudden mental stress results in the lack of blood flow and oxygen, and thus causes myocardial ischemia [9]. Patients with diabetes, or post MI, or CABG, have worse cardiac conditions because of existing cardiac cell damage and microvascular dysfunction. Therefore, they are more vulnerable to myocardial ischemia when mental stress occurs. Our results indicated that there was no significant difference of developing MSIMI between females and males, and between patients with depression and without. The Vaccarino et al. [35] elucidated that young women with CHD were more likely to develop MSIMI, which was almost fourfold higher than men. Another study of Vaccarino reported similar conclusions that mechanisms in MSIMI could be different in females and males, and the higher morbidity of MSIMI in females might be related with the microcirculatory dysfunction. Samad et al. [39] suggested that the higher morbidity of MSIMI in females might be associated with platelet activity. To our surprise, our results in the present study did not suggest sex as a significant risk factor. This inconsistency might be due to: (1) the different samples and proportion of females and males in each study; (2) the studies were from different regions. More original researches should be done to further study the relationship between sex and MSIMI.
Depression is an independent risk factor of cardiovascular diseases [40]. Jiang et al. [7] suggested that patients with mild to moderate depressive symptoms were at higher risk of MSIMI. In this research, depression was assessed by Center for Epidemiological Studies-Depression scale (CES-D). However, only four articles mentioning depression were included in the present meta-analysis, and no significant importance was found in depression as a risk factor for MSIMI.
In addition, anger [5], sever left ventricular dysfunction, and anxiety [6] have been considered as severe factors in MSIMI, but the evidence is not enough.

Diagnostic methods of MSIMI
We found significant differences in MSIMI prevalence detected by different mental stress, diagnostic techniques and diagnostic measurement. Univariate metaregression elucidated the potential link between types of mental stress and MSIMI. We postulated some potential reasons for this association. First, the activation of different signal pathway may lead to different consequences. The mechanism of MSIMI involves the strong interaction between heart and the brain. Mental stress can activate hypothalamic pituitary adrenocortical axis, sympathetic nervous system, adrenomedullar hormonal system, and parasympathemic nervous system via releasing different hormones or neurotransmitters which can have different impact. Second, individual differences may play an important role. In Table 1, we described the types of mental stress in all the included studies. The common types involve mental arithmetic, anger call, public speech, mirror trace, Stroop color-word test et al. We found that the prevalence of MSIMI induced by one type of mental stress was 22-34%, two types of mental stress was 37%, and three types yield 43% (Table 3). According to our own clinical observations, trace mirror seemed to be a pleasure rather than emotional stress for those who are good at designing or drawing, while mental arithmetic could be a serious stress to them for most of them are afraid of mathematics; vise verse for those who are skilled at mental arithmetic. The phenomenon implied that we should consider individual differences in the consequence caused by different types of mental stress task, which is consistent with Bremner et al's study. Bremner [41] conducted a study with the intent of revealing the association between brain and MSIMI. It was found that mental arithmetic was associated with left insula activation, while public speaking was associated with right pre/ post-central gyrus and middle temporal gyrus activation. In the context of MSIMI, different types of mental stress might active or deactivate different brain regions, which would promote or inhibit cardiac responses. Therefore, we suggest that researchers should consider individual differences in different types of mental stress task while assessing MSIMI, and make a standard together. In our  opinion, two different types of mental stress tasks would be better to diagnose MSIMI for the reason that one type might not be eligible to provoke MSIMI, while more than two types might be time and economic consuming. In addition, we took diagnostic techniques as a pivotal factor in diagnosing MSIMI. As Table 3 showed that the prevalence of MSIMI diagnosed by SPECT was 26%, ECG yielded 16%, while echocardiography yielded 41%, RNV yields 38%. SPECT is a direct way to observe myocardial ischemia via myocardial perfusion defects, demonstrating its vital role in diagnosing MSIMI. Good reproducibility of SPECT has also been identified [42]. However, some patients with MSIMI assessed by echocardiography might be missed.
ECG is a convenient technique, but it is been proved not sensitive enough for MSIMI [36]. Jiang et al. [25] investigated both ECG and echocardiography in distinguishing MSIMI, while no myocardial ischemia was discovered by ECG. Therefore, the false negative of ECG presented low prevalence of MSIMI.
Echocardiography is economical and practical in clinical practice, which could detect LVEF response and wall motion during mental stress simultaneously. Though LVEF decrease could result from myocardial ischemia induced by mental stress and also be consistent with SPECT [43], LVEF response is also influenced by hemodynamics and the basic left ventricular function [18]. Therefore, echocardiography is likely to generate false positive results.
Peripheral arterial tonometry (PAT) is applied to assess microcirculation dysfunction, which is expected to detect myocardial ischemia induced by mental stress. CAD patients with MSIMI have lower PAT ratio according to the studies comparing SPECT and PAT. Some researchers suggested that PAT might have similar detection efficiency compared to SPECT and RNV [20,44], and more researches remain to further explore the potential role of PAT in detecting MSIMI and make it standardized.

Conclusions
In conclusion, the pooled prevalence of MSIMI in CAD patients is 32%. The present meta-analysis implicates that patients with diabetes, or post MI or CABG are more vulnerable to develop MSIMI and different types of mental stress and diagnostic techniques might influence the prevalence of MSIMI. Therefore, it is necessary to formulate a standard diagnostic method for MSIMI, which should be adequate, assessable, and affordable all around world.