(View plain language summary)
Excess body weight is associated with increased risk of colorectal cancer (CRC). However, it is currently not clear how weight loss strategies alter future CRC risk. Obesity (or bariatric) surgery (OS) is increasingly performed in individuals who are ‘morbidly’ obese (body mass index [BMI] >40 Kg/m2). A human mucosal biomarker study (Gut 2011;60;893-901) and a Swedish population-based epidemiological (Ann Surg 2013) study by the Applicants both demonstrated that CRC risk actually increases with time after OS (in a cohort of 15095 individuals). Increased CRC risk was particularly evident greater than ten years after OS (consistent with the natural history of CRC) and was similar for gastric bypass, gastric banding and gastroplasty. By contrast, CRC risk in non-operated obese individuals remained stable.
To confirm and extend the important (but counter-intuitive, in the presence of weight loss) finding from the Swedish population-based study that CRC risk actually increases after OS, by performing a larger British population study based on individual-level linkage of the Hospital Episode Statistics (HES) database, National Cancer Data Registry (NCDR), Office for National Statistics (ONS) data and the General Practice Research Datalink (CPRD).
Ethically approved database linkage will be performed using an established National Cancer Intelligence Network (NCIN) linkage approach. HES data (ICD codes for OS [G30-33] or obesity alone [E66; ICD-10]) from 1997-2010 will be linked to NCDR data on CRC diagnosis, tumour location and stage at diagnosis until 2014. Separate linkage of NCDR data to the CPRD (covering approximately 5% of the population) will validate HES coding for the OS procedure or obesity alone for a subset of the study population. Where possible, we will use longitudinal BMI data from the CPRD to link changes in body weight over time with CRC risk. The CPRD will also allow us to investigate colorectal adenoma (polyp) risk after OS. The National Bariatric Surgery Register (NBSR) cannot be used as it does not have approval for use of individual data.
The risk of developing CRC will be compared between those who underwent OS and non-operated obese individuals. Incidence rates will be compared to the expected risk of developing CRC in the English population as a whole. The expected number of CRCs will be calculated for each cohort by multiplying the observed person-time by age, sex and calendar year-specific cancer incidence rates in the corresponding background population. Standardised incidence ratios will then be calculated. Person-years at risk in each cohort will be calculated from the date of entry into the study to the diagnosis of a cancer, death (using ONS data) or the end of the study period (2014).
HES data predict that, during the study period from 1997 until 2010, the OS cohort with greater than 10 year post-BS exposure will be approximately 6700 with an estimated 70 incident CRCs based on an age-standardised CRC rate of 50/100000/year and an age range of 40-60 years during follow-up (cf. 49 CRCs in the Swedish population study).
A total of 1,056,392 patients were initially identified. The final dataset consisted of 1,002,607 individuals, including; 39,747 (4%) as having OS and the remainder 962,860 (96%) as having an episode of hospital care due to obesity without OS. The majority of patients in both groups were female; 77% in the OS group and 63% in the obese no OS group. As expected, the OS group was younger than the obese no surgery group, with a mean age of 45 and 53 years, respectively. The majority of OS (91%) took place after 2006 and this restricted the potential follow-up time after surgery to six years for the majority of this population. The OS group had a median follow-up period of three years (range 1-16 years) and 144,677 person-years of follow-up in which 43 new diagnoses of CRC were made. The equivalent figures for the obese no surgery group were a median follow-up time of 2.5 years (range 1-16 years), and 3,608,882 person-years at risk and 3,237 new diagnoses of CRC.
The absolute cumulative incidence of CRC in the OS group was lower (30 per 100,000 person-years) than that in the obese no surgery group (91 per 100,000 person-years), which is likely explained by the younger age of the OS cohort. There was a trend towards an increased risk of CRC following OS group (SIR: 1.26, 95% CI: 0.92-1.71), but the association was not statistically significant. There was, however, statistically significant slightly increased CRC risk in individuals who did not undergo surgery (SIR: 1.12, 95% CI: 1.08-1.16). In the OS group, a significantly increased risk of CRC was observed in the oldest (≥50 years) age group (SIR: 1.47, 95% CI: 1.02-2.06). In the obese who did not undergo OS, the SIR of CRC was higher in males (SIR: 1.21, 95% CI: 1.15-1.26) than in females (SIR: 1.02, 95% CI: 0.97-1.08) and it decreased with age. The risk of CRC was significantly higher for the obese no surgery group in the latest calendar period (2006–13) (SIR: 1.16, 95% CI: 1.12-1.21) compared to the earliest period (SIR: 0.94, 95% CI: 0.86-1.02). Finally, the SIR increased with follow-up time in the OS group (SIR: 1.24, 95% CI: 0.81-1.81 for 1-2 years vs SIR: 1.32, 95% CI: 0.77-2.11 for ≥2 years), and decreased in the obese no surgery group (SIR: 1.14, 95% CI: 1.09-1.20 for 1-2 years vs SIR: 1.10, 95% CI: 1.04-1.15 for ≥2 years), the latter attaining statistical significance.
