Diabetes-Related Increased Cancer Risk May Be Statistical Artifact

VIENNA – Most of the increased risk of cancer associated with diabetes appears in the first year or so after diabetes diagnosis, suggesting that it’s mainly attributable to increased medical surveillance.

Three studies presented at the annual meeting of the European Association for the Study of Diabetes came to similar conclusions, which seemed to hold for both genders, for obesity-driven and non–obesity-driven cancers, and for patients with type 1 as well as type 2 diabetes – a new finding, according to Mr. Bendix Carstensen.

His large, population-based study found no overall excess risk of cancers among type 1 diabetes patients, compared with the general population.

“Based on this, we can exclude a major carcinogenic effect of exogenous insulin among those with type 1 diabetes, because if there was such an effect we would see some substantial increases in at least some cancers,” said Mr. Carstensen, an epidemiologist at the Steno Diabetes Center in Gentofte, Denmark.

Cancer and type 1 diabetes

The study examined cancer rates in patients with type 1 diabetes from five countries: Australia, Denmark, Finland, Scotland, and Sweden. Type 1 diabetes was defined as a diabetes diagnosis that occurred before age 30. In general, these registries contained patients who were still younger than 60 years. Despite the very large 4.6 million person-years of follow-up, the databases represent a population that does not exhibit the typical age-related increase in cancer risk – a slight limitation of the study, Mr. Carstensen noted.

The databases identified 9,369 cases of cancer among patients with type 1 diabetes. They were classified by gender, age, date of cancer diagnosis, and cancer rate, compared with that of the country’s entire population stratified by the same variables.

The crude rate of cancers in all the diabetes patients combined was no different from that of the general populations, with a risk ratio of 1.00 for men and 1.04 for women.

When cancers were examined by site and gender, some significant differences did arise between the diabetic and nondiabetic groups. Stomach cancer was 19% more likely in men and 75% more likely in women. The risk of pancreatic cancer was 70% increased in men and 36% in women. The risk of liver cancer was about doubled in each gender, and for kidney cancer, the risk was 29% in men and 42% in women. For women, there was a 53% increase in the risk of endometrial cancer.

In the time-dependent analysis, Mr. Carstensen found that almost all of the cancers diagnoses were made in the first year after diabetes was diagnosed and dropped rapidly thereafter. The extended time curve showed no lasting impact of diabetes on overall cancer incidence.

A more specific analysis looked at the time-dependent rate ratios of prostate and colorectal cancers for men, and breast, endometrial, and colorectal cancers in women. Each curve showed the high rate ratio in the first few years, followed by a drop-off and no lasting impact.

There were also no lasting impacts of diabetes on lung cancer, melanoma, or non-Hodgkins lymphoma in either gender.

“We did see an elevated site-specific cancer pattern in patients with type 1 diabetes, but no overall excess,” he said. “For these patients, the total cancer occurrence is not really different from population rates.”

Detection bias in diabetes and obesity

Another large, population-based study came to a similar conclusion for those with type 2 diabetes. In fact, “Detection bias may be the main cause of the increased cancer incidence among patients with diabetes,” Dr. Kirstin De Bruijn said.

She presented a subanalysis of the ongoing Rotterdam Study. The Rotterdam Study, launched in 1990, is investigating determinants of disease in residents aged 55 years and older. It now comprises about 11,000 subjects. Dr. De Bruijn of Erasmus University, Rotterdam, the Netherlands, investigated the association of cancer and diabetes in 10,181 patients. Of these, 906 had an incident type 2 diabetes diagnosis, and 2,238 had an incident cancer diagnosis during the mean follow-up period of 11 years. She looked at the incidence of breast, prostate, pancreatic, lung, and colorectal cancers.

In the overall analysis, the risk of any cancer was 30% increased in the patients with diabetes. This risk was attenuated, but remained statistically significant, in the fully adjusted model (hazard ratio, 1.2). In a cancer-specific analysis, however, only the risk of pancreatic cancer was significantly elevated (HR, 3.6 in the adjusted model). The risk of prostate cancer was 30% lower than that of the general population, but that was not a significant finding, she added.

