Thirty-four of the thirty six studies included for review showed a positive association between ionizing radiation from medical imaging and increased risk of cancer[19,21-27,29-51,53-56].

A recent direct study of CT scan use and cancer risk by Pearce et al[21] showed a leukaemia relative risk of 3.18 (95%CI: 1.46-6.94) for children and young adults who received a cumulative dose of > 30 mGy and a brain tumour relative risk of 2.82 (95%CI: 1.33-6.03) for children and young adults who received a cumulative dose of 50-74 mGy. This corresponded to an estimated absolute risk of about 1 excess leukaemia case and one excess brain tumour case for every 10000 patients who undergo one head CT scan before the age of 10[21].

Another large retrospective study in the United States[23] reported a modest increase in cancer risk secondary to low dose (50-100 mSv) and high dose (> 100 mSv) radiation from CT scans in the elderly. In this study, an estimated 1659 (0.03%) and 2185 (0.04%) cancers were related to ionizing radiation from two cohort populations of over five million patients each[23]. Berrington de González et al[31] reported that an extra 29000 (95%UL 15000-45000) cancer cases could be attributable to the 57 million CT scans performed in the United States during 2007. Nearly 30% of the scans were estimated to be performed in patients aged 35-54 years, 13% in those aged 18-34 years and 7% in persons aged 18 or less[41]. The projected risks in females were higher for scans that exposed the chest due to the additional risk of breast cancer and higher lung cancer coefficients[41].

Two studies, Blettner et al[52] and van Walraven et al[28] reported no statistically significant increased risk of brain tumours and secondary abdomino-pelvic malignancies following medical ionizing radiation respectively. Patients in the van Walraven et al[28] study were assessed for secondary abdomino-pelvic malignancies associated with abdomino-pelvic CT scans use in the follow up of previous testicular cancer. Patients received a median radiation dose of 110 mSv (IQR 44-190) from medical radiation imaging at 5 years follow up, a dose which was associated with increased risk of cancer in other studies included for review[22-26,31,33,35,38,41,46].

A study of 18-35 years old participants by Zondervan et al[27] suggested that the majority of CT-induced cancers were from sporadic rather than frequent scanning. Whilst frequent scanning is associated with a significant cancer risk, it is usually reserved for the very ill, a population where a large proportion die before any radiation induced cancer may factor into their health[27].

Several studies assessed the risk of radiation exposure in children and young adults[21,24,27,29,32,41,42,44,49,51]. An Israeli study by Chodick et al[49] reported an absorbed brain dose (from a head CT) of 130 mGy for children aged < 3 years old to 30 mGy at age 16-18 years and a stomach dose (from an abdominal CT) of 51 mGy at age < 3-24 years mGy at age 16-18 years. Increasing age was associated with a reduction in cancer risk with the highest excess risk of 0.52% estimated for children aged < 3 years and 0.21% at age 16-18 years[49]. Berrington de González et al[41] estimated a mean lifetime cancer risk of 1 for every 1000 head CT scans at age 3 years and 1 year every 2000 head CT scans at age 15. For abdomino-pelvic CT scans, a lifetime cancer risk of 1 for every 500 scans was predicted at ages 3 and 15 and 1 every 1000 scans at age 30[41].

CT colonography is regarded as sensitive as optical colonoscopy and is sometimes used to detect large adenocarcinomas of the colon[57]. A CT colonography screening study estimated, using standard protocols, that patients would receive a dose of 8 mSv and 7 mSv for women and men respectively[31]. Assuming a CT colonography screen every 5 years from the age of 50-80 years, 150 radiation-related cancers resulted for every 100000 patients screened (95% CT uncertainty interval, 80-280)[31]. The number of colorectal cancers prevented from CT colonography, based on three microsimulation models, varied between 3580 to 5190 cases per 100000 patients screened, resulting in a benefit-risk ratio of 24:1 (95% CT uncertainty interval, 13:1-45:1) to 35:1 (95% CT uncertainty interval, 19:1-65:1)[31]. The benefit-risk ratio was higher for patients aged 65-80 relative to those aged 50-64[31].

A retrospective cohort study of 1424 patients by Woo et al[22] also showed a positive benefit-risk ratio when examining the mortality benefit from preventing a pulmonary embolism vs mortality risk from radiation induced cancer (benefit-risk ratio of 25 for patients in the emergency department or outpatient setting and 187 for inpatients).

