Patients with Crohn’s disease (CD) had a higher overall incidence of cancer than matched members of the general population in most treatment cohorts, including treatment-naive patients, but much of the excess was attributable to nonmelanoma skin cancer, according to a nationwide population-based cohort study published in The American Journal of Gastroenterology.
After nonmelanoma skin cancers were excluded, the study found that the absolute excess cancer incidence was small, amounting to approximately one additional cancer case per 1,000 person-years — roughly one extra case for every 1,000 patients followed for one year — in the largest cohorts. The remaining difference was attributed to lung, small bowel, hepatobiliary, and hematologic cancers.
Real-world data quantifying absolute cancer risk in CD by treatment status “are lacking but are essential for patient counselling,” the authors wrote.
The researchers used Swedish nationwide registry data from 2007 through 2023 to examine cancer incidence among 38,733 patients with CD and more than 360,000 matched general population comparators. Patients were followed for a median of 7.3 years. Each patient with CD was matched by birth year, sex, and place of residence with up to 10 people without inflammatory bowel disease. Of 50,855 patients with CD identified, 12,122 were excluded because of contraindications to immunomodulatory drugs, such as HIV, advanced liver or kidney disease, or transplantation, or because of prior use of those drugs.
Treatment cohorts comprised patients naive to immunomodulatory drugs and new users of thiopurines, tumor necrosis factor inhibitors (TNFi), combined thiopurine and TNFi therapy, vedolizumab, or ustekinumab. The researchers used a “once exposed, equal to always exposed” design, meaning patients could contribute follow-up time and events to more than one treatment group and were not censored when they discontinued a treatment.
The study measured excess cancer incidence — the number of additional cancers per 1,000 person-years in each CD cohort compared with its matched comparators. Cancers required histopathologic confirmation, and the researchers used a one-year lag between CD diagnosis or treatment start and cancer assessment to reduce detection bias.
Compared with matched comparators, excess incidence of any cancer ranged from 2.42 to 3.14 cases per 1,000 person-years in the treatment-naive, thiopurine, TNFi, and combination therapy cohorts. The excess in the vedolizumab and ustekinumab cohorts, the two smallest groups, was not statistically significant.
“Most of the observed excess cancer risk in CD in our study was due to elevated rates of basal and squamous cell skin carcinomas,” the authors wrote.
After nonmelanoma skin cancers were excluded, the excess was 1.27 cases per 1,000 person-years in the treatment-naive cohort, 1.14 in the thiopurine cohort, and 0.96 in the TNFi cohort. It was not statistically significant in the other three cohorts.
The study also found elevated lymphoma incidence compared with the general population in the thiopurine and TNFi cohorts, at 0.14 and 0.34 extra cases per 1,000 person-years, respectively. In contrast, the researchers found no elevated incidence for common cancers including colorectal, breast, and prostate cancer in any of the treatment cohorts.
For several other cancer types, including cervical, uterine, and nervous system cancers, the authors also reported no elevated cancer rates regardless of treatment group.
The researchers found that age had an important relationship with absolute cancer risk. Excess incidence was not significantly elevated among pediatric patients but was elevated among adults. In the thiopurine cohort, for example, the excess rose from 0.41 cases per 1,000 person-years in patients younger than 18, which was not statistically significant, to 6.65 in patients 60 and older.
Absolute excess incidence increased with age, but the relative increase in risk compared with the general population followed the opposite pattern: It was highest in younger patients and near general population levels in patients 60 and older, the authors reported.
Because cancer incidence increases steeply with age in the general population, “similar relative risks may translate into larger absolute differences in older patients,” the authors noted.
“Importantly, age modified both baseline cancer risk and the absolute impact of therapy and disease severity,” the authors wrote. They added that shared decision-making should consider patient age, comorbidities, and inflammatory burden.
“From a clinical perspective, these findings are reassuring. The absence of an increased overall cancer incidence in pediatric patients supports the continued use of thiopurines and biologics when clinically indicated, as the absolute cancer risk seems to be very low in this age group,” the authors wrote.
For most patients with CD, they wrote, the absolute excess cancer risk was small and should be weighed against the benefits of sustained control of inflammation.
The authors cautioned against using the findings to compare malignancy risk among individual treatments. Differences between the treatment cohorts could be affected by disease duration, disease phenotype, age, and other factors, and “this study does not provide any direct comparison across treatment groups.”
Other limitations included the median 7.3-year follow-up, which the authors wrote may be insufficient to capture malignancies with long latency periods. The once-exposed-always-exposed design also did not account for actual treatment duration and could underestimate risk because patients remained classified as exposed after discontinuing therapy.
As an observational study, the analysis also could not fully account for potential confounding from smoking, obesity, disease severity, cancer screening participation, genetic and environmental factors, and other comorbidities.
The authors also noted that overall cancer incidence is somewhat lower in Sweden than in the United States — age-standardized rates of about 310 vs. 367 new cases per 100,000 person-years in 2022 — and that differences in lifestyle, screening, and diagnostic practices may affect cancer incidence. Swedish registries also do not capture race, they wrote.
The study was funded by the Swedish Research Council, the Swedish Cancer Society, and other Swedish and European research grants. Several authors reported grants or advisory, consulting, or speaker fees from companies including AbbVie, Bristol Myers Squibb, Janssen, Pfizer, and Takeda.