She then looked at a time-dependent model that split the follow-up period into epochs according to time since diabetes diagnosis (up to 3 months, 3 months to 5 years, and more than 5 years).

The overall risk of a cancer diagnosis was more than three times higher in the first 3 months. It remained significantly elevated, though less so, in the first 5 years (HR, 1.5), and then fell off.

When the individual cancers were considered, only pancreatic cancer was significantly more common among the diabetes patients, and was almost 29 times more likely to be diagnosed in the first 3 months. However, Dr. De Bruijn noted, it’s tough to tease out that particular relationship, since the obesity that accompanies type 2 diabetes can also drive the development of pancreatic cancer.

Ellena Badrick, a researcher at the University of Manchester (England), also examined the idea of detection bias in a population database, exploring the time-dependent relationships for obesity-related cancers, compared with those not related to obesity.

She found 10,315 patients who were diagnosed with type 2 diabetes from 1995 to 2010. These were compared with 20,630 controls chosen from the same period. Obesity-related cancers were considered to be breast, endometrial, ovarian, renal, esophageal, pancreatic, and gallbladder.

There were 1,349 cancers among the patients with diabetes, of which 323 were related to obesity. Among the controls, there were 3,218 cancers, of which 634 were obesity related.

She also split her follow-up period into epochs since diabetes diagnosis: up to 6 months, 6-12 months, 12-24 months, 24 months-5 years, and beyond. Cancer incidence was reported as cases per 1,000 person/epoch.

There was a much higher detection rate overall in the first 6 months after diagnosis – more than 100 cases per 1,000 person-epoch. After 6 months, this dropped to less than 20 cases per 1,000 person-epoch. The incidence did increase over the follow-up period, but she said that reflected the expected age-related pattern.

The same pattern emerged when looking at obesity- vs. non–obesity-related cancers. In the first 6 months, the incidence of obesity-related cancer hovered around 30 per 1,000 person-epoch. By 1 year this had dropped to near zero. For non–obesity-related cancers, the incidence rate was much higher in the first 6 months, at nearly 80 per person/epoch. But this also dropped to near zero by 1 year. Both incidence curves followed the same slow increase as the subjects aged.

Over the entire follow-up, 27% of all the obesity-related cancers and 73% of the non–obesity-related cancers were diagnosed within the first year after a diagnosis of type 2 diabetes.

When cancers diagnosed during the first 2 years were excluded, patients with type 2 diabetes had a 43% increase in the risk of developing an obesity-related cancer during the study. There was no increased risk, however, in developing a cancer not related to obesity.

“This suggests that in patients with type 2 diabetes, cancer risk-reduction strategies should be targeted against obesity-related cancers,” Ms. Badrick said.

Mr. Carstensen is an employee of the Steno Diabetes Center, which is owned by Novo Nordisk. He disclosed that he owns stock in the company. Dr. De Bruijn and Ms. Badrick had no financial disclosures.

VIENNA – Most of the increased risk of cancer associated with diabetes appears in the first year or so after diabetes diagnosis, suggesting that it’s mainly attributable to increased medical surveillance.

Three studies presented at the annual meeting of the European Association for the Study of Diabetes came to similar conclusions, which seemed to hold for both genders, for obesity-driven and non–obesity-driven cancers, and for patients with type 1 as well as type 2 diabetes – a new finding, according to Mr. Bendix Carstensen.

His large, population-based study found no overall excess risk of cancers among type 1 diabetes patients, compared with the general population.

“Based on this, we can exclude a major carcinogenic effect of exogenous insulin among those with type 1 diabetes, because if there was such an effect we would see some substantial increases in at least some cancers,” said Mr. Carstensen, an epidemiologist at the Steno Diabetes Center in Gentofte, Denmark.

Cancer and type 1 diabetes

The study examined cancer rates in patients with type 1 diabetes from five countries: Australia, Denmark, Finland, Scotland, and Sweden. Type 1 diabetes was defined as a diabetes diagnosis that occurred before age 30. In general, these registries contained patients who were still younger than 60 years. Despite the very large 4.6 million person-years of follow-up, the databases represent a population that does not exhibit the typical age-related increase in cancer risk – a slight limitation of the study, Mr. Carstensen noted.

