Intraoperative cerebral high intensity transient signals and postoperative cognitive function: a systematic review

Summary

Microemboli during surgery have been hypothesized to cause postoperative cognitive changes. The purpose of this article was to systematically review the available literature related to intraoperative microemboli, measured with transcranial Doppler ultrasound and postoperative cognitive function. The literature remains largely undecided on the role of microemboli and cognitive impairment after surgery, because most studies underpowered to show a relationship.

Introduction

Cognitive decline after cardiac surgery is a common complication and has been reported in approximately 50% of patients at discharge, 36% at 6 weeks, 26–33% at one year, and 42% at five years[1,2] postoperatively. Cerebral microemboli, due to gaseous, organic, or inorganic particles generated during the procedure, have been hypothesized to be a primary predictor of cognitive decline following cardiac surgery[3]. Transcranial Doppler Ultrasound (TCD) is a portable, non-invasive tool for the measurement of cerebral hemodynamics which has been used in the operative setting to measure the high intensity transient signals (HITS) which are felt to represent particulate or gaseous microembolic particles[4,5]. The focus of this paper was to systematically examine the evidence relating intraoperative HITS to cognitive decline postoperatively.

During cardiac surgery, numerous embolic sources including pre-exisiting atherosclerotic plaque, thrombogenesis from the foreign surfaces of cardiopulmonary bypass, air from the cardiopulmonary bypass circuit, and pericardial fat globules are generated.[5] Macroemboli (>200µm) occlude larger arteries that supply focal vascular territories and produce a clinically apparent stroke. Microemboli may only occlude small arterioles and produce no or only subclinical findings with no apparent physiologic deficits. Within an artery, embolic material has different reflective properties than the surrounding red blood cells and produce distinct HITS from the background flow velocity pattern. These HITS are felt to represent microemboli.

Pathophysiologically, microemboli may cause neurological injury. Autopsy of patients who died after cardiac surgery found evidence of microemboli in all of the brains examined[6] Further, Aly-Omar, et al[7] used magnetic resonance imaging to demonstrate that HITS can result in acute or subacute brain infarction. However, the size of the microemboli make clinically evident stroke relatively infrequent after surgery[8]. Most studies have focused the relationship of HITS to more subtle, postoperative cognitive functioning.[9],[10] The purpose of this paper was to systematically review the literature to determine if intraoperative HITS were related to cognitive function after surgery. Because of the time commitment and resource commitment required to assess preoperative and postoperative cognitive function and to measure and quantify HITS, we predicted that most studies of intraoperative HITS and cognitive function would be small (<100 patients). We hypothesized that 1) operative procedures with direct connection to the cerebral vasculature (cardiac surgery and carotid endarterectomy) would be more studied and provide the best evidence, 2) the evidence would derive from smaller studies, which are less likely to demonstrate a significant change in cognitive function, and 3) the lack of a unified testing battery and common outcome limit the ability to combine the available studies into a meta-analysis.

Methods

A pubmed search of studies (1965-July 2007) was conducted. We combined 3 broad category searches (cognition, surgery, and TCD) to identify appropriate abstracts for review. The cognition search category combined the results of the following the medical subject headings (MeSH) and keywords: “cognition”, “dementia”, “postoperative cognitive dysfunction”, “delirium”, or “encephalopathy”. The surgery search category combined the following results “surgery”, “cardiac surgery” “carotid endarterectomy”, or “CABG”. The TCD category combined the following results: “transcranial Doppler”, “Doppler ultrasound”, “embolism”, “intracranial embolism”, “blood flow velocity”, or “microemboli”. The three categories were examined for intersection. Further, manual searches of references from relevant study articles and reviews were also conducted.

Selection of English language articles was performed by three authors (KKM, JBW, JLR). We included all surgical procedures, because the understanding of HITS in one surgical procedure may provide valuable evidence to the relationship in other studies. We selected prospective studies that used cerebral TCD to count HITS during the surgical procedure. Studies that performed neuropsychological evaluation before and after surgery were included. Randomized studies of interventions (stenting vs. surgery; different arterial filters for cardiopulmonary bypass) were included if they described the relationship of intraoperative HITS and neuropsychological performance.

