SEARCH

SEARCH BY CITATION

Keywords:

  • anti-tumor necrosis factor;
  • small bowel resection;
  • Crohn's disease

Abstract

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Background:

Our primary aim was to determine if the rate of small bowel resection (SBR) has declined over time among Crohn's disease (CD) patients seen at a single academic institution. A secondary aim was to establish whether the indication for surgery has changed.

Methods:

Patients with a primary or secondary ICD-9 code for CD (555.0–555.9) who underwent SBR at the University of Pittsburgh were included. Patients were divided into 4 separate time periods based on when they had surgery: 1995–1998 (Period 1), 1999–2001 (Period 2), 2002–2004 (Period 3), and 2005–2007 (Period 4). Medical records were reviewed for the 6 months preceding surgery. Use of 5-ASAs, immunomodulators (IMs), tumor necrosis factor (TNF) antagonists, and corticosteroids were noted. Disease behavior was defined as nonstricturing, nonpenetrating (B1), stricturing (B2), and penetrating (B3). Proportions of patients undergoing SBR were calculated according to calendar cohort and these rates were examined for time trends.

Results:

In all, 227 unique patients were analyzed for a total of 236 surgeries. The rates of 5-ASA, IM, and corticosteroid use were similar across the 4 time periods. By contrast, TNF antagonist usage progressively increased over time (0%, 18%, 34%, 35%; P = 0.0002). The annual rate of SBR per period did not change (1.6%, 1.9%, 1.6%, 1.9%; P = 0.93). Similarly, the disease behavior did not change over time.

Conclusions:

While the frequency of TNF antagonist use in CD at the University of Pittsburgh has increased over time, the rate of SBR and indication for surgery has remained unchanged. These findings may be explained by long-standing, complicated disease refractory to medical therapy. Inflamm Bowel Dis 2009

Up to 75% of patients with Crohn's disease (CD) will require intestinal resection at some point in their disease course.1 Of those undergoing resectional surgery, the need for reoperation is as high as 70% after 20 years of disease.2 The past decade has seen an increase in the use of immunomodulators (thiopurines and methotrexate) and antitumor necrosis factor (TNF) such as infliximab, adalimumab, and certolizumab pegol. Despite these advances, many patients continue to require surgery due to stricturing and penetrating complications.

It is unclear whether or not immunomodulators (IMs) have altered the rates of resectional surgery or the indication for surgery (behavioral phenotype) in CD. Cosnes et al3 noted that annual rates of intestinal resection at a French tertiary care center between 1978 and 2002 did not appear to change with the secular increase in azathioprine use. Similarly, the indication for surgery did not change over time, although the most recent cohort experienced a reduction in the proportion of operations performed for failure of medical therapy (nonstricturing, nonpenetrating disease). However, this study may have been limited in part by the paucity of patients who received at least 3 months of azathioprine prior to surgery. Additionally, the effect of anti-TNF agents could not be determined as their documented use was negligible. The primary findings of Cosnes et al are supported by those of Bewtra et al,4 who analyzed the National Hospital Discharge Survey between 1990 and 2003 and found that bowel surgery rates among CD patients did not change over time. By contrast, a more recent pediatric study concluded that early azathioprine use is associated with lower rates of surgery.5

To date, published studies examining the effects of anti-TNF agents on small bowel resection rates have largely been limited to 1-year follow-up results from randomized trials.6, 7 Patients randomized to receive infliximab maintenance in ACCENT I required fewer surgeries compared to those who were randomized to receive episodic therapy.8 Similar results were found for adalimumab in the CHARM trial.9 Most recently, Schnitzler et al10 performed an observational study on 614 consecutive CD patients who received infliximab and were followed for a median of 55 months. Of the 347 patients who had a sustained clinical benefit on infliximab, the rate of major abdominal surgery prior to infliximab was significantly higher than the rate after the start of infliximab (P < 0.001). The reverse was true of the 185 patients who were either primary nonresponders or who stopped infliximab secondary to a loss of response; their surgical rate increased after starting infliximab.

The University of Pittsburgh medical center is a large academic institution with a growing referral-based inflammatory bowel disease (IBD) program. Our primary hypothesis was that the volume of total CD patients as well as the rates of infliximab use have increased over the past 12 years and this would correlate with a decrease in the rate of small bowel resectional surgery. Our secondary hypothesis was that, over time, fewer surgeries would be performed for failure of medical therapy (nonstricturing, nonpenetrating disease).