The databases identified 9,369 cases of cancer among patients with type 1 diabetes. They were classified by gender, age, date of cancer diagnosis, and cancer rate, compared with that of the country’s entire population stratified by the same variables.

The crude rate of cancers in all the diabetes patients combined was no different from that of the general populations, with a risk ratio of 1.00 for men and 1.04 for women.

When cancers were examined by site and gender, some significant differences did arise between the diabetic and nondiabetic groups. Stomach cancer was 19% more likely in men and 75% more likely in women. The risk of pancreatic cancer was 70% increased in men and 36% in women. The risk of liver cancer was about doubled in each gender, and for kidney cancer, the risk was 29% in men and 42% in women. For women, there was a 53% increase in the risk of endometrial cancer.

In the time-dependent analysis, Mr. Carstensen found that almost all of the cancers diagnoses were made in the first year after diabetes was diagnosed and dropped rapidly thereafter. The extended time curve showed no lasting impact of diabetes on overall cancer incidence.

A more specific analysis looked at the time-dependent rate ratios of prostate and colorectal cancers for men, and breast, endometrial, and colorectal cancers in women. Each curve showed the high rate ratio in the first few years, followed by a drop-off and no lasting impact.

There were also no lasting impacts of diabetes on lung cancer, melanoma, or non-Hodgkins lymphoma in either gender.

“We did see an elevated site-specific cancer pattern in patients with type 1 diabetes, but no overall excess,” he said. “For these patients, the total cancer occurrence is not really different from population rates.”

Detection bias in diabetes and obesity

Another large, population-based study came to a similar conclusion for those with type 2 diabetes. In fact, “Detection bias may be the main cause of the increased cancer incidence among patients with diabetes,” Dr. Kirstin De Bruijn said.

She presented a subanalysis of the ongoing Rotterdam Study. The Rotterdam Study, launched in 1990, is investigating determinants of disease in residents aged 55 years and older. It now comprises about 11,000 subjects. Dr. De Bruijn of Erasmus University, Rotterdam, the Netherlands, investigated the association of cancer and diabetes in 10,181 patients. Of these, 906 had an incident type 2 diabetes diagnosis, and 2,238 had an incident cancer diagnosis during the mean follow-up period of 11 years. She looked at the incidence of breast, prostate, pancreatic, lung, and colorectal cancers.

In the overall analysis, the risk of any cancer was 30% increased in the patients with diabetes. This risk was attenuated, but remained statistically significant, in the fully adjusted model (hazard ratio, 1.2). In a cancer-specific analysis, however, only the risk of pancreatic cancer was significantly elevated (HR, 3.6 in the adjusted model). The risk of prostate cancer was 30% lower than that of the general population, but that was not a significant finding, she added.

She then looked at a time-dependent model that split the follow-up period into epochs according to time since diabetes diagnosis (up to 3 months, 3 months to 5 years, and more than 5 years).

The overall risk of a cancer diagnosis was more than three times higher in the first 3 months. It remained significantly elevated, though less so, in the first 5 years (HR, 1.5), and then fell off.

When the individual cancers were considered, only pancreatic cancer was significantly more common among the diabetes patients, and was almost 29 times more likely to be diagnosed in the first 3 months. However, Dr. De Bruijn noted, it’s tough to tease out that particular relationship, since the obesity that accompanies type 2 diabetes can also drive the development of pancreatic cancer.

Ellena Badrick, a researcher at the University of Manchester (England), also examined the idea of detection bias in a population database, exploring the time-dependent relationships for obesity-related cancers, compared with those not related to obesity.

She found 10,315 patients who were diagnosed with type 2 diabetes from 1995 to 2010. These were compared with 20,630 controls chosen from the same period. Obesity-related cancers were considered to be breast, endometrial, ovarian, renal, esophageal, pancreatic, and gallbladder.

There were 1,349 cancers among the patients with diabetes, of which 323 were related to obesity. Among the controls, there were 3,218 cancers, of which 634 were obesity related.

She also split her follow-up period into epochs since diabetes diagnosis: up to 6 months, 6-12 months, 12-24 months, 24 months-5 years, and beyond. Cancer incidence was reported as cases per 1,000 person/epoch.