Results

Our pubmed search identified 337 abstracts which were reviewed and 56 articles were identified for review. From these, we selected 14 cardiac surgery, 5 carotid endarterectomy (CEA) and 2 orthopedic surgery articles for presentation. Each subset of surgery is detailed below. In agreement with our hypothesis about smaller studies, 12 of 14 (86%) cardiac surgery studies, 2 of 5 (40%) CEA studies, and 2 of 2 (100%) orthopedic studies had fewer than 100 patients.

Cardiac Surgery

Table I describes the studies that examined intraoperative HITS and subsequent cognitive dysfunction after cardiac surgery. Several features of these studies warrant discussion. First, intraoperative HITS occur during nearly all cardiac surgical procedures,[9,11,12]. Among the fourteen studies identified, there is considerable heterogeneity with respect to the cognitive assessments and batteries, which limits the ability to make definitive conclusions about the relationship of HITS and cognitive deficits. For example, each study utilized a different number and combination of neuropsychological tests and the methodology used to define the outcome of cognitive decline is not consistent among the studies. Furthermore, the testing batteries were delivered at differing times postoperatively (range: 5 days – 5 years). Additionally, the statistical analysis of the HITS is variable between the studies. For example, three studies dichotomize HITS, which reduces statistical power, while others utilize nonparametric methods or correlations to examine HITS as a continuous variable. Further, the size of the cognitive battery and non-normal distribution of HITS result in challenges to the power of the smaller studies. Overall, HITS were associated with neuropsychological performance in 4 of 14 cardiac surgery studies.

Table I.

TCD Studies Measuring HITS and Cognitive Function in Cardiac Surgery

Author & Year	Study Type	Surgical Procedure	n	Cognitive Assessments Definition of Decline	NP Time	Cognitive Findings in Relation to TCD
Motallebzadeh, 2007[13]	Randomized Prospective	CABG -on-pump -off-pump	212	8 NP tests Composite score	Preop; Postop discharge, 6 weeks, & 6 months	HITS: on-pump (1605) vs. off- pump (9) (p<.01) After adjustment for age, education, and on-pump status, HITS did not remain in the cognitive decline regression model
Bokeriia, 2007 [54]	Prospective	CABG Valve ± CABG	62	11 NP tests Z-scores on individual tests	Preop; Postop 2–4 weeks	HITS: CABG (463) vs. Valve ±ABG (881) (p<.01) Total HITS not associated with cognitive function Left ME may predict verbal memory performance (p=.07) Right ME may affect psychomotor function (p=.07)
Stroobant, 2006[55]	Prospective	CABG -on-pump -off-pump	50	7 NP tests 20% decline on 20% of tests	Preop; Postop 6 days and 6 months	HITS: on-pump (335) vs. off- pump (145) (p<.01) No correlation between number of HITS and NP tests
Abu-Omar, 2006[41]	Randomized Prospective	CABG -on-pump -off-pump	25	1 Verbal Memory Task	Preop; Postop 4 weeks	HITS: on-pump (254) vs. off- pump (35) (p<.01) No difference in preop and postop testing HITS negatively correlated with MRI prefrontal activation (r=−.61, p<.01)
Stygall, 2003[9]	Longitudinal Prospective	CABG	107	11 NP Tests Residualized change scores Simple change scores	Preop Postop 5 years	ME predicted of NP decline at 5 years (unadjusted)
Whitaker, 2003 & 2006 [56,57]	Randomized Prospective	CABG -3 arterial filters	192	9 NP tests ≥1 SD decline in ≥2 tests	Preop; Postop 6–8 weeks	Arterial filter decreases HITS HITS not correlated with NP change ≤35 HITS associated with better NP recovery
Lund, 2003[58]	Randomized Prospective	CABG -on-pump -off-pump	52	8 NP tests 20% decline on ≥2 tests	Preop; Postop 3 months	HITS: on-pump (90) vs. off-pump (16) (p<.01) No correlation between HITS and NP impairment
Eifert 2003[59]	Prospective Randomized	CABG +/− Valve -Aortic filter	24	4 reaction time tests for attention and alertness	Preop; Postop 7 days	No difference between filter and non filter group with respect to HITS or NP function HITS were not correlated with NP performance
Nadareishvili, 2002 [60]	Prospective	Valve Replacement	36	4 Cognitive screening measures Measures analyzed individually	Postop 3 months	No correlation between HITS and cognitive decline (p > 0.05)
Browndyke 2002[61]	Prospective Controlled	CABG or Valve	27	5 NP Measures Measures analyzed individually	Preop; Postop 10 days and 30 days	Surgical subjects perform worse than controls Surgical subjects performed worse at 10days than at 30 days No correlations between HITS and NP performance
		Controls	25
Fearn, 2001 [62]	Prospective	CABG	65	Cognitive Drug Research Battery (4 tests)	Preop; Postop 1, 8, & 24 weeks	57% of patients had >200 HITS HITS correlated with decreased memory accuracy
Jacobs, 1998[42]	Prospective	CABG ± Valve	18	10 NP Tests	Preop; Postop 10 days, 3 months	12 patients with completed TCD PET scan uptake correlated with HITS No relationship of HITS and NP performance
Braekken, 1997[63]	Prospective	CABG and Valve Replacement;	40	10 NP Tests ≥1 SD decline in ≥1 test	Preop; Postop 2 months	Valve repair patients with cognitive deficits were more likely to have increased HITS counts. CABG patients had lower HITS counts
Clark, 1995 [64]	Randomized Prospective	CABG	41	Neurobehavioral Cognitive Status Examination Cognitive decline not defined	Preop Postop 5–10 days	Gradient effect of HITS with NP decline in 35% of those with >60 HITS

