Background.A retrospective study of the clinical, epidemiologic, and virologic features of norovirus gastroenteritis in 12 adult allogeneic hematopoietic stem cell transplant (HSCT) recipients.
Methods.Norovirus infection was diagnosed by reverse-transcriptase polymerase chain reaction. Strains were genotyped by nucleic acid sequence of the most highly conserved region of the norovirus gene encoding the capsid S (shell) domain.
Results.Ten of 12 patients presented with vomiting of short duration, but diarrhea was present in all. The median time from onset to norovirus diagnosis was 1 month (range, 0.25–6.0 months). Eleven patients were receiving immunosuppression when norovirus infection was diagnosed: 8 for graft-versus-host disease (GVHD) in an organ other than gut, 1 for previous gut GVHD, and 2 for presumed gut GVHD that proved to be norovirus gastroenteritis. Six patients required enteral or parenteral nutrition for severe weight loss. In 10 patients, diarrhea lasted a median of 3 months (range, 0.5–14 months) and virus was shed at a high level throughout. The remaining 2 patients died after 4 months of diarrhea (one died of unrelated complications, and the other died of malnutrition). The noroviruses found were GII (untyped), GII-3, GII-4, and GII-7 in 1, 1, 9, and 1 patients, respectively. Eleven of the 12 patients had acquired their infection in the community. Phylogenetic analysis of the GII-4 strains demonstrated that all differed.
Conclusions.Noroviruses are a hitherto unsuspected cause of prolonged morbidity and mortality in adults after allogeneic HSCT. The use of reverse-transcriptase polymerase chain reaction to detect high viral load levels in feces distinguishes norovirus gastroenteritis from gut GVHD.
Published data suggest that diarrhea complicates 79% of allogeneic hematopoietic stem cell transplantations (HSCTs) . It is often secondary to conditioning therapy, other drugs, or graft-versus-host disease (GVHD). However, in a proportion of patients, it has an infectious origin, and viral gastroenteritis—most commonly due to enteric adenoviruses and rotaviruses—has been reported [1-5]. Remarkably, noroviruses, the most common cause of nonbacterial gastroenteritis worldwide, have not so far been identified as a significant cause of diarrhea in adults after allogeneic HSCT.
Noroviruses are a genus of the Caliciviridae family. Three norovirus genogroups (I, II, and IV) infect humans, and these are subdivided into ∼30 genotypes; variants can be identified within the genotypes  since genetic variability rapidly evolves because of the accumulation of point mutations and frequent recombination among different viruses. In immunocompetent individuals, vomiting and diarrhea due tonorovirus infection usually last only a few days, although asymptomatic shedding of virus in feces may continue for up to 3 weeks . However, in immunocompromised individuals after intestinal transplantation or chemotherapy, there are only a few reports, all involving children [8-10] (apart from 1 in an adult heart transplant recipient ), of norovirus infection causing a protracted diarrheal illness with prolonged viral shedding in feces. There is also a single report of prolonged gastroenteritis in a child after allogeneic HSCT . The present retrospective analysis describes the clinical, epidemiologic, and virologic features of norovirus gastroenteritis in 12 adult allogeneic HSCT patients.
Patients and Methods
Clinical setting. The Department of Haematology at University College Hospital in London, England, performs ∼50 allogeneic HSCTs per annum on adult patients. Of these procedures, ∼60% involve reduced-intensity conditioning.
Patients. Twelve allogeneic HSCT patients with norovirus gastroenteritis (Table 1) had each received 1 of 3 categories of conditioning before transplantation : 2 regimens included T cell depletion—that is, reduced intensity (fludarabine, melphalan, and in vivo alemtuzumab) or full intensity (total body irradiation, fludarabine, cyclophosphamide, and ex vivo alemtuzumab); the third regimen did not involve T cell depletion and consisted of a full-intensity regimen with total body irradiation and cyclophosphamide or etoposide. In the hospital, the patients were treated in single rooms with en suite facilities. All underwent regular long-term follow-up at University College Hospital.
Microbiological screening of fecal samples. At the onset of diarrhea, stool specimens were tested using standard diagnostic laboratory techniques for Salmonella, Shigella , and Campylobacter species; Escherichia coli 0157; Clostridium difficile; and ova, cysts, and parasites . Until June 2006, fecal samples were also screened by electron microscopy for viruses causing gastroenteritis, such as adenovirus 40/41, astrovirus, norovirus, and rotavirus. In July 2006, this was replaced by reverse-transcriptase polymerase chain reaction (RT-PCR) for norovirus RNA. For detection of the other viruses causing gastroenteritis, either electron microscopy or adenovirus 40/41 and rotavirus enzyme immunoassays were used.
