Serology: Serology is useful to determine previous viral exposure and identify patients at risk, as only seropositive patients (positive IgG antibodies in blood) may develop CMV disease[30,31].

Antigenemia assay: It detects viral protein pp65 produced in peripheral blood polymorphonuclear leukocytes. It is a simple and rapid technique, with a sensitivity of 60%-100% and a specificity of 83%-100%[32,33]. However, it does not differentiate between latent infection and active disease[23], no association exists with virus reactivation in the intestinal mucosa[31,34], and false negatives can occur in neutropenic patients[35].

PCR DNA amplification assay in blood: This test is supplanting antigenemia. Sensitivity ranges 65%-100% and specificity ranges 40%-92%[32,36,37]. This diagnostic method cannot differentiate between latency and activity states, so it is necessary to determine a cut-off above which active infection is diagnosed in patients with IBD[30,35], as well as in solid organ transplantation, where a viremia > 1000 copies/100 000 leukocytes indicates symptomatic CMV infection[38]. This method can be used both to detect disease and to monitor treatment response; a slow or absent decline in DNA levels after treatment could be an early indicator of drug resistance[39], and positive result after therapy indicates continuing treatment[40]. Most studies in patients with IBD have reported a correlation between identification of CMV by PCR in blood and detection colonic CMV by haematoxylin and eosin (HE) or immunohistochemistry (IHC)[34,41-43].

PCR assay in stool: A qualitative and quantitative PCR assay for CMV DNA has been performed on human faecal specimens from immunocompromised patients[44,45].

Histological diagnosis is considered the “gold standard” for diagnosing CMV disease in the gastrointestinal tract[46-51]. Several methods are available.

Histology by HE: Typically reveals cytomegalic cells that are 2 or 4 fold larger than normal cells, containing basophilic intranuclear inclusions in eccentric location surrounded by a clear halo, giving it an “owl’s eye” appearence. Cells show a thickened nuclear membrane and smaller granular intracytoplasmic inclusions. Colonic biopsies should be taken from inflamed mucosa near or within the ulcer. This method has very high specificity (92%-100%), but its main problem is a poor sensitivity (10%-87%), making the diagnosis requires many samples and a trained pathologist[46,47]. Anecdotally cytomegalic cells have been described in colon biopsies from normal mucosa in healthy individuals[48,49].

IHC in colon biopsies: It involves the identification of CMV antigens in infected cells. It has a higher sensitivity than HE (78%-93%)[50,51]. Other techniques, however, do not rely on histology.

PCR DNA amplification assay in colon mucosa: PCR DNA amplification assay in colon mucosa has the greatest accuracy for virus detection[29,52,53] and may be used as a qualitative or quantitative method. PCR DNA levels in the colon are not related to viremia levels measured in blood[31,43,54,55]. Some studies detected prevalences greater than 30% in IBD patients[56], although the significance of a positive result in absence of histological signs of CMV infection is unclear. Very few studies have shown a correlation between histology (HE/IHC) and PCR results[31,43]. This suggests that detection of low DNA levels could determine latent infection and requires a cut-off level of viremia to distinguish infection from disease[30]. In other words, a positive result in colon does not necessarily reflect the involvement of CMV in UC flare-ups. A recent study from France suggests a cut-off of > 250 copies/mg for CMV disease[54] with a sensitivity of 100% and a specificity of 66%. To avoid false positives, and awaiting further studies, this determination should be performed only in patients with active UC refractory to conventional treatment.

Viral culture: Viral culture was previously regarded as the gold standard in CMV detection. Culture has a sensitivity of 45%-78% and a very high specificity (89%-100%). The virus is placed in a fibroblast tissue culture and diagnosis is made once the virus causes cytopathic changes in fibroblasts. The problem is that the result takes days to weeks[10], so this method is not used in clinical practice[57].

Typical endoscopic findings of CMV colitis: Typical endoscopic findings of CMV colitis are microerosions, deep ulcers and pseudotumoral lesions[58]. Most studies in patients with IBD, specifically in active UC, have not found specific endoscopic features[43,54,59,60]. Anecdotal studies describe some characteristic endoscopic (large, irregular, punched-out or longitudinal ulcers) with virus[61]. Discrepancy can be explained by the different criteria used to define CMV infection or disease.

In conclusion, different methods of detection of CMV have different sensitivities and specificities. In the setting of a severe UC flare-up refractory to conventional treatment, colonic IHC or PCR in blood should be performed, PCR presents a good correlation with the results of HE and IHC[30]. Serology and antigenemia only are useful as negative predictors for CMV infection, since they do not correlate with actual CMV colitis. The role of PCR positivity in colon remains to be determined, it is not known how to interpret its positivity in the absence of histological changes. However, recent studies imply positivity of this method in the disease prognosis[54]. Current European guidelines[62] recommend the use of tissue PCR or IHC to detect CMV in patients with UC resistant to immunomodulators, whereas older American guidelines[63] recommend performing sigmoidoscopic biopsy and viral culture in refractory cases of UC.