CABG - coronary artery bypass graft; + CABG - indicates with or without coronary artery bypass graft; HITS - high intensity transient signals; ME - microemboli; MRI - magnetic resonance imaging; NP-neuropsychological; On-Pump- utilizing the cardiopulmonary bypass machine/pump; Off-pump - surgery not requiring use of the cardio-pulmonary bypass machine/pump; Preop – preoperative; Postop- postoperative; SD- standard deviation

The three largest studies (n>100) are mixed with the regard to the relationship of intraoperative HITS and cognitive outcomes. Motallebzadeh et al. found that off-pump coronary artery bypass grafting (CABG) was associated with significantly fewer median HITS compared to on-pump CABG (off-pump median 9; intraquartile range [IQR] 4, 28 vs. on-pump median 1,605; IQR 750, 2,475).[13] On-pump CABG was associated with significantly worse cognitive decline at discharge (−0.25 SD), but not at 6 weeks or 6-months. Upon adjustment for age, education, and pump-status, HITS were not associated with cognitive decline.[13] Whitaker, et al found no relationship between HITS and cognitive decline in 192 patients at 6–8 weeks and 3 months postoperatively. However, there were significant losses to follow-up and the primary focus of the study was to compare the effects of three arterial filters on HITS, not cognitive decline. Stygall et al. determined that intraoperative HITS were independently associated with cognitive decline 5-years postoperatively[9,14]. While the length and quality of follow-up were admirable, there was significant attrition and the data on HITS at the 6 days and 8 week timepoints are not presented. Because of the length of time between exposure (HITS) and follow-up, a major issue with the Stygall study is whether HITS, atherosclerosis predisposition, or other interim factor caused subjects to develop cognitive impairment. Atherosclerosis is associated with HITS and cognitive decline and thus, may confound the relationship of HITS and cognitive decline.

One of the larger studies (n=395) of microemboli and cognitive function during cardiac operations was not included in the table, because the study used a unilateral Doppler ultrasound on the carotid artery to measure HITS, not TCD[15]. The study found increased neuropsychological deficits in patients with more than 100 HITS.

Carotid Endarterectomy (CEA)

CEA is considered a high-risk procedure for stroke and cerebral HITS due to the direct connection to the cerebral vasculature[8,16]. Microembolism reportedly occurs during a majority of CEAs and has been associated with neurological sequelae from stroke and transient ischemic attacks[17–19] to more clinically subtle and silent lesions[19]. In addition to these studies, however, microembolism is seen as a major culprit in cognitive decline often witnessed post-carotid endarterectomy[20,21]. While, TCD has been established as a useful tool to monitor the occurrence of intraoperative HITS from microembolism during CEA[22,23], few studies have evaluated the interconnectedness of CEA, intraoperative HITS and post-operative cognitive decline. Table II describes five studies that have examined this relationship between intraoperative HITS and cognitive decline after CEA. One study was able to find an association with HITS and neuropsychological decline. In subgroup analysis, two studies found that higher intraoperative HITS (>10 HITS) were associated with some degree of cognitive decline. Another study comparing CEA and carotid angioplasty found no relationship between HITS and cognitive decline in either procedure[24]. Thus, conclusions about HITS and cognitive function are limited by the limited findings of an association, the low number of subjects in subgroup analysis, and the low number of intraoperative HITS relative to cardiac surgery.

