Editorial: smoking status and 5‐year survival in patients with colorectal cancer

Smoking and tobacco have been associated with lung, colon, head and neck, pancreatic, and bladder cancers, and cardiopulmonary disease. Smoking is expected to result in a billion deaths by 2100 and a vigorous campaign against smoking will be a true ‘moonshot’ against cancer. The question raised by Sharp and colleagues is whether continued smoking, after a colon cancer diagnosis, makes one more likely to die of colon cancer. Sharp and colleagues performed a multivariate analysis, using multiple imputation to handle missing data. They showed that, after adjusting for five covariates and stratifying by stage and age, smoking increases the risk of dying of colon cancer after treatment with surgery alone, but not if patients also receive chemotherapy within a year of diagnosis (surgery and chemotherapy group) or are treated with chemotherapy alone. Surgery alone is the mainstay of treatment for early stage (≤ low-risk stage II) colon cancer, and chemotherapy alone is used to treat stage IV. Surgery and chemotherapy together are used for adjuvant treatment of stage III and for some high-risk stage II patients, but the authors’ category of surgery and chemotherapy is even broader. They define it as receipt of chemotherapy within a year of surgery. Therefore, it also includes patients whose cancers relapsed within a year and who were then treated with chemotherapy, and patients with de novo stage IV cancer who underwent surgery for obstruction. It is challenging to make sense of these results. Why should smoking worsen cause-specific survival following surgery alone, but not with chemotherapy, and not with chemotherapy alone? We suggest three possibilities. The first is that there is residual confounding in the analysis. In this study, current smokers were younger, more likely to be male, with higher deprivation scores, more likely to have poorly differentiated histology, and to have presented at higher stage. The authors adjusted for these factors, but it is possible that residual confounding remained. One potential factor is that patients for whom adjuvant therapy was indicated and who were also smokers may have been less likely to receive adjuvant therapy (e.g. because providers view them as poorly compliant, or that they might not tolerate chemotherapy, or that their competing risks were great and overall prognosis was poor). These patients might otherwise have been in the surgery and chemotherapy group but, without adjuvant therapy, were moved to the surgery-only group. Furthermore, patients with very advanced disease who presented with obstruction and/or bleeding, might have received surgery alone for palliation. Thus, the surgery-only group may have become enriched with people destined to do poorly. Although the authors made attempts to overcome this by stratifying by stage, they may need to further separate low from high-risk stage II to overcome this concern. The second possibility is that smokers with relapsed metastatic cancer may be less likely than nonsmokers to have that detected within a year. After a patient is treated for early stage colon cancer with surgery alone, the patient is advised to follow a strict surveillance schedule for the first 5 years. If recurrence is found, chemotherapy often follows. In this cohort, if recurrence was found within a year, the patient would have moved from the surgery-only to surgery and chemotherapy group. If a smoker, because of higher deprivation and other barriers to healthcare, had missed the 12-month surveillance and was diagnosed and treated later, he/she would have been categorised as surgery-only, instead of surgery and chemotherapy. While this might have little consequence for that patient, any imbalance in adherence to surveillance could have worsened outcomes for the surgeryonly group. This could be an artefactual reason why we might see decreased survival among smokers in the