Chapter 12 – Biomarkers of Postoperative Cognitive Dysfunction: Finding the Signal amid the Noise

Chapter 12 Biomarkers of Postoperative Cognitive Dysfunction: Finding the Signal amid the Noise

Miles Berger , MD, PhD, S. Kendall Smith , MD, PhD, and Anver Khan , BA

Introduction

A clinical consensus on how to define and diagnose PND continues to evolve (see Chapter 11), although PND likely represents the extreme end of the continuous distribution of postoperative cognitive change.¹ Nevertheless, efforts to identify effective biomarkers of PND remain an active area of research. According to the NIH Biomarkers Definitions Working Group, a biomarker, or biological marker, is a “characteristic that is objectively measured and evaluated as an indicator of normal biological processes, pathogenic processes, or pharmacologic responses to a therapeutic intervention.”² Despite interesting mechanistic hints from animal models of postsurgical memory deficits, human PND is a cognitive syndrome of unknown etiology.¹ Thus, identifying biomarkers for human PND may help enhance our understanding of its etiology and identify possible therapeutic targets.

There are at least four conceivable categories of biomarkers for PND (Table 12.1). Risk factor markers are those indicators that would potentially identify patients that are most vulnerable to PND. Etiologic markers identify a process that is causally linked to PND or leads to PND at a later time point. Pathophysiologic markers, in contrast, reflect the presence of the actual brain or body state of PND itself. These terms are analogous to the way the terms trait marker and state marker are used in the psychiatry literature. A “trait marker” represents the properties of the behavioral and biological processes that play an antecedent, possibly causal role in the pathophysiology of the psychiatric disorder whereas a “state marker” reflects the status of clinical manifestations in patients.³ Finally, end-product markers would reflect the effect of PND on the brain and/or body, which may remain after PND itself resolves.

Table 12.1 Biomarker Classification, Definitions, and Time Frame

PND Biomarker Type	Definition	Time Frame	Hypothetical Example
Risk factor marker	A marker of a vulnerable patient; semianalogous to the way the term trait marker is used in the psychiatry literature	Can be measured before perioperative care	Decreased CSF amyloid-beta levels³¹
Etiologic marker	A marker of a process which causally contributes to PND or leads to PND at a later time point	Can be measured at the time of perioperative care	?
Pathophysiologic marker	A marker of the actual brain or body state of PND itself – analogous to the way the term state marker is used in the psychiatry literature	Can be measured at the same time as PND is detected by neurocognitive testing (typically weeks after perioperative care)	?
End-product marker	A marker of the effect of PND on the brain and/or body, which remains even after PND itself resolves	Can be measured after PND resolves	?

Biomarkers that are potentially useful for PND diagnosis can be identified in a wide variety of sources (Table 12.2). Thus far, research has demonstrated promise in enhancing our understanding of PND by studying metabolites and proteins of interest in urine, blood, CSF, and saliva. In a broader sense, genetic variation could be considered a biomarker (most likely as a risk factor biomarker), and neuroimaging could theoretically provide any of the four types of PND biomarkers listed above. Advantages and disadvantages inherent to each of these sources are summarized in Table 12.2, and specific literature examples are given in Table 12.3.

Table 12.2 Biomarker Source, Advantages, and Disadvantages

Biomarker Source	Advantages	Disadvantages	Studies That Used This Biomarker Type
Urine	Obtained easily and noninvasively as long as there is no underlying pathology	Contamination possible; unclear relationship between urine marker concentrations and effect site concentrations within the brain	⁴^–⁶
Blood	Easily obtained pre-, intra-, and postoperatively; likelihood of finding markers is high as research has shown various PND markers are found in the serum	Minimally invasive procedure required; large dynamic range of analytes; due to the presence of the blood-brain barrier, unclear relationship between serum marker concentrations and effect site concentrations within the brain	⁹^–¹¹^, ¹⁴^–¹⁶
CSF	Allows access to various neuroinflammatory markers that play a role in PND	Patient recruitment may be challenging; pain during sampling; rare but severe risks including meningitis, post dural puncture headache Complex neural activity patterns within circuits are thought to give rise to cognition (rather than mere concentrations of factors within the brain or CSF); thus, unclear what CSF biomarker levels may tell us about the alterations in neuronal activity patterns that presumably underlie cognitive deficits in PND	²⁷^, ³¹
Saliva	Easily obtained, noninvasive	Not a proven source for markers that correlate with PND
DNA	Easily obtained	Unclear to what extent genetic markers may correlate with PND, given the likely multiplicity of causes for PND	³⁵^, ³⁶
Neuroimaging biomarker (i.e., functional or structural MRI)	Noninvasive	Patients with MRI-incompatible implants (i.e., pacemakers) cannot be scanned; claustrophobia is a contraindication; time consuming, and expensive. fMRI has limited temporal and spatial resolution, and only measures changes in hemoglobin oxygenation (which are themselves an indirect proxy for neuronal activity)	⁴⁹^, ⁵⁰
Intraoperative physiologic monitoring (i.e., processed or raw EEG, cerebral oximetry, cerebral blood flow monitoring, etc.)	Easily obtained, relatively risk free	Relevant commercial monitor may not be available at all institutions/locations; poor sensitivity/specificity, although EEG monitoring may⁴¹ or may not⁴⁰ alter PND risk at a population level	⁴⁰^, ⁴¹^, ⁵¹