Table II.

HITS and Cognitive Function after Carotid Endarterectomy

Author & Year	Study Type	Surgical Procedure	n	Cognitive Testing Definition of Decline	Testing Time	Cognition Findings in Relation to TCD
Bossema, 2005 [65]	Prospective	CEA	58	8 NP Tests >1SD decline from preop	Preop; Postop 3 months	No association between HITS and cognitive decline 3 months postop
Lloyd, 2004 [66]	Randomized Prospective	CEA	100	NP Battery of 7 cognitive domains	Preop; 6 months Post-op	Patients subjected to >10 HITS during carotid dissection had decline in at least two tests of cognition (p< 0.05; relative risk = 1.125)
Fearn, 2003[67]	Prospective Controlled	CEA	159	Cognitive Drug Research Battery (4 tests)	Preop; Postop 5 days and 2 months	Overall, no association with HITS Patients with >10 HITS prior to dissection (n=6) tended to be slower in memory
		Controls	20
Crawley, 2000[24]	Prospective Matched	CEA	26	10 NP Tests ≥1 SD decline in ≥2 tests	Preop; Postop 6 weeks & 6 months	More HITS in PTCA No correlation between HITS and NP score
		Carotid PTCA	20
Gaunt, 1994 [16]	Randomized Prospective	CEA	100	30-point NP Assessment	Pre-op; 5–7 days Post-op	>10 particulate embolic events detected by TCD correlated to post-operative cognitive decline (p<0.05)

CEA-Carotid endarterectomy; HITS – high intensity transient signals; NP-neuropsychological; On-Pump- utilizing the cardio-pulmonary bypass machine/pump ; Off-pump – surgery not requiring use of the cardio-pulmonary bypass machine/pump; Preop – preoperative; Postop- postoperative; PTCA – angioplasty; SD – standard deviation

Orthopedic Surgery

Although cardiac and carotid operations offer a more direct potential impact of circulating microemboli by their anatomical proximity and access to the cerebral arterial system, it has long been known that fat and thromboembolism occurs after long-bone fractures, total hip arthroplasty[25,26], knee athroplasty[27–29], and other skeletal surgical procedures[30–32]. Although previously believed to be captured by the lung, it has been shown that microemboli may transverse a patent foramen ovale and enter systemic circulation through right-to-left shunts in patients during orthopedic surgeries [33].

Microemboli that occur during orthopedic trauma and surgeries have been postulated to be a possible cause in the cognitive decline of 5%–41% of orthopedic patients[34,35], however, the exact etiologies are still unknown. As in cardiac and carotid procedures, TCD is an effective method of detecting intraoperative HITS.

Although several studies have confirmed intraoperative HITS during orthopedic surgery or documented cognitive decline post-orthopedic surgery, there are only two recent studies which have examined the possible interconnectedness of orthopedic surgeries, the intraoperative quantity of HITS and any clinically perceived neurocognitive decline (Table III). Compared to cardiac surgeries, the number of cerebral HITS from orthopedic surgeries was low (total HITS = 4–9)[35,36]. Both studies demonstrated higher number of HITS counts in subjects with right to left shunts which have bypassed the pulmonary vasculature. However, neither study was able to demonstrate a significant association between intraoperative HITS and cognitive function. Further discussion provided by these researchers hypothesized that their studies were limited by the decreased quantity of HITS in their particular subjects or that the study was diluted by their classification of “cognitively impaired”[36]. Interestingly, a study by Abraham et al.[37] reported that HITS from an operated leg persisted even a week following the operation. As such, potential HITS may occur and cause cognitive sequelae, yet would go undetected under the current methodology of most TCD and surgical studies.

Table III.

