The effect of smoking, obesity and diabetes on recurrence‐free and overall survival in patients with stage III colon cancer receiving adjuvant chemotherapy

Abstract Background The association between smoking, diabetes and obesity and oncological outcomes in patients with stage III colon cancer treated with surgery and adjuvant chemotherapy is unclear. Aim To evaluate whether smoking, obesity and diabetes are associated with the disease‐free survival and overall survival rates of patients with stage III colon cancer who have received adjuvant chemotherapy. Methods Patients were selected from the prospectively maintained Australian Cancer Outcomes and Research Database (ACCORD). All stage III colon cancer patients who received adjuvant chemotherapy between January 2003 to December 2015 were retrospectively analyzed. The three primary exposures of interest were smoking status, body mass index (BMI) and diabetic (DM) status. The primary outcomes of interest were disease‐free survival (DFS) and overall survival (OS). Results A total of 785 patients between 2003 and 2015 were included for analysis. Using Kaplan‐Meier survivorship curves, there was no association between OS and smoking (P = .71), BMI (P = .3) or DM (P = .72). Similarly, DFS did not reveal an association with smoking (P = .34), BMI (P = .2) and DM (P = .34). Controlling for other covariates the results did not reach statistical significance in adjusted multiple regression models. Conclusion Smoking, obesity and DM were not shown to influence DFS or OS for patients with stage III colon cancer who have received adjuvant chemotherapy.


| INTRODUCTION
Colorectal cancer (CRC) is the third most commonly diagnosed cancer in Australia and the second highest cause of cancer death. 1 Over twothirds of patients diagnosed with CRC will have a colonic primary, and of these, approximately 24% will have locoregional nodal involvement [American Joint Committee on Cancer (AJCC) stage III] at diagnosis. 2,3 Over the past three decades, there has been a significant improvement in the survival rates for patients diagnosed with stage III colon cancer, owing largely to the development of effective adjuvant systemic treatments. Earlier trials demonstrated a survival advantage with adjuvant fluorouracil (5-FU) and levamisole over observation alone. 4 More recently, the addition of oxaliplatin to a fluoropyrimidine has been shown to offer further survival benefit and become accepted as a standard of care. 5 Currently, the 5-year survival rate for patients with stage III colon cancer is estimated to be 71%. 6 Smoking, obesity and diabetes mellitus (DM) are patient-related factors that are associated with an increased risk of developing CRC. [7][8][9] It has been suggested that smoking may increase the risk of colon cancer by the induction of tumor angiogenesis as well as by the inhibition of apoptosis. 10 Contrastingly, obesity and DM may promote carcinogenesis through chronic systemic inflammation, hyperinsulinemia and increased levels of circulating insulin-like growth factors. 8,9 Therefore, it is not unreasonable to consider the impact of these factors on the disease-free (DFS) and overall survival (OS) of patients who have been treated for colon cancer. Furthermore, there is some evidence suggesting that smoking, obesity and DM may reduce the efficacy of chemotherapeutic agents. [11][12][13] Nicotine has been found to decrease the antiproliferative and pro-apoptotic effects of 5-FU in in vitro CRC cells, while hyperglycemia has been associated with the diminished efficacy of a number of chemotherapy agents in animal models. 11,13 Furthermore, visceral obesity may adversely influence pharmacokinetics and drug-volume distribution resulting in potentiating side effects and subtherapeutically dosing systemic treatments. 12 There have been several studies that have investigated the impact of patient-related factors on DFS and OS; however, the majority of these draw data from heterogeneous cohorts with respect to both tumor site (colon and rectal cancers) and disease stage. [14][15][16] Studies evaluating smoking and survival for stage III colon cancer patients receiving chemotherapy have shown mixed results and may lack broader generalizability to other populations from around the world. 17,18 The aim of this retrospective Australian cohort study is to determine whether smoking, obesity and DM are associated with DFS and OS rates of patients with stage III colon cancer who have received adjuvant chemotherapy.  (25-29.9) or obese (>30) based on BMI ranges described by the World Health Organization. Diabetic status was classified as present or absent irrespective of type. OS was defined as time of diagnosis to death and DFS was the time from surgery to recurrence at any site. The primary outcome was the association between smoking status, BMI or DM on DFS and OS in patients with stage III colon cancer.

| Statistical analysis
Patient characteristics by overall survival status were summarized as mean (SD), median (minimum, maximum) or number (%) according to type and distribution. Quality of the survival cohort was assessed using the Clarke index, a ratio measure of the sum of observed observation time/potential observed time, based upon overall survival. 19 Table 1 and further subdivided according to primary exposure in Table 2

| Smoking status
Smoking status modeled as a three-level nominal variable (never, ex and current) was not associated with overall survival when unadjusted (P = .71, Figure 1A) or adjusted using a multivariable regression model. As a binary model, there was no difference in mean restricted survival time at 2 or 5 years (

| Body mass index
There was no evidence for a difference in survival with increasing BMI. The Kaplan-Meier survivorship curve and log rank test (P = . 30) are presented in Figure 1B. BMI. The Kaplan-Meier survivorship curve and log rank test (P = 0.20) are presented in Figure 2B. The PH assumption was violated for multivariable Cox regression model. Both the IPW hazard ratio and RMST indicate no evidence for an association between BMI and DFS (Table 3).  Figure 1C.

| Diabetes mellitus
The multivariable Cox proportional hazards model and the IPW mode both indicated no association between DM and overall survival. The PH assumption was met in all models. There was no difference in RMST between exposure levels at either 2 or 5 years (Table 3).   they did not find an association with recurrence for either cancer type. 33 The authors suggested that less aggressive treatment of diabetic patients as well as a potentially reduced response to chemotherapy may have led to this reduction in survival. Determining the true impact of diabetes on colon cancer survival and recurrence is complex as much of the data is retrospective and lacks objective markers of disease severity and levels of hyperglycemia, such as glycosylated hemoglobin levels (HbA1c). There is also some evidence showing that treatment with metformin may offer a protective effect and reduce CRC-specific deaths. It has been suggested that metformin reduces the circulating levels of insulin like growth factors, as well as the synthesis of certain proteins that are key in the production of malignant cancer cells and angiogenesis. 33 Therefore, documentation and analysis of diabetic treatment regimens of study cohorts may need to be examined closely to assess whether these variables may impact on the survival outcomes.
This study is limited by its retrospective nature and presence of missing data. Imputation techniques to account for missing data were not performed due to their inherent problems in adequately accounting for selection bias. This study is also limited by the lack of objective markers of disease severity of diabetes such as glycosylated hemoglobin, and exposure severity of smoking such as a pack year history. Our study's strength is the size of cohort and the homogenous cohort with the inclusion of only patients with stage III colon cancer who received postoperative chemotherapy.

| CONCLUSION
In our Australian cohort of patients with stage III colon cancer who received adjuvant chemotherapy, there was no association between smoking, obesity or DM and DFS or OS outcomes.

CONFLICT OF INTEREST
The authors have stated explicitly that there are no conflicts of interest in connection with this article.

PATIENT CONSENT STATEMENT
This was a retrospective audit involving access to existing medical records and the research personnel reviewing these records would normally have access to these records. Therefore, patient consent was waivered.

DATA AVAILABILITY STATEMENT
The data that support the findings of this study are available on request from the corresponding author. The data are not publicly available due to privacy or ethical restrictions. All de-identified data accessed from the ACCORD database has been confidentially stored and is accessible upon reasonable request to the principle author.