PATIENTS AND METHODS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Patients

Following University of Pittsburgh Institutional Review Board (IRB) approval and in keeping with the Health Insurance Portability and Accountability Act, inpatient and outpatient records were retrospectively obtained from the Medical Archival System (MARS, Medical Archival System, Pittsburgh, PA). MARS is a repository for information forwarded from the University of Pittsburgh Medical Center health system's electronic clinical, administrative, and financial databases.11 MARS data are indexed on every word and will recover all encounters with a given patient between specific dates. Data were de-identified (De-ID Software, University of Pittsburgh) by an honest broker (a person independent of the research) and obtained with a waiver of informed consent. Patients were identified from the University of Pittsburgh Medical Center, Presbyterian campus (the primary setting of most major surgeries in the health system) inpatient discharge abstracts (between January 1, 1995 and June 30, 2007) by using a primary or secondary diagnosis code of 555.1–555.9 (CD) from the International Classification of Diseases, Ninth Revision, Clinical Modification (ICD-9-CM). Among this group of CD patients, we then determined who had undergone a small bowel resection (SBR); ICD-9 procedure codes searched: 45.33, 45.51, 45.61, 45.62, 45.63, 45.72, 45.73, 45.90, 45.91, 45.92, 45.93, 46.01, 46.20, 46.21, 46.22, 46.23, 46.40, 46.41, 46.51, 46.73, 46.74, 46.79, 46.81, 46.93, 46.99, 57.83, 57.84. Only excisional small intestinal surgeries were taken into account. We limited patients to those who underwent an SBR in order to define a more homogeneous phenotype. These patients were then divided into 4 consecutive time periods based on when they had their surgery: January 1, 1995 to December 31, 1998 (Period 1); January 1, 1999 to December 31, 2001 (Period 2); January 1, 2002 to December 31, 2004 (Period 3); and January 1, 2005 to June 30, 2007 (Period 4). Timing of the individual periods was adjusted to attain a relatively equal distribution of surgeries per period. Patients who underwent more than 1 small bowel resection on separate admissions were noted.

Following de-identified document review, patients were included if their ICD-9 diagnosis code was consistent with a diagnosis of CD, if the procedure code matched the operation performed, and if a small bowel specimen was sent to pathology. Exclusion criteria included patients found to have ulcerative colitis (UC) or a disease other than CD; surgery for isolated colonic or perianal disease; history of any type of transplant surgery; isolated strictureplasty; dysplasia or cancer found on pathology; surgery for reasons other than CD (e.g., a mechanical obstruction secondary to adhesions where no inflammation or stricture is found at surgery); or an incomplete medication list.

Study Design

For each surgery, de-identified patient information was reviewed for the 6 months preceding the surgery. This included all progress notes, letters, discharge summaries, medication use (contained within the body of any note), imaging and colonoscopy reports, as well the operative and surgical pathology reports. For each surgery the following information was abstracted: age, sex, race, smoking status (active in the last 6 months), and disease duration. History of a previous small or large bowel surgery was also recorded (minor surgery for perianal fistulotomy or any abscess drainage was not included). CD medication use in the 6 months preceding surgery (including 5-aminosalicylic acid [5-ASA] products, IMs, anti-TNF agents, corticosteroids) was tabulated.

During document abstraction, disease behavior was noted and classified according to the definitions provided by the Montreal Working Party.12 Disease behavior was determined from radiographic, endoscopic, surgical, and histologic findings. According to the Montreal Classification, disease behavior B1 refers to nonstricturing, nonpenetrating disease; B2 refers to stricturing disease; and B3 refers to penetrating disease. Perianal disease is listed separately. During our review, any mention of “narrowing,” “stenosis,” or “stricturing” on any modality qualified as B2 disease. Patients were classified with B3 disease if there was note of a “fistula,” “sinus tract,” “abscess,” or “inflammatory mass,” even in the presence of concurrent stricturing disease. Presence or history of perianal fistulae or abscesses did not qualify as B3 disease.