HITS and Cognitive Function after Orthopedic Surgery

Author & Year	Study Type	Surgical Procedure	n	Cognitive Testing	Testing Time	Cognition Findings in Relation to TCD
Koch, 2007[36]	Prospective A-V shunt matched	THA or TKA	24	11 NP tests 20% decline in ≥2 tests	Preop Postop 3 days & 2 months	All patients had HITS No correlation between HITS and cognitive performance
Rodriguez, 2005[35]	Prospective	TKA	37	13 NP tests 0.5 SD decline in ≥3 tests	Preop Postop 1 week and 3 months	22 of 37 patients had HITS No association between HITS and cognitive performance

A-V – arterial-venous; HITS- high intensity transient signals; NP – neuropsychological; Preop- preoperative; Postop- postoperative; SD- standard deviation; THA- total hip arthroplasty; TKA – total knee angioplasty;

Discussion

This systematic review examined the literature relating intraoperative HITS, measured with TCD, and their relationship to postoperative cognitive function. At this time, there is insufficient evidence to describe a causal link between intraoperative HITS and postoperative cognitive function. However, this systematic review was able to identify several important points. First, compared to cardiac surgeries (total HITS range= 0–5260), the number of cerebral HITS from orthopedic surgeries (total HITS range = 0–40) and CEA (total HITS range 0–700) was lower. Second, the studies use heterogeneous methods to analyze HITS and assess cognitive outcomes. For example, within each study a different battery of cognitive tests is utilized and a different methodology is used to define cognitive decline. Third, future work in populations at high risk for HITS and postoperative cognitive decline could address some of the limitations in the current literature. Finally, while cardiac surgery and CEA are high-risk for HITS because of the direct connection to the cerebral vasculature, there are many potential sources of HITS during surgery which may be modifiable.

The lack of a causal pathway does not mean that HITS do not contribute to postoperative cognitive function. The brain is a highly redundant and complex organ which may protect against our recognition of cognitive impairments caused by HITS. Clinical manifestation of HITS in one patient may not correlate well with symptoms of HITS in another patient despite similar load and location. As a result, a subclinical infarction from a microembolism may remain subclinical. However, it is still an infarction. Given the choice between a microembolism and no microembolism, we postulate that most patients would prefer no microembolism. With this postulate, we discuss implications for future studies of intraoperative TCD monitoring for HITS.

The basic pathophysiology (i.e. microembolism occludes a small arteriole resulting in cognitive decline) suggests that there should be a detectable cognitive decline with increased microembolic load. However, there may be differences in the composition of the HITS stemming from cardiopulmonary bypass, CEA, and orthopedic surgery. Embolic, air, and inorganic (calcium plaque) HITS are generally treated identically despite potential dissipation of the air emboli. HITS will have greater significance if established guidelines for microemboli detection are applied during studies.[38] Multifrequency TCD technology has been reported to accurately distinguish particulate from gaseous HITS[3] by comparing the reflected frequency of both blood and embolic material at two insonation frequencies [39], [40]. Additionally, functional magnetic resonance imaging (fMRI) cerebral intensity[41] and cerebral glucose metabolism measured with positron emission tomography (PET)[42] have been found to be correlated with number of HITS. While fMRI and PET are superior modalities for localizing microemboli and determining the alterations on cerebral function, use of these modalities in the immediate operative/postoperative setting is limited by patient acuity, the time to obtain the scan, and the physical requirements of the scanner. There is potential research and clinical value in validating the output of TCD, fMRI, and PET imaging modalities against each other to determine the functional implications and composition of HITS. Presently, in vivo data on the composition of the particles is limited.

The studies to date provide an excellent basis for a more comprehensive, prospective study to determine if HITS cause cognitive impairment related to cardiac surgery. TCD studies of HITS and cognitive function are resource intensive. As a result many of the studies reviewed are smaller (<100 patients). This poses statistical challenges to the power of these studies, because HITS during surgery is frequently a skewed variable with many cases having a relatively small number of HITS and fewer numbers of cases with larger HITS counts. As a result, there is a wide standard deviation in HITS counts, often greater than the mean. HITS rate (number of HITS / surgical duration) has been used for comparison of HITS counts between procedures of different duration, but still requires assessment of normality. To address the skew, many studies dichotomize HITS, which further reduces statistical power. Thus, studies with <100 patients would likely lack statistical power to detect a significant difference in HITS between those with and without cognitive decline. Smaller studies also limit our ability to adjust for multiple potential confounding patient variables such as age, prior history of stroke, baseline cognitive functioning, etc. While small studies may provide pertinent information about this evolving field, larger studies, which are still noticeably infrequent, will be necessary to formulate a conclusive opinion about the neurocognitive effects of intraoperative microemboli.

