Temporal trends in imaging and associated radiation exposure in inflammatory bowel disease

Authors


  • Disclosure: None declared.

Correspondence to:

Richard Pollok,

Department of Gastroenterology, Knightsbridge Wing, St George's Hospital NHS Trust, Blackshaw Road, SW17 0QT, London, UK

Tel.: + 0208 725 1206

Fax: + 0208 725 3520

Email: richard.pollok@nhs.net

Summary

Objectives

Increasing use of diagnostic imaging in inflammatory bowel disease (IBD) has led to concerns about the malignant potential of ionising radiation in a cohort that have an increased lifetime risk of gastrointestinal malignancy. The aim was to quantify radiation exposure in IBD patients referred from primary care, determine predictors of high exposure and evaluate temporal trends in diagnostic imaging over a 20-year period.

Methods

This was a retrospective cohort study whereby IBD patients were recruited from the outpatient clinic and evaluated retrospectively. The total cumulative effective dose (CED) received from tests was calculated for each subject. Cox regression was performed to assess factors associated with potentially harmful levels of ionising radiation defined as total CED > 50 milli-sieverts (mSv; equivalent to five CT abdomen scans).

Results

The cohort included 415 patients. Median total CED was 7.2 mSv (IQR: 3.0–22.7) in Crohn's disease and 2.8 mSv (IQR: 0.8–8.9) in ulcerative colitis patients, respectively. A total of 32 patients (8%) received a CED > 50 mSv. A history of IBD-related surgery was associated with high exposure (HR 7.7). During the study period, usage of abdominal CT increased by 310%.

Conclusion

Approximately 1 in 13 patients in the study cohort were exposed to potentially harmful levels of ionising radiation. Strategies to minimise exposure to diagnostic medical radiation in IBD patients are required.

What's known

  • Increasing use of diagnostic imaging in IBD has led to concerns about the malignant potential of ionising radiation in a cohort that is already predisposed to malignancy. There is limited data, largely from referral centre studies quantifying radiation exposure from diagnostic imaging in IBD patients and several clinical predictors have been implicated. No studies have evaluated time to excessive exposure or demonstrated trends in imaging over the survey period.

What's new

  • This large study from the UK designed to minimise referral centre bias has demonstrated that about 1 in 13 patients with IBD were exposed to potentially harmful levels of diagnostic medical radiation. In addition, Kaplan–Meier analysis evaluated time to excess exposure.

Introduction

Inflammatory bowel disease (IBD) includes Crohn's disease (CD) and ulcerative colitis (UC) which are lifelong idiopathic disorders characterised by gastrointestinal (GI) inflammation. Extraintestinal manifestations may be present in up to 40% of patients [1]. The disease burden of IBD is significant, with 1.4 million and 2 .2 million affected in United States and Europe, respectively [2].

Diagnostic imaging plays a vital role in the diagnosis and management of IBD. Repeated examinations are often required, to determine the disease extent and severity, diagnose complications, monitor response to therapy, during perioperative evaluation and to evaluate extraintestinal manifestations of IBD. Patients with IBD are already at increased risk of colorectal and small intestinal cancers [3-7] from longstanding disease and therapy may increase the risk of malignancy [8, 9]. There is therefore concern regarding the increased risk of exposure to potentially harmful levels of diagnostic medical radiation (DMR) and associated cancers in this group. The risk of radiation-induced malignancy is higher in younger patients particularly those less than 35 years [10] which is pertinent as 20% of patients with IBD are diagnosed in childhood [11].

Studies of atomic bomb survivors and nuclear industry workers indicate that protracted exposure to ionising radiation increases the risk of malignancy [11-15]. Radiation exposure as little as 50 milli-sieverts (mSv) from DMR has been implicated in the development of solid tumours, particularly of the colon and urogenital tract [12]. It is estimated that up to 2% of malignancies could be attributed to DMR [16], with 5500 deaths attributable to radiation-induced cancer in the United States each year in the general population [16, 17]. The US National Research Council estimates that one patient will develop a radiation-induced cancer in their lifetime out of every 1000 patients undergoing a 10 mSv CT abdominal scan [18]. There is therefore concern that DMR may be contributing to the excess cancer risk amongst IBD patients. The aim of this study was to estimate radiation exposure in IBD patients referred directly from primary care over a 20-year period in a single UK centre. In addition, we sought to determine clinical predictors of those receiving excess radiation exposure and examine temporal trends in diagnostic imaging modalities.