There was an overall decreased risk of breast cancer after OS (SIR: 0.76, 95% CI: 0.62-0.92), and slightly increased risk in obese individuals who did not undergo OS (SIR: 1.08, 95% CI: 1.04-1.11). The risk of endometrial cancer was increased by nearly three-fold for both groups (SIR: 2.98, 95% CI: 2.25-3.90, for OS and SIR: 2.60, 95% CI: 2.48-2.73, for obese no surgery groups) compared with the background population. The risk of kidney cancer was increased approximately 3-fold after OS (SIR: 3.06, 95% CI: 2.08-4.34) and almost two-fold in obese individuals who did not undergo OS (SIR: 1.78, 95% CI: 1.68-1.89). However, the risk of lung cancer was reduced in the OS group (SIR: 0.70, 95% CI: 0.46-1.03) while it was slightly raised for obese patients who did not undergo OS (SIR: 1.09, 95% CI: 1.05-1.13).
In summary, although we confirmed the established relationship between obesity and CRC, we have been unable to confirm or refute the findings of the earlier Swedish study that suggested an increase in CRC risk after OS due to the smaller than expected number of CRCs that occurred after OS. The obese patient cohort identified through HES was much larger than that in the Swedish study but OS was performed less frequently than in Sweden and limited post-operative follow-up for the majority of OS cases in England meant that the actual number of post-OS CRC cases was less than the Swedish study. Our stringent methodology, which included exclusion of surgical HES codes that would be unlikely to be OS and exclusion of cancer cases within 12 months of surgery, also contributed to the modest number of post-OS CRC cases.
The non-CRC data are perhaps the most interesting and suggest that breast cancer risk is reduced after OS. The SIRs for endometrial (uterine) and kidney cancer were high in both OS and no surgery obese cohorts highlighting the strong link with obesity for these particular cancers.
Excess body weight (termed overweight or obese depending on its degree) is associated with increased risk of some cancers, notably bowel cancer and post-menopausal breast cancer. Therefore, one might expect that weight loss would lead to decreased future cancer risk. Obesity surgery (OS) is a recognised treatment for obesity when other measures have failed. Obesity surgery induces significant, long-lasting weight loss. However, counter-intuitively, a study performed in Sweden suggested that risk of bowel cancer (but not other cancers) actually increased after OS, perhaps related to changes in diet and gut microbes after surgery.
We aimed to determine the risk of bowel (and other) cancer risk after OS in patients in the English NHS in comparison with obese patients who did not undergo OS.
All individuals who underwent OS or had a hospital episode (admitted to hospital or a day-case procedure) with a diagnosis of obesity, but no OS, 1997–2013 were identified using the Hospital Episode Statistics (HES) database, which contains details of all admissions and outpatient appointments in English NHS hospitals. Subsequent diagnosis of bowel cancer and the time after OS ‘at risk’ were determined by linking individuals to National Cancer Registration and Analysis Service records and Office of National Statistics (ONS) data on date of death, respectively. Approval for use of patient identifiable data without individual consent was obtained from the Confidentiality Advisory Group. Because OS patients and obese patients were so different, we could not compare the cancer rates directly. Instead, we calculated the standardised incidence ratio (SIR), which compares the cancer rate to that of the normal population with the same age and gender.
The final dataset contained just over 1 million (1,002,607) obese patients, of whom 4% (n=39,747) underwent OS. In the obese population that did not undergo OS, 3,237 developed bowel cancer, which gave a SIR 1.12 – equivalent to a 12% increased bowel cancer risk. In those who underwent OS, 43 developed CRC, which gave a SIR of 1.26. By contrast, OS was associated with decreased breast cancer risk (SIR 0.76 – a 24% risk reduction), unlike obese individuals who did not undergo OS. The risk of kidney and uterus cancer (but not lung cancer) was significantly higher in obese individuals than the normal population.
We confirmed that bowel cancer risk is increased in individuals with at least one hospital episode coded as obesity. Unfortunately the number of individuals who developed bowel cancer after OS was limited which meant that the accuracy of our risk estimates is limited. Risk of breast cancer after OS was reduced compared with obese individuals who do not have OS. This suggests that OS affects cancer risk differently at different body sites. In our study, there was a strong relationship between obesity and risk of other obesity-related cancers such as kidney cancer.
Changes in cancer risk after OS should be further investigated so that appropriate monitoring can be introduced in the ever-increasing number of people undergoing OS.