To the Editor,

We would like to thank Chang et al.1 for identifying an error in our manuscript entitled “Anesthetic technique and cancer outcomes: a meta-analysis of total intravenous versus volatile anesthesia”.2 We have re-checked all analyses in the manuscript and can confirm that this was an isolated error due to a data transcription error relating to the Oh et al. study3 within our recurrence-free survival analysis. As stated by Chang et al., when the reported estimates in Oh et al. are corrected, the pooled hazard ratio (HR) is now 0.87; 95% confidence interval (CI), 0.66 to 1.15; P = 0.32 (compared with the erroneously reported HR of 0.78; 95% CI, 0.65 to 0.94; P < 0.01). Nonetheless, although now not statistically significant, the effect estimate still suggests that total intravenous anesthesia (TIVA) may be protective, albeit with a wider confidence interval that now includes 1.

It is important to note that the Oh et al. study incorporates substantial heterogeneity (I2 > 60%; P = 0.02), so we performed a sensitivity analysis as we did for the other outcomes reported in the manuscript. When removing the Oh et al. study (as an outlier), as described in our methods, the effect estimate is 0.79 (95% CI, 0.62 to 1.0; P = 0.05) indicating a borderline protective effect with an acceptable inconsistency score (I2 = 40%; P = 0.16; see attached revised forest plot).

Figure
figure 1

Recurrence-free survival following sensitivity analysis

Full size image

A recent editorial4 related to the important question of whether anesthetic technique impacts cancer outcomes suggested that heterogeneity in magnitude of surgical procedure within study cohorts provided “context for divergent study outcomes”; with a protective effect by propofol-TIVA on cancer recurrence and survival more likely to be seen in patients having major cancer surgery. As discussed, the study by Yoo et al.5 has substantial heterogeneity in surgical magnitude. This study also contributes to the statistical heterogeneity in our meta-analysis. When this study is also removed from analysis, the pooled estimate is statistically significant (HR, 0.71; 95% CI, 0.59 to 0.86; P < 0.01) with no heterogeneity (I2 = 0%; P = 0.41), further supporting our conclusions.

A major limitation of our manuscript is that our findings are based on mostly observational studies. As stated in our conclusion, there is an imperative need for robust randomized-controlled trials in this area, which will help to strengthen the current evidence on impact (or lack thereof) of anesthetic technique on cancer outcomes.