November 13, 2017
Heather Yeo, MD, MHS
In 2005 when the anti-tumor necrosis factor (anti-TNF) drug infliximab was approved by the Food and Drug Administration for treatment of ulcerative colitis (UC), it was considered a new hope for patients with UC who had failed conventional therapy, with short-term, randomized control trials showing success in achieving clinical remission. After ten years of experience with biologics, anywhere from 30% to 66% of patients still require surgical intervention (Aratari, et al; Gustavson, et al; Kohn, et al; Reich, et al). Researchers in the Department of Surgery at Weill Cornell Medicine sought to further examine the controversy and sort our if infliximab is helping as much originally hoped, and if things are getting better or worse.
“Clinically, many surgeons have noticed that when patients finally come for surgical intervention, they are sicker,” said Dr. Heather Yeo, the Nanette Laitman Clinical Scholar in Healthcare Policy and Research/Clinical Evaluation at Weill Cornell Medicine and an author on the study. “We wanted to determine whether this was true at a national level.”
The comprehensive, longitudinal study sought to compare colorectal surgical outcomes in UC patients before the approval of infliximab against those who received colorectal surgery after a course of infliximab treatment. Patients with UC undergoing colorectal surgery were identified using relevant ICD-9 diagnosis and procedure codes. Patients were only included if they underwent any of the following as their first colorectal surgery:
- Total abdominal colectomy
- Total proctocolectomy
- Ileostomy
- Pouch construction
Patients with prior diagnosis of colorectal cancer or had a total abdominal colectomy for a diagnosis code other than UC were also excluded. Demographic information plus other identifiers were catalogued, including concurrent diagnosis of C. diff, comorbidities, presence of weight loss, and hospital volume (broken into low (0-5 procedures annually), medium (6-21) and high volume (22 or more) tertiles).
The primary outcome studied was in-hospital death. Secondary events studied include in-hospital major events (stroke, pulmonary embolism, shock, etc.) prolonged length of stay, procedural complications such as bleeding, 30-day readmissions, other-than-routine discharges, and number of subsequent procedures within 1 year of the index procedure. After comparing baseline characteristics and adjusting for differences, the research group compared procedure utilization including types of procedures performed. Outcomes were then adjusted using a generalized linear mixed model, which included adjustments for hospital volume, hospital clusters, age, sex, insurance status, comorbidities, admission status, procedure group, concurrent C. diff diagnosis, minimally invasive procedures, and initial creation of the pouch.
Key observations in the results included the following:
- Patients underwent more procedures after 2005. There were on average 346 procedures per year from 1995 to 205 and 408 procedures per year from 2006 to 2013. More patients underwent 3 or more procedures from 2006-2013 than in the prior period (14% vs. 9%, p < 0.01).
- Patients were sicker after 2005. A higher proportion of patients in the post-2005 group had 2 or more comorbidities, specifically (p = < 0.01 for all below):
- Hypertension (18% before vs. 23%)
- Obesity (2% vs. 5%)
- Cardiopulmonary disease (9% vs. 12%)
- Anemia (18% vs. 21%)
- Renal failure (0% vs. 5%), and
- Concurrent C. diff infection (6% vs. 11%)
A higher rate of diabetes was also found after 2005 (9% vs. 11%, p = 0.02)
- Patients had a higher adjusted risk of readmission and complications after 2005. Rates of 30-day readmission, major events, and procedural complications after 2005 all featured significant increases no matter whether the procedure was elective or emergent.
The research team acknowledges limitations with the study, including a lack of individualized prescription information to determine which patients in the later cohort were exposed to biologics, and the potential for clerical errors in the administrative database used to source data. They still believe scale of the data studied and the robust adjustments for numerous variables give the data credibility to draw comparisons between the two time periods.
“Our hypothesis that staged procedures increased after 2005 was proven correct by this longitudinal study,” said Dr. Yeo. “Even with robust adjustments for comorbidities, insurance type, and different procedure histories, since the introduction of infliximab, patients also have worse outcomes.”
Dr. Yeo’s team thinks that biologics may prevent those with mild or moderate disease from ever progressing to the point of surgery, leaving only the patients with the most advanced disease to advance to surgery. Patients may also be referred too late for a single stage procedure to be effective, leading to a greater number of staged procedures, pushing the procedure count up. Similarly, surgeons may also be acting more cautiously, doing procedures in multiple stages that they might’ve opted to do single-stage in the past. C. diff infection did rise after 2005, but it was accounted for in adjusted analyses so it’s not likely to be the primary driver. Lastly, the immunosuppressive properties of anti-TNFs like infliximab may make patients more susceptible to poor outcomes.
Dr. Yeo sees the study of these potential factors as a logical next step for research:
“I think the next question answer is who is more likely to respond to biologics like infliximab and who would not benefit. If a predictive model can be determined, it could lead to personalization of treatment pathways, giving those unlikely to benefit from biologics the chance to proceed to surgery at an earlier stage of disease when chance for successful clinical remission are higher.”