There was a much higher detection rate overall in the first 6 months after diagnosis – more than 100 cases per 1,000 person-epoch. After 6 months, this dropped to less than 20 cases per 1,000 person-epoch. The incidence did increase over the follow-up period, but she said that reflected the expected age-related pattern.

The same pattern emerged when looking at obesity- vs. non–obesity-related cancers. In the first 6 months, the incidence of obesity-related cancer hovered around 30 per 1,000 person-epoch. By 1 year this had dropped to near zero. For non–obesity-related cancers, the incidence rate was much higher in the first 6 months, at nearly 80 per person/epoch. But this also dropped to near zero by 1 year. Both incidence curves followed the same slow increase as the subjects aged.

Over the entire follow-up, 27% of all the obesity-related cancers and 73% of the non–obesity-related cancers were diagnosed within the first year after a diagnosis of type 2 diabetes.

When cancers diagnosed during the first 2 years were excluded, patients with type 2 diabetes had a 43% increase in the risk of developing an obesity-related cancer during the study. There was no increased risk, however, in developing a cancer not related to obesity.

“This suggests that in patients with type 2 diabetes, cancer risk-reduction strategies should be targeted against obesity-related cancers,” Ms. Badrick said.

Mr. Carstensen is an employee of the Steno Diabetes Center, which is owned by Novo Nordisk. He disclosed that he owns stock in the company. Dr. De Bruijn and Ms. Badrick had no financial disclosures.

VIENNA – Most of the increased risk of cancer associated with diabetes appears in the first year or so after diabetes diagnosis, suggesting that it’s mainly attributable to increased medical surveillance.

Three studies presented at the annual meeting of the European Association for the Study of Diabetes came to similar conclusions, which seemed to hold for both genders, for obesity-driven and non–obesity-driven cancers, and for patients with type 1 as well as type 2 diabetes – a new finding, according to Mr. Bendix Carstensen.

His large, population-based study found no overall excess risk of cancers among type 1 diabetes patients, compared with the general population.

“Based on this, we can exclude a major carcinogenic effect of exogenous insulin among those with type 1 diabetes, because if there was such an effect we would see some substantial increases in at least some cancers,” said Mr. Carstensen, an epidemiologist at the Steno Diabetes Center in Gentofte, Denmark.

Cancer and type 1 diabetes

The study examined cancer rates in patients with type 1 diabetes from five countries: Australia, Denmark, Finland, Scotland, and Sweden. Type 1 diabetes was defined as a diabetes diagnosis that occurred before age 30. In general, these registries contained patients who were still younger than 60 years. Despite the very large 4.6 million person-years of follow-up, the databases represent a population that does not exhibit the typical age-related increase in cancer risk – a slight limitation of the study, Mr. Carstensen noted.

The databases identified 9,369 cases of cancer among patients with type 1 diabetes. They were classified by gender, age, date of cancer diagnosis, and cancer rate, compared with that of the country’s entire population stratified by the same variables.

The crude rate of cancers in all the diabetes patients combined was no different from that of the general populations, with a risk ratio of 1.00 for men and 1.04 for women.

When cancers were examined by site and gender, some significant differences did arise between the diabetic and nondiabetic groups. Stomach cancer was 19% more likely in men and 75% more likely in women. The risk of pancreatic cancer was 70% increased in men and 36% in women. The risk of liver cancer was about doubled in each gender, and for kidney cancer, the risk was 29% in men and 42% in women. For women, there was a 53% increase in the risk of endometrial cancer.

In the time-dependent analysis, Mr. Carstensen found that almost all of the cancers diagnoses were made in the first year after diabetes was diagnosed and dropped rapidly thereafter. The extended time curve showed no lasting impact of diabetes on overall cancer incidence.

A more specific analysis looked at the time-dependent rate ratios of prostate and colorectal cancers for men, and breast, endometrial, and colorectal cancers in women. Each curve showed the high rate ratio in the first few years, followed by a drop-off and no lasting impact.

There were also no lasting impacts of diabetes on lung cancer, melanoma, or non-Hodgkins lymphoma in either gender.

“We did see an elevated site-specific cancer pattern in patients with type 1 diabetes, but no overall excess,” he said. “For these patients, the total cancer occurrence is not really different from population rates.”