Proportions of patients undergoing small bowel surgery were calculated according to calendar cohort, and these rates were then examined for time trends. The at-risk cohort, or denominator, was the total number of unique CD patients seen at the University of Pittsburgh Medical Center, Presbyterian campus, within a given time period (Periods 1–4). This included patients with any combination of upper, small bowel, colonic, or perianal disease. While most patients were seen by a gastroenterologist at our institution, some were referred directly for surgery.

In a secondary analysis we sought to determine whether disease behavior was reclassified as a result of surgical findings. Specifically, we focused on the conversion from B2 to B3 disease; conversion from B1 to B2 or B3 disease was not analyzed as there was only 1 B1 patient in our sample. Patients who had documented preoperative B3 disease, or who had imaging (including computerized tomography, small bowel follow-through, or barium/gastrograffin enema) prior to surgery that showed or suggested penetrating disease were classified in the “preoperative B3” group. The “postoperative B3” group consisted of all B3 patients, including those who were newly found to have penetrating disease on the basis of operative findings or surgical pathology. Three patients were excluded from this analysis because the preoperative phenotype was unclear from the medical records.

Statistics

Means and standard deviations were calculated for all continuous variables; proportions were calculated for all binary and categorical variables. Overall surgical rates and the effect of medications across the 4 time periods were compared using a chi-square test of trend with alpha set at 0.05.13 All comparisons were made using SPSS 16.0 (SPSS, Chicago, IL).

RESULTS

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

Demographics

In all, 360 patients with a total of 403 surgeries met our inclusion criteria. Of these, 133 patients with a total of 167 surgeries were excluded. The most common reasons for exclusion included a diagnosis of indeterminate colitis or UC, incomplete medication data, or missing operative or histology reports. This left 227 CD patients with a total of 236 small bowel surgical procedures who were included in the study. Nine patients had 2 surgeries in separate time periods.

The demographic and clinical features according to time period are presented in Table 1. Patient characteristics, including gender, race, age, smoking status, prior surgery, and duration of disease were generally similar over time. Of the 122 total patients who had a prior surgery, 88 had a small bowel or ileocolonic resection, 19 had a small bowel and separate colonic resection, and 15 had an isolated colonic resection.

Table 1. Demographics
 Period 1 1995-1998Period 2 1999-2001Period 3 2002-2004Period 4 2005-2007
  1. No significant differences across time for any variables. Because of incomplete smoking data, denominator for Periods 1 to 4: 44, 55, 47, 61. The denominator for disease duration for Periods 1 to 4: 54, 52, 45, 57. SD, standard deviation.

Total surgeries58605365
Females, n (%)34 (58.6)33 (55)30 (56.6)33 (50.8)
Caucasian, n (%)56 (96.6)55 (91.7)50 (94.3)60 (92.3)
Age, years, mean + SD37.3 + 13.938.8 + 13.036.8 + 13.338.7 + 12.1
Smoking, last 6 mo, n (%)16 (36.4)24 (43.6)17 (36.2)23 (37.7)
Prior surgery, n (%)31 (54.4)33 (55)25 (47.2)33 (50.8)
Disease <1 yr4 (7.4)4 (7.7)8 (17.8)6 (10.5)
Duration 1-5 yr5 (9.3)8 (15.4)3 (6.7)10 (17.5)
n (%) 6-10 yr11 (20.4)10 (19.2)8 (17.8)13 (22.8)
   >10 yr34 (63)30 (57.7)26 (57.8)28 (49.1)

Medication Use

5-ASA use primarily included mesalamine and sulfasalazine; 2 patients were on olsalazine and 1 was on balsalazide. Most IMs use consisted of azathioprine or 6-mercaptopurine; only 3 patients were on methotrexate. The anti-TNF patients were all on infliximab except for 1 patient who received adalimumab. The relative frequency of medication use in the 6 months preceding surgery is shown in Figure 1. Mesalamine use did not change over time, with 62.1% of patients using it in Period 1 and 49.2% using it in Period 4. The chi-square test of trend was not significant (P = 0.61). Although the frequency of IM use appeared to increase over time linearly (37.9%, 43.3%, 50.9%, 58.5% for Periods 1–4, respectively), the test of trend was not significant (P = 0.55). TNF antagonist use increased from 0% in Period 1 (preinfliximab era) up to 35.4% in Period 4. This increase was significant by chi-square test of trend at P = 0.0002.

thumbnail image

Figure 1. Medication use in the 6 months preceding small bowel resection surgery. **P = 0.0002 for test of trends across time periods. 5-ASA, 5-aminosalicylate; IM, immunomodulator; anti-TNF, antitumor necrosis factor; CS, corticosteroids.