Methods

Study design

This retrospective study was conducted at the Departments of Gastroenterology and Radiology, St George's Hospital (London, UK). The study was approved by the local ethics committee and the Trust's Research and Development department.

Study population

To avoid selection bias, subjects attending clinic between 5 January 2011 and 30 June 2011 were consecutively recruited and evaluated retrospectively. The cohort comprised only those with a known diagnosis of IBD made using established criteria [19]. All patients with a concomitant chronic condition which may require additional imaging were excluded. Demographic and clinical data were obtained by scrutinising medical records including surgical operations, disease characteristics and current or previous medical therapy. Those with less than 12 months follow-up were excluded allowing time trend evaluation of investigations stratified by calendar year. Tertiary referrals are seen at our hospital, but to avoid tertiary centre bias, only patients referred directly from primary care were included. The follow-up period was defined as the period from the date of diagnosis, as evidenced from medical records to the end of study period, 1 July 2011.

Outcome measures

Number and type of imaging procedures performed were obtained from the radiology database containing data from 1990. Patients diagnosed before this time were not included in this study. As extraintestinal manifestations may occur in 40% of patients with IBD [1], which may require radiological assessment and patients may have had imaging unrelated to their IBD diagnosis, we included non-GI imaging in our analysis. For the purposes of this study, all procedures were assigned to anatomical categories using a method that has been described previously [20].

Radiation doses by respective imaging modalities are shown in Table 1. The effective dose of radiation from each test was estimated from published standardised tables [21]. Cumulative effective dose (CED) was calculated for each subject by summing the effective doses of radiation from the time of diagnosis until the end of study period. A CED > 50 mSv was defined as potentially harmful as this definition has been used in previous studies allowing comparisons to be made, but more importantly epidemiological data suggest that a protracted exposure above this level is associated with an increased risk of cancer [10].

Table 1. Radiation doses for common diagnostic tests [21]
Diagnostic imagingEffective dose (mSv)
CT abdomen10
Small bowel follow-through3
Plain abdominal X-ray0.7
Plain chest X-ray0.02
Thoracic spine0.64

Statistical analysis

The database was constructed using Excel 2010 (Microsoft Corp, Redmond, WA) and STATA version 11.0 was used for statistical analysis. Comparisons were made using Fisher's exact test for categorical variables, unpaired t-test for normally distributed continuous variables, and the Mann–Whitney test for continuous variables with a skewed distribution. Predictors of exposure to potentially harmful levels of ionising radiation defined by a total CED > 50 mSv was evaluated using univariable and multivariable cox regression. As disease duration increases the likelihood of radiation exposure, we performed a Kaplan–Meier analysis assessing cumulative probability of exposure to total CED > 50 mSv in the whole cohort stratified by IBD type and whether or not they had IBD-related surgery. Patients were censored when and if they reached the threshold of CED > 50 mSv or until end of follow up.

Results

The cohort included 423 patients, but 8 were excluded because of a concomitant chronic condition requiring regular imaging. The study population included 415 patients in total. The demographic and clinical characteristics are presented in Table 2. There were significant differences observed between patients with CD and UC in relation to their age at diagnosis, disease duration, occurrence of IBD-related surgery, therapy and smoking status (p < 0.05). The median total CED in all IBD patients was 4.4 mSv (IQR: 0.8–15.3). The median total CED was 7.2 mSv (IQR: 3.0–22.7) and 2.8 mSv (IQR: 0.8–8.9) in CD and UC patients, respectively. A total of 32 patients (8%) were exposed to a total CED > 50 mSV, consisting of 29 CD and 3 UC patients, respectively. In these 32 patients, mean time to reaching 50 mSv was 10.2 years (SD ± 5.3) and range 0.4–20 years. Distribution of total CED in the study population is shown in Figure 1A, B.