Detection bias in diabetes and obesity

Another large, population-based study came to a similar conclusion for those with type 2 diabetes. In fact, “Detection bias may be the main cause of the increased cancer incidence among patients with diabetes,” Dr. Kirstin De Bruijn said.

She presented a subanalysis of the ongoing Rotterdam Study. The Rotterdam Study, launched in 1990, is investigating determinants of disease in residents aged 55 years and older. It now comprises about 11,000 subjects. Dr. De Bruijn of Erasmus University, Rotterdam, the Netherlands, investigated the association of cancer and diabetes in 10,181 patients. Of these, 906 had an incident type 2 diabetes diagnosis, and 2,238 had an incident cancer diagnosis during the mean follow-up period of 11 years. She looked at the incidence of breast, prostate, pancreatic, lung, and colorectal cancers.

In the overall analysis, the risk of any cancer was 30% increased in the patients with diabetes. This risk was attenuated, but remained statistically significant, in the fully adjusted model (hazard ratio, 1.2). In a cancer-specific analysis, however, only the risk of pancreatic cancer was significantly elevated (HR, 3.6 in the adjusted model). The risk of prostate cancer was 30% lower than that of the general population, but that was not a significant finding, she added.

She then looked at a time-dependent model that split the follow-up period into epochs according to time since diabetes diagnosis (up to 3 months, 3 months to 5 years, and more than 5 years).

The overall risk of a cancer diagnosis was more than three times higher in the first 3 months. It remained significantly elevated, though less so, in the first 5 years (HR, 1.5), and then fell off.

When the individual cancers were considered, only pancreatic cancer was significantly more common among the diabetes patients, and was almost 29 times more likely to be diagnosed in the first 3 months. However, Dr. De Bruijn noted, it’s tough to tease out that particular relationship, since the obesity that accompanies type 2 diabetes can also drive the development of pancreatic cancer.

Ellena Badrick, a researcher at the University of Manchester (England), also examined the idea of detection bias in a population database, exploring the time-dependent relationships for obesity-related cancers, compared with those not related to obesity.

She found 10,315 patients who were diagnosed with type 2 diabetes from 1995 to 2010. These were compared with 20,630 controls chosen from the same period. Obesity-related cancers were considered to be breast, endometrial, ovarian, renal, esophageal, pancreatic, and gallbladder.

There were 1,349 cancers among the patients with diabetes, of which 323 were related to obesity. Among the controls, there were 3,218 cancers, of which 634 were obesity related.

She also split her follow-up period into epochs since diabetes diagnosis: up to 6 months, 6-12 months, 12-24 months, 24 months-5 years, and beyond. Cancer incidence was reported as cases per 1,000 person/epoch.

There was a much higher detection rate overall in the first 6 months after diagnosis – more than 100 cases per 1,000 person-epoch. After 6 months, this dropped to less than 20 cases per 1,000 person-epoch. The incidence did increase over the follow-up period, but she said that reflected the expected age-related pattern.

The same pattern emerged when looking at obesity- vs. non–obesity-related cancers. In the first 6 months, the incidence of obesity-related cancer hovered around 30 per 1,000 person-epoch. By 1 year this had dropped to near zero. For non–obesity-related cancers, the incidence rate was much higher in the first 6 months, at nearly 80 per person/epoch. But this also dropped to near zero by 1 year. Both incidence curves followed the same slow increase as the subjects aged.

Over the entire follow-up, 27% of all the obesity-related cancers and 73% of the non–obesity-related cancers were diagnosed within the first year after a diagnosis of type 2 diabetes.

When cancers diagnosed during the first 2 years were excluded, patients with type 2 diabetes had a 43% increase in the risk of developing an obesity-related cancer during the study. There was no increased risk, however, in developing a cancer not related to obesity.

“This suggests that in patients with type 2 diabetes, cancer risk-reduction strategies should be targeted against obesity-related cancers,” Ms. Badrick said.

Mr. Carstensen is an employee of the Steno Diabetes Center, which is owned by Novo Nordisk. He disclosed that he owns stock in the company. Dr. De Bruijn and Ms. Badrick had no financial disclosures.