Download figure to PowerPoint

The proportion of patients on prednisone (minimum of 5 mg daily) or budesonide (minimum of 3 mg daily) at some point in the 6 months preceding surgery were 74.1%, 63.3%, 62.3%, and 64.6% for Periods 1–4, respectively; test of trend was not significant (P = 0.93).

Surgical Rates

The total number of unique CD patients seen at the University of Pittsburgh increased over the 4 time periods: 229 per year in Period 1, 378 per year in Period 2, 387 per year in Period 3, and 581 per year in Period 4. Despite this rise in patient volume, as well as the increase in TNF antagonist use, the rate of small bowel resections per year per period remain unchanged over time, as shown in Table 2: 1.6%, 1.9%, 1.6%, 1.9% for Periods 1–4, respectively (P = 0.93).

Table 2. Surgical Rates
Time PeriodTotal CD Patients per YearTotal SB Surgeries per YearSurgical Rate per Year
  1. Rate of small bowel resectional surgery over time. CD, Crohn's disease; SB, small bowel.

1229151.6
2378221.9
3387191.6
4581271.9

Disease Behavior over Time

The relative percentage of postoperative disease behavior in patients undergoing surgery over the 4 time periods is shown in Figure 2. Based on our low threshold for defining B2 (stricturing) disease, only 1 patient (in Period 1) was found to have B1 (nonstricturing, nonpenetrating) disease. We found that the relative frequency of B2 and B3 disease did not change over time (P = 0.98).

thumbnail image

Figure 2. Disease behavior. Breakdown of B1, B2, and B3 disease over time. Numbers in bars are the percentage breakdown per time period. B1 disease represents 1.7% of patients in Period 1, 0% for Periods 2 to 4. B1, nonstricturing, nonpenetrating disease. B2, stricturing disease. B3, penetrating disease.

Download figure to PowerPoint

Change in Behavioral Phenotype Based on Surgical Findings

Preoperatively, 108/233 (46.4%) patients were grouped as having B2 (stricturing) disease, while 125/233 (54.7%) had B3 (penetrating) disease. Based on the gross and microscopic findings at surgery, 37 of the 108 (34.3%) patients with B2 disease were reclassified as having B3 disease. Thus, postoperatively, 71/233 patients (30.5%) remained B2s, while 162/233 (69.5%) were categorized as having B3 disease. This is shown in Table 3. Given that just over half of the patients had had a prior resection, and that the pathology for these surgeries was largely unavailable, we were concerned that we were overestimating the proportion of newly discovered B3 disease. Thus, we repeated our analysis for patients who were surgery-naïve; the results were unchanged (data not shown). Finally, conversion from B2 to B3 disease did not change over time: 21.1%, 33.3%, 21.1%, 15.9% for Periods 1–4, respectively; test of trend, P = 0.49.

Table 3. Disease Behavior Change at Surgery
 Postop B2Postop B3
  1. Conversion from B2 to B3 disease as a result of operative findings. Number of surgeries (percentage of total surgeries performed – denominator = 233 surgeries). Preop, pre operative; Postop, postoperative; B2, stricturing disease; B3, penetrating disease.

Preop B271 (30.5)37 (14.8)
Preop B30 (0)125 (54.7)

DISCUSSION

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES

In this study of the experience at a large tertiary care IBD center, we found that TNF antagonist use in the 6 months preceding small bowel resectional surgery for CD has increased over time. However, we observed that the overall rate of small bowel resectional surgery has remained unchanged. Our annual surgical rate of 1.6%–1.9% is lower than what has been noted in the literature.3 This is primarily attributed to excluding patients who underwent anything other than an SBR.