Table 2. Demographic and clinical characteristics
ParameterCrohn's disease (n = 217)Ulcerative colitis (n = 198)
  1. Values in parenthesis represent percentages (except for age, disease duration).

  2. a

    At recruitment for CD, at diagnosis for UC.

  3. IQR, interquartile range; SD, standard deviation.

Age at diagnosis (year) mean (SD)30.8 (± 14.2)36.9 (± 15.1)
Male92 (42)96 (49)
Disease duration (years) median (IQR)8.3 (4.0–14.5)7.7 (3.6–13.5)
Montreal classificationaAge
A1 (<16 years) = 24 (11)Proctitis = 36 (18)
A2 (17–40 years) = 153 (71)Left-sided colitis = 67 (34)
A3 (> 40 years) = 40 (18)Pancolitis = 95 (48)
Location 
L1 Ileal = 24 (11) 
L2 Colonic = 77 (35) 
L3 Ileocolonic = 116 (54 
Behaviour 
B1 Inflammatory = 113 (52) 
B2 Stricturing = 66 (30) 
B3 Penetrating = 38 (18) 
P Perianal = 32 (15) 
Extraintestinal manifestations22 (10)11 (6)
IBD-related surgery0 = 121 (56)0 = 183 (97)
1 = 71 (33)1 = 5 (3)
> 1 = 25 (12)> 1 = 0 (0)
Family history of IBD8 (4)4 (2)
Smoking (at diagnosis)Never = 205 (94)Never = 195 (98)
Ex-smoker = 1 (< 1)Ex-smoker = 2 (1)
Current smoker = 11 (5)Current smoker = 1 (1)
5ASA131 (61)177 (90)
Immunomodulator144 (66)88 (45)
Biologics31 (14)5 (3)
Corticosteroids within 3 months of diagnosis48 (23)69 (35)
Figure 1.

(A) Distribution of estimated radiation exposure in patients with Crohn's disease. (B) Distribution of estimated radiation exposure in patients with ulcerative colitis

Factors associated with a CED > 50 mSv are presented in Table 3. Univariate analyses suggested that gender, disease type, corticosteroid use within 3 months of diagnosis, immunomodulators, extraintestinal features and history of IBD-related surgery were significantly associated with total CED > 50 mSv. After multivariate analysis, only gender [HR 2.21 (1.07–4.56)] and IBD-related surgery [HR 7.76 (3.34–18.10)] were found to be independent predictors. Males were at an increased risk compared with females, with the occurrence of a high radiation dose over twice that for females. Patients undergoing IBD-related surgery were eight times more likely to have a total CED > 50 mSv.

Table 3. Cox regression analysis associated with total CED > 50 mSv
VariableUnivariableMultivariable
Hazard ratio (95% CI)p-valueHazard ratio (95% CI)p-value
  1. a

    Hazard ratio given for 10-year increase in age.

  2. IBD, inflammatory bowel disease.

Male sex2.16 (1.05, 4.45)0.042.21 (1.07, 4.56)0.03
Age (at diagnosis) a0.82 (0.66, 1.11)0.23 
Crohn's disease7.57 (2.30, 24.9)0.001 
Steroids within 3 months of diagnosis0.25 (0.08, 0.83)0.02 
Immunomodulator use2.87 (1.23, 6.68)0.02 
Biologics2.14 (0.88, 5.25)0.10 
Extraintestinal features2.51 (1.02, 6.13)0.04 
IBD-related surgery7.72 (3.32, 18.0)< 0.0017.76 (3.34, 18.10)< 0.001

In total, 3742 procedures were performed in the whole cohort over a 20-year period which included: plain radiographs, CT scans, magnetic resonance imaging (MRI) scans, nuclear medicine imaging, fluoroscopic procedures and ultrasound. Of these procedures, 2296 were performed to evaluate the GI tract and 1446 were non-GI related. The relative percentage contribution of all procedures by anatomical category is presented in Figure 2. Abdominal exposure contributed to 89% of total CED and thereafter chest imaging at 7%.

