Diabetes is a risk factor for colorectal cancer and several reports suggest worse cancer-specific outcomes in diabetes patients. Recent studies in multiple tumour types indicate metformin may positively impact on cancer-specific and overall survival. A population-based series of stage II colorectal cancer patients treated and followed from 2000 to 2013 were analysed for baseline characteristics, treatment, and outcomes. 1116 patients with stage II colon cancer were identified, 55.5% were male and median age was 70.9 years (range 20.5–101.2). The diabetes patients (21.6%,
A marked rise in the prevalence of diabetes has been noted in recent years [
The potential impact of diabetes medication on cancer outcomes remains an area of intense study. The multiple potential anticancer properties of metformin [
Here, we present data from a prospective database, where comprehensive data on a consecutive series of patients with colorectal cancer was entered by clinicians at four large cancer centres, allowing us to explore the impact of diabetes and diabetes medication on cancer presentation, treatment, and outcome.
Prospective data from a comprehensive colorectal cancer database from four metropolitan hospitals (Royal Melbourne Hospital, Melbourne Private Hospital, Western Hospital, and Western Private Hospital) were analysed. All patients underwent routine preoperative investigations, including computerised tomogram scans of chest, abdomen, and pelvis, with treatment and follow up based on standardised protocols. Data regarding patient status, including any recurrence or new primary cancer, was prospectively collected, with disease status documented at each follow-up visit.
Data fields examined related to initial colon cancer presentation, comorbidity including diabetes, cancer pathology, operative outcomes, adjuvant therapy use, cancer recurrence, and survival data. Baseline renal function was assessed in diabetic cohorts for underlying renal impairment. This study was approved by the Melbourne Health Human Research Ethics Committee.
We elected to focus on patients with stage II colon cancers to allow a more detailed analysis of clinicopathological characteristics and cancer-specific outcomes in a single cancer stage where there is little impact from chemotherapy. We excluded Stage I colon cancers due to the very low recurrence rate. We excluded node positive colon cancers due to the variation in delivery of adjuvant treatment (none versus fluoropyrimidine alone, versus oxaliplatin based combination treatment), and the potential confounding effect of diabetes and the frequently associated comorbidities, which might influence chemotherapy choice, chemotherapy dose, and chemotherapy completion rate. An initial search identified all patients diagnosed with stage II colon cancer between January 1, 2000, and December 31, 2013. The subset of patients with type 2 diabetes was then defined and analysed according to diabetes medication, confirmed from review of the hospital pharmacy database and individual patient files. Diabetes patients were compared with nondiabetes patients, and for diabetes patients the impact of individual antidiabetes medication was examined.
Statistical significance was analysed using SAS Enterprise Guide 6.1 (SAS institute Inc, Cary, NC, USA). Descriptive statistics including median and frequencies were used to describe the study population in each treatment category. Kaplan-Meier method was used to analyse overall survival. The Cox proportional hazards model was used for univariate and multivariate analyses of clinicopathological factors for recurrence free survival.
We identified 1121 patients diagnosed with stage II colon cancer; 55.5% were males and median age was 70.9 years (range 20.5–101.2). A total of 246 patients (22.0%) with type 2 diabetes were identified. Five patients (2.0%) were excluded from analysis of the impact of diabetes medication, as at the time of colorectal cancer diagnosis their diabetes treatment could not be reliably determined from pharmacy records and chart review due to inconsistencies in data. Diabetes patients were older than nondiabetes patients (median age; 74.0 year versus 69.6 years,
Diabetes medication for the 241 patients with type 2 diabetes included in the analysis is shown in Figure
Flow chart—study population. aMetformin with or without sulfonylurea; binsulin with or without sulfonylurea; cno diabetes medication; dsulfonylurea alone or other combinations; edipeptidyl peptidase-4 inhibitor, sitagliptin used in this one patient.
The impact of diabetes and diabetes medication on colon cancer pathology and presentation is shown in Tables
Clinicopathological features of nondiabetes versus diabetes patients.