Other than failure of medical therapy, there are a number of potential explanations as to why we did not observe a decrease in the rate of surgery over time. First, in a majority of patients the duration of disease was greater than 10 years. Given the natural history of CD behavior over time, after 20 years 80% of patients will have developed complications of stricturing or penetrating disease.14 Thus, it is unlikely that medical therapy with immunomodulators or anti-TNF agents at this stage might impact the need for surgery. To better evaluate the natural course of disease with anti-TNF agents would require following patients on anti-TNF treatment at the diagnosis of disease for at least 1 decade. Second, the at-risk cohort patients who received infliximab included primary and secondary nonresponders as well as sustained responders. In theory, the sustained responders had lower surgical rates, in concordance with the results of Schnitzler et al10; however, our study was not designed to answer this question. Third, even if medications were contributing to a decrease in surgical rates, this difference could be neutralized by a rise in complex medication-refractory tertiary referrals who warrant surgery. Finally, with the advent of laparoscopic techniques, more patients may be undergoing a limited SBR earlier in the disease course. This would be consistent with the findings of Beaulieu et al,15 who found that patients diagnosed with CD after 2000 actually had twice the risk of undergoing surgery within 2 years of diagnosis than those patients diagnosed in the 1990s.

In terms of disease behavior, we found that the indications and findings at surgery have remained constant over time, with a stricturing to penetrating ratio of 1:2, a ratio unchanged by the increasing use of TNF antagonists. Notably, nearly all the patients in our cohort underwent surgery for complicated disease, whether stricturing or penetrating. In fact, only 1 surgery was clearly performed for nonstricturing, nonpenetrating disease (B1). Penetrating disease (even if occult and found at the time of surgery) was found in two-thirds of the patients. This ratio remained stable over the 4 consecutive time cohorts. These findings are in line with those of Cosnes et al,3 who observed that the indication for intestinal resection among a subset of 565 patients did not significantly change over time. Unlike Cosnes et al, however, we were not able to detect any trend in fewer surgeries being performed for failure of medical therapy (inflammatory, B1 disease).

Interestingly, we found that over one-third of patients who were preoperatively diagnosed with stricturing (B2) disease had penetrating (B3) disease. One explanation for this is that prestenotic dilation gives rise to a subtle perforating lesion that may not be easily diagnosed on imaging. However, conversion rates from B2 to B3 disease did not decrease over time; this suggests that improvements in imaging and detection of penetrating disease cannot fully predict all B3 findings at the time of surgery. The clinical implications of previously undetected disease are emerging: Sachar et al16 found that patients with B3 phenotype at the time of surgery were far more likely to have complicated, symptomatic disease in the 3 years following surgery than patients with B2 disease.

The strength of this study is the large sample size derived from a comprehensive electronic medical records database. Additionally, the discernment of behavioral phenotype was based not on symptoms, but the findings at surgery. The primary limitation of our study was the inability to better characterize the at-risk cohort. Our at-risk cohort was defined by the total number of unique CD patients seen at the University of Pittsburgh within a given time period. However, we did not have medication, phenotypic, or demographic data on this cohort. It is certainly possible that immunomodulator and/or anti-TNF use over time in the at-risk cohort was different from the rates seen among those who underwent SBR. Additionally, although we limited our analysis to patients with small bowel disease, the at-risk cohort included patients with any combination of small bowel, colonic, perianal, or upper GI disease. Furthermore, it is conceivable that a patient in the at-risk cohort was seen only once and subsequently had surgery at a different institution. Finally, even if we can associate emerging TNF antagonist use with surgical rates, we cannot make any statements about causality.

Another important limitation is that our results do not necessarily apply to a community practice setting. Most likely, both medical and surgical therapies were more aggressive in the setting of a tertiary referral-based institution. Secondary limitations included an inability to quantify the full duration or dose of the analyzed medications. We also likely underestimated the frequency of B1 disease; per our protocol, any mention of narrowing on any modality qualified for stricturing (B2) disease.

In sum, we found that the rate of small bowel resectional surgery for CD at the University of Pittsburgh has remained unchanged despite increased utilization of anti-TNF agents. Also, the indications for surgery have remained unaltered; two-thirds of patients were ultimately found to have penetrating disease. As azathioprine, 6-mercaptopurine, and infliximab have been demonstrated to prevent postoperative endoscopic recurrence,17–19 it will be interesting to see if reoperation rates will decline in the setting of early medical intervention.

REFERENCES

  1. Top of page
  2. Abstract
  3. PATIENTS AND METHODS
  4. RESULTS
  5. DISCUSSION
  6. REFERENCES