Brenner et al[55] investigated the effects of full-body CT examinations which have become more popular in private independent radiology clinics. This study showed a single full-body CT scan in a 45-year-old adult would result in an estimated lifetime attributable cancer mortality risk of around 0.08% with 95% credibility limits being a factor of 3.2 in either direction[55]. An annual examination up till age 75 (30 examinations in total) was reported to increase the lifetime risk to 1.9% with 95% CT credibility limits being a factor of 2 in either direction[55].

The main objectives of this review was to provide an overview of the radiation risks involved with medical imaging and use this as a framework to better understand risks associated with the use of CT scans for surveillance in patients diagnosed with colorectal cancer.

The majority of studies included for review showed a positive association between ionizing radiation from medical imaging and increased cancer risk[21-27,29-51,53-56]. As with all medical procedures the dilemma lies in balancing the potential harm vs the benefit medical imaging provides. Meer et al[23] suggested that despite using conservative estimates and worst-case scenario methodology, the cancer risk was low in the elderly United States population even in patients who received dosages over 100 mSv. Whilst the risks are apparent, they need to be taken in context and two studies[22,31], which assessed the use of CT scans to detect potentially life-threatening illnesses (colorectal cancer and pulmonary embolism), showed a clear positive benefit-risk ratio. Instances in which medical imaging may not be justified include the use of full-body CT examination as a “screening” tool, where there is potential radiation associated cancer risk[55] but poor evidence regarding its effectiveness and life-prolonging benefits[58-60]. A typical dose from a single full-body CT scan was estimated to be 16, 14 and 10 mGy to the lung, GI tract and bone marrow respectively but subject to variability due to differences in CT scanners and protocols[55]. This equates to an effective dose (weighted average dose to all oragns) of around 12 mSv and an excess lifetime cancer mortality risk of 1.9% if 30 such scans were undertaken over a lifetime[55].

In children and young adults, an age dependent cancer risk was reported, with the risk decreasing as the patients became older, particularly for head CT scans[41-49]. A recent direct study investigating CT scan use and cancer risk in patients less than 22-year-old also showed a leukaemia and brain tumour risk approximately three times higher when receiving a cumulative dose of > 30 mGy and 50-74 mGy respectively[21]. This approximates to 5-10 and 2-3 head CT scans in children < 15 years for the corresponding leukaemia and brain tumour risks stated above, respectively[21]. Children are considered more radiosensitive to the oncogenic effects[61-74], may have a longer lifetime risk to develop cancer (particularly sarcoma, lymphoma and breast carcinoma)[49] and receive higher doses relative to adults due to their smaller body size and relative attenuation[75].

Two studies[28,52] did not show a statistically significant association between medical imaging radiation and increased cancer risk. There could be several explanations for this. The results may be of face-value and there may be no association between medical imaging radiation and brain tumours[52] or secondary abdomino-pelvic malignancies[28]. Self-reported information and recall bias may under or overestimate radiation dose received in the study carried out by Blettner et al[52]. The number of diagnostic procedures is also a crude estimate of actual radiation exposure due to the variability in radiation dose, even for the same procedure[40,52]. Van Walraven et al[28] suggested that the relationship between radiation and cancer risk may not be linear, rather requiring a particular threshold rate at which cellular repair mechanisms are overwhelmed and start carcinogenesis[76-85].

Most studies included in the review used a linear, no-threshold (LNT) model as proposed in the BEIR VII report[15] to calculate cancer risks. The LNT model is based on atomic bomb survivors in the Japanese population (the Life Span study) and proposes that any radiation dose increases the risk of developing cancer[15,21]. Therefore, it is perhaps unsurprising that a majority of studies included in the review showed a positive association between medical imaging radiation and increased cancer risk.

The LNT model is not full proof, particularly with regard to application of data to non-Japanese populations and at low doses (< 50 mSv) of radiation where there is no convincing epidemiological evidence of a linear model[86]. The LNT model is still debated and a review by Pauwels and Bourguignon discusses these issues in detail[86].

Another limitation may involve the retrospective nature of many studies and a number of authors have commented on the practicality of following up hundreds of thousands patients for their entire lifetime[31,43]. These difficulties have also been noted in another study[87]. As described above, recall bias and under or overestimation of radiation dose received may also be another limitation in case-control studies[30,52].