Figure 2.

Relative percentage contribution of all procedures by anatomical category performed over a 20-year period

Time trends in imaging over a 20-year period are presented in Figure 3A, B. Over the study period, CT of the abdomen and/or pelvis increased from 0% (1990–1994) to 31% (2008–2010), an increase of 310% overall. The 5-, 10- and 20-year estimates of the cumulative proportion of patients with a CED > 50 mSv for CD were 3%, 11% and 32% and 0%, 0% and 9% for UC, respectively (log rank p = 0.0001). For those who had IBD-related surgery, the 5-, 10- and 20-year cumulative probability of exposure to total CED > 50 mSv was 5%, 19% and 45%, respectively, compared with 0%, 2% and 10% for those who did not have surgery (log rank p = 0.001) (Figure 4A, B).

Figure 3.

(A and B) Relative percentage of gastrointestinal investigations used in the management of Crohn's disease and ulcerative colitis by calendar period

Figure 4.

(A and B) Kaplan–Meier analysis showing cumulative probability of being exposed to CED > 50 mSv from diagnosis according to inflammatory bowel disease (IBD) type and previous IBD-related surgery

Discussion

Principal findings

In this cohort of IBD patients, about 1 in 13 patients were exposed to potentially harmful levels of radiation. Radiation exposure was significantly higher in CD patients than in UC patients, which reflects the natural history of CD, whereby patients are more likely to have repeated imaging during their disease course to differentiate disease activity from complications. This is a cause for concern as protracted exposure to low levels of radiation maybe associated with lifetime risk of malignancy [13, 14, 16, 17, 22]. The likelihood of receiving a total CED > 50 mSv was about eight times greater for those with a history of IBD-related surgery [HR 7.76 (3.34–18.10)]. This is probably explained by the fact that imaging is usually required pre- and postoperatively for several reasons and CT imaging remains the prominent modality as it allows rapid assessment and is widely available in the acute setting. Males appeared to be at a marginally increased risk [HR 2.21 (1.07, 4.56)]. This small difference is unlikely to be of clinical significance. Time trends in imaging over a 20-year period revealed an evolution in keeping with recently published BSG [23] and ECCO [24] guidelines recommending increased use of MRI and intestinal ultrasonography to reduce radiation exposure. However, during the same period, CT usage also increased significantly. Kaplan–Meier analysis indicated that cumulative probability of exposure to a CED > 50 mv increased with disease duration after stratification by IBD type and IBD-related surgery.

Findings in relation to other studies

Our results are comparable to those of Newnham et al. [25] who calculated ionising radiation exposure in 100 consecutive patients and found 11% of the cohort received CED > 50 mSv. Levi et al. [20] studied 324 patients and found that 7% were exposed to potentially harmful levels of radiation and predictors identified were surgery, prednisolone use, disease duration, first year of disease and age. Three other studies [11, 26, 27] of similar design from single tertiary centres reported a lower prevalence of those exposed to a CED ≥ 50 mSv of 5%. Desmond et al. [28] have reported the highest prevalence with 24% receiving a CED > 50 mSv; their long study period and more severe disease may explain this higher exposure. In our cohort, CT usage increased 310% from 0% between 1990 and 1992 increasing to 31% of all procedures in 2008–2010. A similar population-based study from North America found that CT usage increased 840% from 375 studies in 2003 to 3166 studies in 2007 [29]. Our results were comparable to previous studies in that CT contributed most to the total radiation dose. In the United States, CT has almost completely replaced small bowel follow-through (SBFT) as the primary diagnostic modality to image the small bowel in CD [29]. By contrast, a recent UK survey has revealed that SBFT remains the most frequently requested and performed imaging to assess small bowel [30].

