Study Design
This retrospective analysis was performed at the Affiliated Hospital of Xuzhou Medical University (Xu Zhou, China) from December 2016 - July 2017. The clinical research ethics committee of the Affiliated Hospital of Xuzhou Medical University approved this study (Certification No. XYFY2018-KL035-01 approved date: 26/7/2018), which was registered at https://register.clinicaltrials.gov (number: NCT03703973; Principal investigator: JLC; date of registration: 12/10/2018). This study was conducted following the 1964 Helsinki Declaration. As all follow-up data and blood samples in this study were derived from our previous study, which analyzed the correlation between anosmia and postoperative cognitive dysfunction, so this study obtained waiver of Informed Consent. In the previous study, informed consent was obtained from all participants. It has been clearly pointed out in the informed consent form that blood samples and other data from the subjects might be used to futher explore postoperative cognitive impairment. All patient data remained confidential, and study populations were coded to ensure anonymity. This manuscript was compiled in strict adherence to applicable STROBE guidelines. In total, blood samples and neuropsychological test results from 196 patients were analyzed in this study.
Subject Enrollment
From December 2016 to July 2017, elderly patients ≥60 years of age, referred for major non-cardiac and non-neurological surgical procedures under general anesthesia with an anticipated 5-day hospital stay, were screened for study inclusion. Patients were excluded on meeting any of the the following criteria: (1) significantly impaired hearing, vision, or motor skills (hemiplegia, etc.); (2) abnormal personality characteristics or psychological illnesses with the potential to interfere with the completion of baseline neuropsychological testing; (3) any history of neurological disorders or diseases including depression, schizophrenia, Parkinson's disease, multiple sclerosis, or senile dementia; (4) any surgical procedures or severe trauma within the last 12 months; (5) alcoholism, drug dependence, or severe physical illnesses; (6) vital organ dysfunction; (7) preoperative Mini-Mental Status Exam (MMSE) scores classified as follows: < 17 points for illiterate individuals, < 20 points for individuals with a primary school education, or < 34 points for individuals with secondary education or higher[12,13]; (8) a Geriatric Depression Scale (GDS) grade >2; or (9) an unwillingness to complete the study procedures or to comply with protocols.
Neuropsychological Testing
Individual patients underwent two rounds of standard neuropsychological testing, with baseline testing performed one day before surgery and follow-up testing conducted one-week post-surgery. Testing was conducted in general care wards in a quiet room. The appropriate neuropsychological tests were selected based on two International Studies of Post-operative Cognitive Dysfunction (ISPOCD 1 and 2) [1, 14, 15]. In total, testing included 9 tests with 11 subscales covering cognitive domains associated with memory, psychomotor speed and dexterity, physical motor speed, attentional capacity, and perceptual-spatial function (Table 1). Further details regarding this testing approach have been published previously[16].
To facilitate measurements of dNCR, 30 volunteers not undergoing anesthetization or surgery were recruited using the same criteria as above. These volunteer controls underwent the same neuropsychological testing at identical time intervals, with testing performed by the same research team.
LTL Measurements
Patient blood samples were collected one day before surgery and stored at -80℃ for subsequent analysis. These samples were not subject to any other testing before this study. LTL values were computed using Cawthon's formula [17]. Briefly, for each sample, LTL was calculated by assessing the ratio of the telomere repeat (T) copy number to that of a single-copy gene (S) via qPCR using the SYBR Green dye (T/S ratio). All samples were analyzed two times for statistical analyses by operators blinded to patient sample details.
dNCR detection
The study definition of dNCR used for these analyses was based on the ISPOCD 1 definition from two prior reports[1, 14, 15], with dNCR incidence being calculated as follows: (ΔXi–)/SD.. These changes (ΔXi) were calculated by subtracting preoperative score values from post-operative values for each task. The averaged difference value for controls () was considered indicative of a systematic error and was computed similarly. Practice effects were then accounted for by subtracting ΔXC from ΔXi, and the resulting score was divided by the standard deviation for the change observed in the control group (SD.) as a means of controlling for expected variability. When participants exhibited a Z-score ≤ -1.96 on a minimum of 2 different neuropsychological tests out of 10 possible tests.
Statistical Analyses
Outlier detection and analyses of distribution normality were conducted through univariate analyses. Continuous variables (age, height, BMI, years of education, follow-up duration, MMSE scores, blood loss, anesthesia duration, surgical duration, total fluid administration, LTL) were compared with t-tests or Wilcoxon rank tests. In contrast, categorical variables (prior history of the disease, surgery type, sex) were compared using Fisher's exact or chi-square tests.
Relationships between LTL and dNCR incidence were examined using multivariate logistic regression analyses through a backward stepwise approach. Covariates in these models included established risk factors associated with dNCR, including age, sex, and level of education [1, 15]. The relationship between LTL and dNCR incidence was further explored with a ROC curve. Youden's index was computed to establish the optimal cut-off value for separating patients into groups with longer and shorter LTL values, after which dNCR incidence rates were compared in these groups. SPSS v 22.0 (IBM, NY, USA) was used for all statistical testing.
While this was a prospective analysis, sample size calculations were performed as a component of the study design process using PASS (v 11.0; NCSS, USA) based on two-independent proportions with odds ratios (OR). Based on prior studies of PND and correlations between this condition and changes in gene expression and olfactory dysfunction, dNCR was predicted to affect 20% of patients, with a risk of cognitive decline for individuals with shorter LTL values (T/S ratio < median T/S ratio) being 2-fold higher (OR=3.0) as compared to individuals with longer LTL values (T/S ratio > median T/S ratio) [11]. At a 0.05 significance level and 0.80 power, a minimum of 190 patients would thus need to be enrolled to provide this study with sufficient statistical power.