Seropositivity in CD patients is as high as in other populations (70%)[64,65], however CMV disease is rare in CD, making the virus an unlikely etiological factor in the novo development of IBD[27,64,66,67].

In most surveys, IHC did not detect CMV in CD patients, and PCR in tissue or stool samples found a very small frequency (< 5% of patients)[27,66-69], except in one study that used a highly sensitive PCR (detection < 10 copies of DNA) which detected CMV in 66% of individuals with CD and in 29% of controls, showing no association between CMV DNA and disease activity, suggesting the authors that small quantities of viral DNA are not clinically relevant[29].

Recently some theories have been published that explain these findings based on small studies[70], so results should be interpreted with caution. Tumor necrosis factor-α (TNFα) would be significantly associated with CMV infection or reactivation in IBD. However interferon-γ (IFNγ), which is produced from CD4 + T cells[71], could suppress CMV reactivation. CD is considered a Th1-type inflammatory process with high expression of IFNγ, which does no help the virus reactivation and may explain the different prevalence of CMV disease in UC and CD[70].

Prevalence of CMV infection is about 70%[31], a similar percentage to that of the general population.

Patients with inactive or mild-moderate UC did not show an increased risk of CMV colitis, using HE and IHC[27,31,41] or IHC in colectomy patients undergoing surgery for dysplasia or cancer[72].

The most extensive literature is on severe and/or steroid-refractory colitis. These two terms are used interchangeably in studies and are not defined clearly, so the results are difficult to interpret[30].

Severe colitis: The combination of serological tests and rectal biopsies found a CMV disease prevalence of around 20%[42,73]. Risk factors that have been identified are female gender, older age, pancolonic disease with active inflammation at histology and azathioprine therapy[60,72,73]. According only to antigenemia, prevalence increased to 34%[60], while using HE or IHC in colonic mucosa prevalence decreased to 3%[74].

Severe steroid-resistant colitis: In retrospective studies the prevalence according to HE was 0.5%[75]. The prevalence of CMV increased by combining HE and IHC with antigenemia (20%-40%)[31,41,72,76,77]. The prevalence using blood PCR was 60%[43], whereas PCR results in colon were lower (38%)[78]. There is a poor correlation between peripheral and colonic viral load, this can be explained by theory proposed by Criscuoli et al[24,42] about a specific genotype, which possibly has a particular colonic tropism and pathogenic character, whilst other genotypes are not pathogenic.

Urgent colectomy for colitis: A higher prevalence of CMV is expected in these patients because the greater severity of their disease, but the prevalence using HE or IHC ranged between 11.5% and 27%[77,79-82], which is similar to the previous groups.

Experimental studies have identified three factors that influence the reactivation of CMV infection in active colitis: (1) increased cell proliferation in inflamed tissue with ulcers that attract CMV; (2) inherent impaired natural killer cell activity presented by patients with IBD[83,84]; and (3) use of immunosuppressive drugs[31,42,69]. The third factor is controversial[73-75,80,85], since use of steroids may be either a risk factor or a surrogate marker of severity disease[41]. In vitro data suggest that steroids and cyclosporine could support the replication of CMV[76,86,87]. Infliximab treatment has not been associated with increased risk of CMV reactivation in IBD patients[66]. Moreover, TNFα has been identified as an activator of CMV, and thus infliximab (IFX) could promote viral latency, by lowering the levels of this cytokine[88,89]. This suggests that IFX could be used in patients with severe UC flare and CMV reactivation. However, some cases of tacrolimus-refractory UC that were treated with IFX did not have a favourable outcome[90].

CMV reaches the mucosa by way of monocytes and then colonize the colonic cells, acquiring particular affinity for the inflammatory sites, probably due to the presence of pro-inflammatory cytokines such as IFNγ and TNFα produced by macrophages and T-cells in active UC[91-94]. CMV avoids immune recognition of macrophages and uses immune specific functions to reactivate and spread[95]. Recent studies suggested CMV can appear only in inflamed tissue and is not found in healthy tissue[54].

Hommes et al[94] proposed the following sequence to explain the pathophysiology of CMV disease in patients with active UC: (1) initiation phase, mucosal inflammatory response induces expression of cytokines and chemokines which activates latently infected cells and the migration of monocytes and dendritic cells into the inflamed mucosa; (2) reactivation phase, in which infected monocytes differentiate into tissue macrophages and dendritic cells; and (3) consolidation phase, during which the virus causes an active replication predominantly in endothelial cells that likely exacerbates inflammation. The virus in the first and second phase has not led to histological damage and can only be detected by PCR-DNA. It would be appropriate to detect CMV in these phases, so the optimal treatment would be administered at an early stage before complicating the disease. They do not consider CMV as an innocent bystander but rather involved in resistance to immunosuppressive treatment.

In the same line, other authors suggest that the positivity of CMV DNA in the colonic mucosa in patients with refractory UC indicates uncontrolled intestinal inflammation, suggesting a change in immunosuppressive therapy[96].

Coincidental detection of primary CMV infection at the first appearance of IBD has been reported before[66,97], primary CMV infection in IBD patients without immunosupression[98,99], and even disseminated CMV infection in CD[100].