All patients |
Patients without type 2 diabetes |
Patients with |
|
---|---|---|---|
Median age in years (range) | 70.9 (20.5, 101.2) | 69.6 (20.5–101.2) | 74.0 (26.5–93.3) |
Sex | |||
Female (%) | 497 (44.5%) | 400 (45.7%) | 97 (40.3%) |
Male (%) | 619 (55.5%) | 475 (54.3%) | 144 (59.7%) |
Site | |||
Right colon (%) | 533 (47.8%) | 407 (46.5%) | 126 (52.3%) |
Left colon (%) | 583 (52.2%) | 467 (53.4%) | 115 (47.7%) |
T4 staging (%) | 176 (15.8%) | 131 (15.0%) | 45 (18.7%) |
<12 lymph nodes examined | 275 (24.6%) | 208 (23.8%) | 67 (27.8%) |
Poor differentiation (%) | 225 (20.2%) | 170 (19.4%) | 55 (22.8%) |
Lymphovascular invasion | 266 (23.8%) | 211 (24.1%) | 55 (22.8%) |
MSI status (%) | |||
MSI-stable | 441 (39.5%) | 351 (40.1%) | 90 (37.3%) |
MSI-high | 114 (10.2%) | 92 (10.5%) | 22 (9.1%) |
Not done | 536 (48.0%) | 413 (47.2%) | 123 (51.0%) |
Unknown | 25 (2.2%) | 19 (2.2%) | 6 (2.5%) |
Emergency presentationa | 158 (14.2%) | 122 (13.9%) | 36 (14.9%) |
aEmergency presentation includes bowel perforation or bowel obstruction.
Pathologic features of diabetes patients: metformin and insulin treated subsets.
Metformin group |
Insulin group |
| |
---|---|---|---|
Median age in years (range) | 73.4 (51.1–93.3) | 71.8 (39.6–85.3) | 0.28 |
Sex | |||
Female (%) | 41 (41.4%) | 3 (12.5%) | 0.008 |
Male (%) | 58 (58.6%) | 21 (87.5%) | |
Site | |||
Right colon (%) | 59 (60%) | 14 (58%) | 1.0 |
Left colon (%) | 40 (40%) | 10 (42%) | |
T4 staging (%) | 17 (17.2%) | 4 (16.7%) | 1.0 |
<12 lymph node sampling | 28 (28.3%) | 7 (29.2%) | 0.09 |
Poor differentiation (%) | 25 (25.2%) | 2 (8.3%) | 0.7 |
Lymphovascular invasion | 21 (21.2%) | 4 (16.7%) | 1.0 |
MSI status (%) | |||
MSI-stable | 36 (36.4%) | 9 (37.5%) | 1.0 |
MSI-high | 18 (18.2%) | 4 (16.7%) | |
Not done | 43 (43.4%) | 11 (45.8%) | |
Unknown | 2 (2.0%) | 0 | |
Emergency presentationa | 11 (11.1%) | 6 (25.0%) | 0.09 |
aEmergency presentation includes bowel perforation or bowel obstruction.
The operative morbidity, mortality, and adjuvant treatment details of diabetes patients, including metformin and insulin treated subsets, versus nondiabetes patients (
Operative morbidity and mortality, adjuvant treatment commenced, and completed details of nondiabetes versus diabetes patients, including metformin and insulin treated subsets.
No diabetes |
Diabetes |
Metformin group |
Insulin group |
|
---|---|---|---|---|
Renal function | ||||
|
N/A | 87 | 80 | 92 |
|
N/A | 49 (22%) | 17 (18%) | 32 (25%) |
Surgical complications, |
185 (21.1) | 67 (27.8) | 31 (31.3) | 7 (29.2) |
Operative mortality, |
14 (1.6) | 5 (2.1) | 1 (1.0) | 1 (4.2) |
Adjuvant chemotherapy commenced, |
217 (24.8) | 33 (13.7) | 11 (11.1) | 4 (16.7) |
Adjuvant chemotherapy completed, |
593 (67.8) | 168 (69.7) | 63 (63.4) | 18 (75) |
aSerum creatinine.
bNumber and percentage of patients with estimated glomerular filtration rate less than 60 mL/min/1.73 m2.
Baseline renal function was available in the majority of the diabetes cohort (
During a median follow-up of 43.2 months, new primary colon cancers were found in 32 patients (2.9%), none of whom had diabetes (TNM stage 1;
Univariate and multivariate analyses for recurrence-free survival.
Clinicopathological variablea | Univariate model | Multivariate model | ||||
---|---|---|---|---|---|---|
HRb | 95% CIc |
|
HRb | 95% CIc |
| |
Age | ||||||
|
1.00 | 0.87–2.02 | 0.184 | 1.00 | 0.86–2.14 | 0.193 |
|
1.33 | 1.35 | ||||
Diabetes | ||||||
No diabetes | 1.00 | 0.74–1.60 | 0.667 | 1.00 | 0.65–1.49 | 0.93 |
Diabetes | 1.09 | 0.98 | ||||
T stage | ||||||
T3 | 1.00 | 1.65–3.41 | <0.001 | 1.00 | 1.36–2.94 | <0.001 |
T4 | 2.37 | 2.00 | ||||
Lymph node yield | ||||||
|
1.00 | 1.01–2.02 | 0.043 | 1.00 | 0.89–1.89 | 0.169 |
|
1.43 | 1.30 | ||||
Lymphovascular invasion | ||||||
Absent | 1.00 | 1.57–3.12 | <0.001 | 1.00 | 1.43–2.89 | <0.001 |
Present | 2.21 | 2.03 | ||||
Emergency presentation | ||||||
No | 1.00 | 1.21–2.75 | 0.004 | 1.00 | 1.07–2.49 | 0.022 |
Yes | 1.82 | 1.63 |
aMicrosatellite instability status was not included in the regression models for lack of complete data (complete data in 52% of patients).
bHazard ratio.
cConfidence interval.