Table 12.3 Specific Biomarkers

Specific Biomarker	Statistical Relationship to POCD	POCD Definition	Time Point(s) at Which POCD Measured	Time Point(s) at Which Biomarker Measured	Study Size and Patient Population	Reference(s)
Urine 8-isoprostane	POCD rate 29.2% (n = 21) in 7 days U8-isoPG:Cr higher at 1, 2, and 7 days (p < 0.05) No significant difference in U8-isoPG:Cr levels at baseline, 1, and 2 days (p < 0.05) U8-isoPG:Cr POCD at 7 days was higher than control (p = 0.01)	Changes in neuropsychological tests administered associated with impairments in daily functioning	Day before surgery 7 days after surgery Tests included: Mini-Mental State Examination (MMSE) Hopkins Verbal Learning Test–Revised (HVLT-R) Brief Visuospatial Memory Test–Revised (BVMT-R) Benton Judgment of Line Orientation Test Digit Span Test Symbol-Digit Modalities Test HVLT-R Delayed Recall Test HVLT-R Recognition Discrimination Index BVMT-R Delayed Recall Test BVMT-R Recognition Discrimination Index Trail-Making Test Verbal Fluency Test	Midstream morning urine collected before and after surgery Day 1 Day 2 Day 7	72 patients 65 years and older had elective general or middle orthopedic surgery	⁴
High-mobility group box 1 (HMGB1) and IL-6	POCD rate 34% 1 week POCD higher serum HMGB1 levels at day 1 (12.15 ± 3.12 vs. 9.91 ± 3.15 ng/ml, p = 0.021) and day 3 (11.04 ± 2.88 vs. 8.52 ± 3.31 ng/ml, p = 0.011) POCD higher IL-6 levels at 6 hours (51.18 ± 15.22 vs. 39.20 ± 14.32 pg/ml, p = 0.009) and day 1 (41.59 ± 11.08 vs. 33.81 ± 11.42 pg/ml, p = 0.026) Serum values of IL-6 at 6 hours, HMGB1 at day 1, and education correlations with z-scores HMGB1 day 3 and IL-6 6 hours were independent risk factors	z-score ≥ 1.96 (Recommended by ISPOCD)	1 week after surgery Tests included: Hopkins Verbal Learning Test–Revised (HVLT-R) Brief Visuospatial Memory Test–Revised (BVMT-R) Trail-Making Test (including Parts A and B) Benton Judgment of Line Orientation Digit Span Test, reflecting concentration of attention Symbol-Digit Modalities Test HVLT-R and BVMT-R Delayed Recall Test HVLT-R and BVMT-R Discrimination Index, items Verbal fluency test	Preoperatively 6 hours Day 1 Day 3 postoperatively	53 patients 60 years and older undergoing gastrointestinal surgery	¹⁵
CRP 1059G/C SNP and SELP 1087G/A SNP	Minor alleles of the CRP 1059G/C SNP (OR 0.37, 95% CI 0.16 to 0.78; p < 0.013) and SELP 1087G/A SNP (OR 0.51, 95% CI 0.30 to 0.85; p < 0.011) associated with reduction in POCD European Americans (n < 443) The absolute risk reduction POCD was 20.6% carriers CRP 1059C allele and 15.2% for carriers SELP 1087A allele.	Decline from baseline score of at least 1 standard deviation for one or more of the four domain scores at 6 weeks after surgery	Day before surgery 6 weeks after surgery Tests included: Short Story module of the Randt Memory Test Modified Visual Reproduction Test from the Wechsler Memory Scale Digit Span subtest of the Wechsler Adult Intelligence Scale–Revised (WAIS-R) exam Digit Symbol subtest of the WAIS-R Trail-Making Test (Part B)	Before induction of anesthesia Before aortic cross-clamp release 10 min after cross-clamp release At end of CPB At end of surgery C-reactive protein (CRP) levels were measured before induction of anesthesia 4.5 hours 24 hours 48 hours after cross-clamp removal	513 patients undergoing CABG surgery with cardiopulmonary bypass	³⁶
