25 Junio 2015
Por Mihai Paduraru
This is a clearly presented paper with specific intention and methodology. The authors identify where there are limitations in their research and suggest how this can be improved upon. However, as has already been identified by A. G. Renehan in his commentary in BJS (2014), a key flaw in the validity of the findings of Jung et al, is the lack of attention paid to the impact of confounding factors over a ten year period, the fact being that the first trial was rigorous as an RCT, with different aims, and the second study is retrospectively analyzing data using the same patient group from the original trial, can lead to confusion. He suggests that the statistical methodology used was insufficient in some cases and lacking in others to support the findings and therefore, the conclusion. Renehan has already commented (BJS 2014) on the insufficiency of some of the statistical analysis and lack of attention to confounding factors.
There are some further points to make in addition to Renehan’s comments. Firstly the high number/proportion of patients from the RCT that were excluded from the data analysis of this second study. The authors give a rationale for this but the end result is a smaller sample size than is considered valid for the results to be significant. Since this is a retrospective cohort study, this is an important point. If all patients had been included, then it has to be considered that the results might have been different. One of the reasons stated for the exclusion of certain patient data was the lack of accuracy of the data base used and the inconsistency of recording, compounded by the changes in reporting over the 10 year period. Again, this factor cannot be ignored and throws into question the accuracy of the patient case records.
Secondly, the higher proportion of patents in the non MBP group with stage III tumor is not discussed sufficiently. The difference between stage II and III is that the cancer has spread to nearby lymph nodes in the latter stage. The expectation then would be for this group to have a lower cancer specific survival rate. Furthermore, each stage has sub classifications (a,b and a,b,c respectively) and we do not know, from the paper, if the stages are pre- or postoperative, nor if the study takes into account the number of nodes in relation to the long term oncological results for stage III (stage IV being excluded). The authors do not provide any data as a reference with regard to overall colon cancer survival for patients undergoing resection with type II and type III tumor stage.
Thirdly, the ‘surgeon factor’ is not discussed at all as a factor. The proficiency of the surgeon has been demonstrated in other studies to play an important role in trial outcomes. This could be another confounding factor, especially with regard to practice 10 years ago and with the debate to prepare mechanicaly or not the colon.
Another confounding factor not accounted for and already mentioned by Renehan, are other/additional kinds of oncological treatment – adjuvance.
In agreement with Renehan, the study has some use in generating further research, but these new studies need to be more thorough in identifying and accounting for the confounding factors.
Overall this paper contributes no hard evidence that MBP could improve colon cancer survival rates, with a low level of evidence for the moment and with high level of bias.
On the other hand, if this study had been a strong one, with a high level of evidence and recommendation (Harbour and Miller), to give it validity, what would the implications for practice be: To choose the option of reducing postoperative complications, including mortality, by not undertaking MBP (as recommended and proven by ERAS); or to choose long term better oncological results re-instating the classical MBP for colon resections? Such a recommendation poses a real ethical dilemma and needs much more concrete evidence than this study offers before we take a step back to the future.