The heterogeneity of testing strategies and definitions of cognitive decline limit our ability to draw definitive conclusions about postoperative cognitive function. At present, there is not a single test for the assessment of cognitive function because of the complexity and redundancy of the brain; instead multiple neuropsychological tests are administered. However, each study has chosen a different battery for analysis of cognitive function. Further, the reduction of many continuously-scored neuropsychological tests into a single dichotomous variable removes important detail about domain-specific cognitive function, is statistically misleading, reduces study power, and lacks clinical validity.[43,44] As highlighted in a comprehensive review of cognitive function after non-cardiac surgery, these limitations are difficult to overcome until a consensus neuropsychological battery (with normative values) and scale for postoperative cognitive changes are validated.[45]

Future studies could target high-risk patients with a high burden of atherosclerotic disease who are more likely to be exposed to a larger HITS load. Atherosclerosis, especially in the ascending aorta and carotid arteries, has been shown to be associated with increased risk for negative cognitive sequelae and may benefit from monitoring[46,47]. Mackensen and colleagues demonstrated that microemboli are associated with atheromatous plaques of the ascending and arch portions of the aorta[48]. In a study of more than 900 CEA patients, the presence of fibrous aortic plaque was significantly associated an increase in HITS.[49] Crawley found that CEA was associated with fewer HITS than carotid artery angioplasty[24]. While further studies are needed to elucidate the neurocognitive impact of microemboli on patients with significant atherosclerotic disease, these patients remain an important study population for future TCD monitoring, because of their increased propensity to require the surgeries frequently associated with microemboli and to be more at risk for cognitive damage from HITS during surgery.

Another population that warrants further study is patients with pre-existing cognitive impairment who may be at increased risk for further decline following surgery. Patients with preoperative cognitive deficits lack reserve capacity to recover from cardiac surgery and have been shown to have a higher incidence of postoperative cognitive impairments[50]. Based on the diminished cognitive reserve, HITS can only serve to further impair cognitive function postoperatively in these patients with pre-existing cognitive impairment. As the incidence of the cognitive decline outcome increases, fewer subjects are needed to appropriately power such studies. If HITS are associated with cognitive impairment, using TCD to minimize HITS exposure in patients with cognitive impairment may be warranted.

A further consideration for use of TCD would include quality improvement among surgical teams. Working with the above postulate that no HITS are preferred to HITS, the use of TCD to measure HITS can inform the team members about the HITS consequences of routine procedures such as rapid injection, cross clamp removal, and CPB cannula insertion. For example, surgical teams have developed protocols to use TCD probes to count HITS occurring in the venous and arterial lines of the cardiopulmonary bypass. Termed ‘transpump Doppler’, this procedure can determine the effectiveness of cardiopulmonary bypass at filtering microembolic load and the effect of perfusionist interventions[51]. The combination of ‘transpump Doppler’ with TCD can accurately determine the source of the microemboli and can be used in quality improvement efforts to reduce microemboli.[52] Additionally, HITS associated with newer surgical procedures could be compared to HITS with traditional procedures. For example, no studies have compared HITS counts in carotid artery stenting vs. CEA. Finally, numerous embolic protective devices have been developed and studied in cardiac surgery, CEA, and carotid artery stenting. While most evaluation studies have focused on neurologic outcomes such as stroke, reduction in HITS could be an additional outcome of future studies.[53]. While the clinical consequences of improving technique may not be immediately appreciated, the long term impact may be significant to the patient[9].

Conclusions

In this systematic literature review, we were unable to demonstrate the causal link between intraoperative HITS, felt to represent microemboli, and postoperative cognitive decline. While some studies have reported an association, most studies have not. As a result, intraoperative HITS may only be a small contributing factor to a multifactorial problem. However, TCD could be used in high risk patients and to improve the technique of the surgical team. Large, multi-site studies are needed to further identify the cognitive and functional sequelae of HITS in a variety of operative settings.