(a) Recurrence free survival of diabetes and nondiabetes stage II colon cancer patients. (b) Recurrence free survival of diabetes patients comparing diet alone group (
For the diabetes population there was no difference in recurrence according to diabetes medication (patients receiving both metformin and insulin were excluded from the analysis of medication impact). As shown in Figure
(a) Overall survival of diabetes and nondiabetes stage II colon cancer patients. (b) Overall survival of diabetes patients comparing diet alone group (
A better understanding of the complex relationship between diabetes and colorectal cancer development and outcome requires comprehensive and reliable data. Here we report an analysis of a large population based series that does not suggest diabetes has a major impact on colorectal cancer pathology, completion of adjuvant chemotherapy, or risk of disease recurrence for patients with stage II colon cancer. There was also no significant impact of individual diabetes medication on disease recurrence or survival in this cohort.
If diabetes or diabetes medication was to impact on the risk of colorectal cancer recurrence, this could relate to the aggressiveness of the primary cancers that develop in these patients. However, in our series we could find no association between diabetes or diabetes treatment and patient presentation with clinicopathological features that are associated with an increased risk of recurrence. This is similar to findings from previous series [
In our series, adjuvant chemotherapy use in diabetes patients was lower than in non-diabetes patients (13.7% versus 24.7%,
We did not observe any impact of diabetes on colon cancer recurrence. There was also no observed impact of diabetes medication on risk of cancer recurrence. However, other well-known prognostic factors for cancer recurrence such as T4 stage, inadequate lymph node yield, presence of lymphovascular invasion, and emergency presentation were found to be of significance. There is a modest trend for better outcomes for metformin treated patients and worse outcomes for insulin treated patients, compared to patients on diet control alone. While there was a trend for diabetes patients to have an inferior overall survival (Log rank test;
There was no evidence for diabetes medication impacting overall survival. This finding is in contrast with three previous single-institution studies where metformin use was found to be associated with better overall survival [
In an analysis of a stage III colon cancer clinical trial cohort, an increased risk of cancer recurrence was found in diabetes patients [
When the impact of diabetes medication on overall survival was analysed in our series, a trend was observed in favour of metformin treated patients. Such analysis is confounded by patient factors associated with metformin administration that also are associated with survival. While we found no significant difference in renal function, we did observe that metformin treated patients were younger and more likely female. Reduced survival in insulin treated patients would not be unexpected due to insulin use being associated with obesity and poor diabetes control, both of which would predict increased noncancer deaths. Only one patient was noted to be on a newer diabetes medication, DPP-4 inhibitor, which has only recently become widely available in Australia.
There are several strengths to our data including the prospective nature of comprehensive data collection from the time of diagnosis through to recurrence and death, the standard approach to patient workup and treatment, and the relatively large sample size. Diabetes status was collected at diagnosis, along with diabetes medication, and detailed data on cancer recurrence and new primary cancers was available. We elected to examine stage II colon cancer for several reasons. We felt that examining one stage of disease rather than a combined analysis of all stages would allow a detailed analysis of cancer pathology, and also make interpretation of outcomes more reliable. Also, if diabetes proved to be an adverse prognostic factor then this could be added to the current list of high-risk features that guide adjuvant chemotherapy decision-making for stage II cases. Finally, given the low rate of usage and the marginal impact of adjuvant therapy on stage II colon cancer outcomes, there is minimal confounding due to any differences in anticancer treatment received, dosing, or treatment duration.
The weaknesses of our study include that the relatively low rate of recurrence in stage II colon cancer makes it more difficult to detect small but clinically significant differences between groups. Also despite a relatively large overall sample size the number of metformin and insulin treated patients was quite modest, particularly once patients on multiple medications were excluded.
Our study demonstrates that, for patients with stage II colon cancer, diabetes is not associated with clinicopathological factors that are known to bestow an increased risk of cancer recurrence. We found a difference in the rate of administration of adjuvant chemotherapy, which has implications for series examining outcomes in node positive cancers given adjuvant therapy has far greater impact on survival outcomes in these patients. We did not observe any significant associations between diabetes or diabetes medication and risk of cancer recurrence or death. Future prospective, randomised studies will further clarify the impact of diabetes medication on colon cancer outcomes. The results of such studies are eagerly awaited, both for colorectal cancer and for other cancers where metformin in particular shows promise as an anticancer agent.
The authors declare that there is no conflict of interests regarding the publication of this paper.