Beta-amyloid and C-reactive protein	POCD rate 44% POCD higher MELD scores (25.36 ± 7.03 vs. 16.21 ± 3.77, p = 0.000). POCD more Child-Pugh class C (71.4%, p = 0.007) POCD received more blood (p = 0.012) POCD CRP and β-Ap spike in serum concentration 24 hours after surgery (CRP: 276.1 ± 109.45 vs. 159.94 ± 53.79 ng/ml, p < 0.01) (β-Ap: 164.96 ± 70 vs. 74.9 ± 15.13 pg/ml, p < 0.01)	One standard deviation decline in two of the five neuropsychiatric tests	A single session before surgery The seventh day after surgery Tests included: MMSE Verbal fluency test (VFT) Digits span tests (DSTs) Item memory (IM) Source memory (SM)	Prior to induction of general anesthesia 30 min after the start of anhepatic phase 3 hours after new liver reperfusion 24 hours after surgery	25 patients undergoing liver transplantation	¹⁶
S-100 protein	48 cases POCD (95% confidence interval: 37.5–58.5) POCD higher S-100 (median 0.24 ng/ml; range 0.01–3.3) without POCD (median 0.14 ng/ml; range 0–1.34 ng/ml) 30 min postop (p = 0.01) Preoperatively, median S-100 0.02 ng/ml (range 0–0.2). No age dependency (p < 0.55) POCD higher S-100 No POCD increase in S-100 (p < 0.01). Post vascular surgery max S-100 occurred 30 min postop median = 0.65 ng/ml (range 0.01–1.57) with POCD (n = 9), and 0.21ng/ml (0.02–0.29) without POCD Post abdominal surgery, max S-100 with POCD (n = 21) 0.21 ng/ml (range 0.01–2.07) compared to 0.1 ng/ml (range 0–0.86) without POCD Urological surgery with POCD (n = 16) max S-100 of 0.22 ng/ml (range 0.02–0.05), and without POCD (n = 21) S-100 was 0.14 (0–1.1)	Decline of more than 10% in neuropsychological test results	Day 1 Day 3 Day 6 Tests included: The Digit Symbol Substitution Test (DSS) of the Wechsler Adult Intelligence Scale Concentration Endurance Test Number connection test	Preoperatively 0.5 hours 4 hours 18 hours 36 hours postoperatively	120 patients, scheduled for elective abdominal, vascular, urological (no transurethral resection of prostate) or trauma surgery (exclusive of hip or knee joint replacement)	⁹
Apolipoprotein E4 (APOE)	POCD rate 54.3% at 6 wks POCD rate 46.1% at 1 year No difference in POCD with or without APOE4 allele (56.6 vs. 52.6%; p < 0.58) Continuous cognitive change score (mean ±SD) similar between groups (APOE4: 0.05 ± 0.27 vs. non-APOE4: 0.07 ± 0.28; p = 0.53) POCD 45.9% of APOE4 POCD 46.3% of non-APOE4 (p = 0.95) Cognitive score was similar (APOE4: 0.08 ± 0.27 vs. non-APOE4: 0.05 ± 0.25; p = 0.39) Biomarker not associated APOE4 genotype or cognition at 6 wks or 1 yr	Decline of 1 SD or more in at least one of the four domains. (1) verbal memory, (2) abstraction and visuospatial orientation, (3) visual memory, and (4) attention and concentration	Testing was conducted at baseline 6 weeks 1 year after surgery Tests included: Randt Memory Test The Digit Span subtest of the Wechsler Adult Intelligence Scale–Revised Test Modified Visual Reproduction Test from the Wechsler Memory Scale The Digit Symbol subtest of the Wechsler Adult Intelligence Scale–Revised The Trail-Making Test (Part B)	Baseline preoperatively 4.5 hours 24 hours 48 hours postoperatively	394 patients older than 55 years undergoing major elective noncardiac surgery	³⁷