With regard to the review itself, failure to identify relevant studies in the literature may have resulted in bias[14,88]. Limits were set to search for English language articles in the last 10 years only. In addition, omission of studies where the full-text could not be retrieved may have contributed to the bias[14,88,89]. As studies were not selected in an independent blinded manner, there could also have been some unjustified exclusion of eligible studies[14,88,89].

On the basis of the available literature, there is a small, but increased risk of cancer from medical radiation imaging with the risk increasing in the younger population. Many studies calculated risk using risk projection models mainly derived from atomic bomb survivors in Japan and there is a debate about the applicability of such models in low dose radiation exposure[21,90-93]. The only cohort study to date which directly assessed the risk of cancer and CT scans reported risk estimates which were broadly consistent with data from the atomic bomb survivors in the paediatric population[21]. Whether these data can be applied in the adult population is still unknown[21].

For patients with colorectal cancer who have undergone curative resection, the current guidelines are variable with ASCO[7,8] NCCN[3] and ESMO[5] guidelines recommending annual CT scans for at least three years following diagnosis and initial treatment. However, all of the published guidelines underestimate the frequency of imaging currently employed in many cancer centres since it is now recognised that resection of localized, recurrent disease in either liver, lung or peritoneum can be associated with long term disease control or cure[2]. The outcomes are also improved with long disease free interval from primary diagnosis making the case for ongoing follow up, even when the patients have reached 5 years post treatment.

Using data from the study of Berrington de González et al[41] a crude estimate of 0.013 and 0.015 lifetime excess cancers was determined for 10 CT scans to the chest, abdomen and pelvis in a 50-year-old male and female respectively. In the New Zealand context, this would equate to 39.6 excess cancers if the 1463 males and 1374 females in the 2009 New Zealand colorectal cancer registry received a conservative measure of 10 CT scans each, in the context of 1244 colorectal cancer deaths[94]. While these numbers are low and primarily include elderly patients in whom the adverse effects of radiation are reduced, ten percent of patients presenting with colorectal carcinoma are under the age of 40 years and in these patients, therapy is often most intensive and as results improve, prolonged follow up will be routine with an aggressive approach taken to treat metastatic disease. Eventually, this will require that guidelines for post-treatment surveillance address the need for more intensive surveillance strategies and make comment on extending surveillance beyond 5 years.

It may be possible to utilize non-radiation methods including magnetic resonance imaging to assess the abdomen and pelvis and contrast enhanced ultrasound for the liver although currently CT of the chest remains the gold standard for detecting pulmonary disease. A study by Schmidt et al[95] comparing the use whole body MRI in the follow up of 24 patients with colorectal cancer showed MRI was less sensitive (sensitivity 63%) at detecting lymph node metastases relative to FDG-PET-CT (sensitivity 93%) and had a similar sensitivity for detecting organ metastases (sensitivity 80% and 78% for PET-CT and MRI respectively). Despite the great soft-tissue resolution MRI provides for detection of pelvic recurrences of colorectal cancer[96-100], its use for routine surveillance of the pelvis after curative surgery was “not justified”[101] on the basis that there were no differences in detection of possible cases suitable for surgical resection compared to conventional follow up protocols, rather suggesting MRI be selectively used for imaging patients following clinical, biochemical or colonocopic assessment. The other possibility is to restrict intensive follow up to patients with adverse prognostic factors and higher risk of recurrence. However recent evidence suggests that after 3 years of survival conventional clinicopathologic factors have limited ability to predict long-term survival[102].

Whilst nothing can be definitively concluded from the crude approximations above, clinicians should be aware of the possible risks associated with ionizing radiation when imaging patients with colorectal cancer. As with any medical intervention, the clinician needs to balance the risks and benefits, particularly more so in the younger population due to the increased radiosensitivity in this group[61-74]. A clinically justified CT scan in the context of colorectal cancer is likely to be of benefit due to the fatal nature of the disease. Further studies of medical imaging risks in the adult population, based on empirical data using direct studies, and epidemiological data of radiation risks at low doses, would be beneficial in assessing the potential benefits and risks associated with multiple imaging in colorectal cancer patients.