Study limitations

The retrospective nature of this study may have led to an under estimation of true radiation exposure as tests performed at other centres may not have been captured and we did not contact patients GP's to evaluate this further. The estimated radiation dose may have been greater or less than true exposure as standardised published doses were used. This study was conducted at a single centre and as such may not reflect wider practice; in particular, our centre has an interest in the use of white cell scans [31] and small bowel ultrasound [32], both of which may have impacted on total radiation exposure. Although the centre functions as a tertiary centre, an attempt was made to reflect practice in secondary centres by selecting only those patients referred directly from primary care, but we accept this remains a potential bias. There is also the possibility that we may have over-recruited those with a more severe disease phenotype and therefore higher radiation exposure as this group may have attended more than once over the 6-month recruitment period; however, this was a small minority involving only five patients. Furthermore, we attributed an exposure of 10 mSv from abdominal CT which is relatively high in the context of newer generation CT scanners, but is appropriate given the time frame of the study.

Clinical implications

Our study indicates, in common with others, that a significant proportion of IBD patients are at risk of excessive DMR exposure. The findings are particularly concerning as the majority of patients in these studies are young adults in whom there is an increased risk of radiation-induced malignancy [10].

In addition, these patients, given their young age at presentation, have many more years of potential radiation exposure ahead of them. Knowledge of the risks of DMR amongst clinicians is limited [33-35]. In routine clinical practice, cumulative exposure to radiation is usually not recorded. Healthcare professionals and radiology staff may not have protocols in place to communicate, filter and flag up those at risk of excessive exposure to DMR.

A number of studies [11, 20, 26, 28, 36] including our own, have found that IBD-related surgery was a predictor of increased CED, which is likely to reflect the use of CT imaging out of hours. Kroeker et al. [11] found that 34% of CT scans in IBD patients were performed in the emergency department. Furthermore, these findings raise questions about the need for alternatives to CT in the out-of-hours setting. Recent advances in CT technology may allow significant reductions in radiation dose compared with current techniques. CT-related radiation can be reduced either by lowering the X-ray tube current (mA) or tube potential (kV) without sacrificing diagnostic performance [37].

A recent prospective study demonstrated that MR and CT enterography are equally accurate in the assessment of disease activity and complications in ileocolonic CD [38]. A meta-analysis comparing ultrasonography, MRI and CT for the diagnosis of IBD showed high sensitivity and specificity with no statistical difference between modalities [39]. With increasing awareness of radiation exposure because of CT imaging, and improving availability of MRI facilities, it is likely that MR enterography will continue to emerge as an important diagnostic tool in the routine assessment of patients with IBD. Ultrasound with or without an oral agent also needs to be considered as an alternative or complementary technique that has proven diagnostic accuracy [40-43]. There are also other endoscopic modalities available to assess the small bowel such as capsule endoscopy and single or double balloon or spiral enteroscopy, however, these techniques are not widely available and currently not considered first line.

It is possible that the use of imaging may increase further in the management of patients with IBD pending recent research interests evaluating whether end-points including mucosal healing and radiological remission have an impact on hospitalisation and surgery. We propose clinicians should log cumulative exposure and consider the idea of a radiology diary. It is also possible that further strategies such as the Image Gently Campaign [44], a US-based awareness campaign to reduce DMR exposure in the paediatric population may further impact on radiation exposure. Furthermore, CT imaging should be restricted to cases where there is a robust indication that is likely to change patient management, and when indicated, the use of low-dose CT protocols should be encouraged. In addition, whenever feasible, alternatives including MRI and ultrasound should be considered as an alternative.

In conclusion, about 1 in 13 patients with IBD were exposed to potentially harmful levels of ionising radiation in this cohort and taken together with other studies gives cause for concern about the potential consequent increased cancer risk amongst an already predisposed cohort. We propose clinicians log cumulative radiation exposure and implement strategies to minimise exposure to DMR.

Author contributions

AP conceptualised the study, SC and RCP designed the study, SC and RH collected the data. Statistician performed statistical analysis, SC wrote the article and AP, RG, RCP revised and approved the article.

Acknowledgements

David Johnstone, Principle Physicist, Radiology Protection Centre, St George's Healthcare NHS Trust. Paul Bassett, Statistician, Biostatistics, University College London, London.

Ancillary