Fibrinopeptide A (FPA)	33 cases POCD 58 polypeptides (p < 0.05) Three differential peaks, 1462.81 m/z, 1536.45 m/z, and 2659.95 m/z, which had higher strength were subject to MALDI-TOF/TOF to reveal their identity. Top scores were 86, 45, and 25 of differential peaks at 1462.81 m/z, 1536.45 m/z, and 2659.95 m/z Mass number of 1462.81 was identified as fibrinopeptide A (FPA), only protein with score > 60 (p < 0.05)	Z test scores.	1 day before surgery 7 days after surgery Tests included: MMSE Hopkins Verbal learning test (HVLT-R) Brief visuospatial memory test (BVMT-R) Trail-Making Test Judgment of line orientation test (the models provided by Psychological Assessment Resources, Inc. [PAR]) Digit span test Letter-digit coding test Word recall test Word interference test Figure recall test Figure interference test Verbal fluency test (VFT)	1 day before surgery 7 days after the surgery	68 elderly patients age ≥65 years for arthroplasty surgeries	¹⁸
S-100β	POCD rate 45.9% at 1 day POCD rate 8.1% at 7 days POCD 1 day higher levels of IL-6 at 6 hours (135 ± 32 pg/ml vs. 91 ± 29 pg/ml, p < 0.05) POCD S-100β at 1 hour (1872 ± 385 pg/ml vs. 1289 ± 143 pg/ml), p < 0.05 No significant change detected in TNF-α, IL-1, or CRP	A decline of more than 10% in neuropsychological tests	Preoperatively Day 1 Day 3 Day 7 Tests included: Digit Symbol Substitution Test of the Wechsler Adult Intelligence Scale Concentration Endurance Test d2 Number connection test	Preoperatively 1 hour 6 hours postoperatively	42 patients older than 60 yr scheduled for total hip replacement were enrolled, and 37 patients completed follow-up	¹⁰
Protein S-100B	Pre-op, mean serum S-100B 0.17 ± 0.38 μg/L. End of CPB, increased to mean peak value of 1.66 ± 0.96 μg/L (t = 12.51, p < 0.001 vs. baseline) and decreased to 0.30 ± 0.22 μg/L (t = 2.52, p < 0.007) 48 hours after CPB S-100B level end of surgery did not correlate with cognitive change scores 48 hours after CPB correlated decrease of SKT visual memory scores at 2 to 4 days postoperatively r = −0.30 p < 0.002	Statistically significant change from baseline performance level on a major test	Prior to procedure 2–4 days after 3 months after CABG Tests included: Syndrom Kurztest (SKT)	Baseline End of surgery 48 hours afterward	106 patients scheduled for elective CABG were prospectively enrolled	⁵²
CSF Biomarkers	POCD rate 24.6% at 7 days POCD higher CSF IL-1β, Tau/Aβ1–42, pTau/Aβ1–42, and lower Aβ1–42 compared to no POCD No difference in CSF levels of Tau, pTau, BDNF, and IL-6 between POCD and Non-POCD (p < 0.05). POCD higher MDA compared with Non-POCD at 7 days (p < 0.05)	Decline of more than 10% in neuropsychological test results	Preoperatively 7 days postoperatively Tests included: Digit Symbol Substitution Test of the Wechsler Adult Intelligence Scale Concentration Endurance Test d2 Number connection test	Measured preoperatively Perioperative plasma levels	61 patients between 65 to 85 years of age who underwent total hip replacement surgery	²⁸
Perioperative NO products	POCD rate 40% and 7% control at 4 days (p < 0.01) POCD rate 53% and 23% control at 6 wks (p < 0.03) New cognitive deficit 4 days postop had larger plasma NOx at each perioperative time points (p < 0.05 for each time point) sensitivity, specificity, ppv, and npv T1 33%, 91%, 71%, and 68% T5 40%, 100,%, 100%, and 73% T6 37.5%, 100%, 100%, and 70.5%	Deficit in one or more cognitive domain(s)	Day before surgery 4 days and 6 weeks Patient spouses studied as controlsTests included: Rey Auditory Verbal Learning Test (RAVLT). Trail-Making Test Parts A (TMT A) and B (TMT B) (Halstead-Reitan Neuropsychological Test Battery) Purdue Pegboard test Controlled Oral Word Association Test Digit Symbol Substitution Test	Before surgery (T1) At the end of surgery (T2) 1 hour (T3) 2 hours (T4) 4 hours postoperatively (T5)	42 patients aged 40– 85 yr undergoing inpatient laparoscopic cholecystectomy	¹²