We thank Lacomis et al.1  for their interest in our study published in Anesthesiology,2  and we appreciate the opportunity to respond to their comments.

Their first question is whether the finding that the increased concentration of interleukin-6 in the blood at 1 h after surgical incision may be predictive of delayed neurocognitive recovery on postoperative days 5 to 7 is a true phenomenon. It is known that one can find false-positive results/relationships if multiple comparisons are performed on many sets of data. This issue is applicable to numerous situations including the data generated by conventional methods, such as our study, and by bioinformatics approaches. One method to reduce the possibility of false positive findings is to use an appropriate statistical method. In our case, the serum concentrations of 14 biomarkers including interleukin-6 were analyzed by two-way repeated-measures ANOVA followed by Bonferroni multiple comparison correction. There was a group (comparison between patients with and without delayed neurocognitive recovery) and group × time interaction effect on the serum interleukin-6 concentrations (see fig. 2 of the original publication2 ). The actual P value was 0.002 for the comparison between delayed neurocognitive recovery group and no-delayed neurocognitive recovery group at 1 h after surgical incision. This P value was lower than 0.01, the cutoff value set for significance for the comparison of five time points (i.e., α/n; 0.05 ÷ 5 = 0.01). One could use a much more stringent alpha threshold by considering the combination of all 14 biomarkers and five time points, deriving a total of 70 inferences (i.e., 0.05 ÷ 70 = 0.00071). Such a consideration is often viewed as being too conservative. Our chosen threshold was the consideration of a balance between type-I and type-II errors. In addition, univariate and multiple logistic regression analyses suggest that the interleukin-6 concentrations at 1 h after surgical incision are a predictive factor for delayed neurocognitive recovery.2  The odds ratio and its CI for this parameter were calculated based on all data and the consideration of it as a continuous predictor (i.e., we did not artificially dichotomize the interleukin-6 concentrations). We derived our conclusion that the serum interleukin-6 concentrations at 1 h after surgical incision may be predictive of delayed neurocognitive recovery based on these analyses and defined conditions. Readers may apply different thresholds or considerations to evaluate our results. Nevertheless, we feel that our finding may be significant because the time point (1 h after surgical incision) may be early enough for clinicians to consider strategies to reduce the occurrence of delayed neurocognitive recovery. Of course, additional studies are necessary to determine the predictive value of interleukin-6 concentrations at 1 h after surgical incision for postoperative neurocognitive disorder.

The second comment from Lacomis et al. is a request to see the data of blood biomarkers arranged based on group assignment to the propofol and sevoflurane groups. These data are shown in fig. 1 of this reply. These data were analyzed by two-way (between-group comparisons with the time and group as two factors to be analyzed), repeated-measures ANOVA followed by Bonferroni multiple comparison correction (α/n; i.e., 0.05 ÷ 5 = 0.01 as the corrected cutoff value). We noted that the concentrations of an oxidative stress biomarker, myeloperoxidase, were different between patients in the propofol and sevoflurane groups (P = 0.012 for the group effect). However, there was no group × time interaction effect on the myeloperoxidase concentrations (P = 0.778; fig. 1). The reason and significance for group difference between patients in the propofol and sevoflurane groups are not known. However, this difference might have existed before anesthesia, suggesting that the difference is not caused by anesthetics applied. The concentrations of other biomarkers throughout the perioperative period were not different between patients in the propofol and sevoflurane groups.

Fig. 1.

Comparison of serum cytokines, oxidative stress biomarkers, or immune modulators between patients in propofol group and sevoflurane group. The data are presented as the means ± SD (n = 26 – 6). The P values are the results from two-way repeated-measures ANOVA with time and group as two factors to be analyzed.

Fig. 1.

Comparison of serum cytokines, oxidative stress biomarkers, or immune modulators between patients in propofol group and sevoflurane group. The data are presented as the means ± SD (n = 26 – 6). The P values are the results from two-way repeated-measures ANOVA with time and group as two factors to be analyzed.

Close modal

The authors declare no competing interests.

1.
Lacomis
CD
,
Williamson
GC
,
McHugh
SM
:
Anesthetics and postoperative cognition: Comment.
Anesthesiology
2021
;
135
:
767
8
2.
Li
Y
,
Chen
D
,
Wang
H
,
Wang
Z
,
Song
F
,
Li
H
,
Ling
L
,
Shen
Z
,
Hu
C
,
Peng
J
,
Li
W
,
Xing
W
,
Pan
J
,
Liang
H
,
Zhou
Q
,
Cai
J
,
He
Z
,
Peng
S
,
Zeng
W
,
Zuo
Z
:
Intravenous versus volatile anesthetic effects on postoperative cognition in elderly patients undergoing laparoscopic abdominal surgery.
Anesthesiology
2021
;
134
:
381
94