Norovirus RT-PCR. RNA was extracted from fecal samples, and complementary DNA was produced using random primers . Samples from patient 1 were tested by nested RT-PCR , but those from patients 2–12 were tested by real-time RT-PCR  and the cycle threshold (Ct) number recorded; this number is inversely proportional to the amount of viral RNA. Thus, viral load levels of fecal norovirus were defined as high if viral RNA was detected in the first round of the nested RT-PCR or if the Ct was <30 by real-time RT-PCR. Conversely, the viral load was defined as low if viral RNA was detectable only in the second round of nested RT-PCR or when the Ct was ⩾30 by real-time RT-PCR.
Norovirus genotyping and analysis of viral variants. Consensus primers based on the nucleic acid sequence of the most highly conserved region of the norovirus gene encoding the capsid S (shell) domain were used in RT-PCR to produce amplicons whose nucleic acid sequence was analyzed to identify genotypes and variants within a genotype . Phylogenetic analysis of the sequences was performed using a neighbor-joining method (Bionumerics, version 5.1; Applied Maths).
Clinical characteristics. In November 2004, soon after HSCT patient 1 presented with severe diarrhea, which was presumed to be acute GVHD, although it was refractory to prolonged steroid therapy. When norovirus infection was diagnosed a month later by electron microscopy, immunosuppression was stopped but symptoms had not resolved at the time of the patient's death because of complications unrelated to gastroenteritis.
Patient 2, who was at relatively low risk of acute GVHD, had profuse watery diarrhea that started with nausea and vomiting in April 2006 1 month after HSCT. Gut GVHD was diagnosed, but the diarrhea did not diminish despite prolonged steroid treatment; there was severe weight loss of ∼14 kg (a quarter of the patient's body weight), and the patient required nasogastric feeding and total parenteral nutrition. In July 2006, norovirus infection was diagnosed by electron microscopy, immunosuppression was withdrawn, and gastrointestinal symptoms slowly resolved.
Patient 3 developed watery diarrhea that immediately responded to steroid treatment, which was consistent with a diagnosis of gut GVHD. As the immunosuppression was being reduced in July 2006, he presented with renewed diarrhea accompanied by vomiting and the steroid dose was increased. However, the vomiting raised the possibility of norovirus infection, and virologic testing of feces was initiated. Because norovirus was not detected by electron microscopy, although clinical suspicion was high, an RT-PCR test was requested from the Health Protection Agency's reference laboratory and the diagnosis of norovirus gastroenteritis was confirmed. The patient's immunosuppression was then reduced and the gastroenteritis resolved rapidly.
At this point (July 2006), norovirus infection was of sufficient clinical concern for RT-PCR to be introduced as the routine screen for its detection in all HSCT patients with diarrhea. During the next year, an additional 9 allogeneic HSCT recipients were identified as having norovirus gastroenteritis. Concurrent infections were excluded as described above in Patients and Methods, and stool samples were not bloody and lacked mucus. Ten of the 12 patients had also been tested for C. difficile toxin with negative results; the 2 exceptions were patients 5 and 12. For patient 7, whose diarrhea lasted for >1 year, an extended pathogen screen , including adenovirus 40/41, astrovirus, Campylobacter species, cryptosporidium,E. coli, Giardia species, rotavirus, Salmonella species, and sapovirus, was applied to 11 of the fecal samples obtained during gastroenteritis, but all test results were negative.
Table 1gives the clinical features of all 12 patients with norovirus gastroenteritis. The median time after transplantation to the development of symptoms was 10.5 months (range, 0.25–96 months). Ten of the 12 patients presented with transient nausea and vomiting. The median time from onset to diagnosis of norovirus infection was 1 month (range, 0.25–6.0 months). In 10 patients, the gastroenteritis resolved, although diarrhea was prolonged, lasting a median of 3 months (range, 0.5–14 months), and in patient 7 symptoms recurred 13 months later. The remaining 2 patients died while still symptomatic (one died of unrelated complications, and the other died of malnutrition because of unresolved norovirus gastroenteritis).
Treatment of norovirus gastroenteritis. As indicated in Table 1, 6 of the 12 patients had a protracted diarrheal illness (median, 7 months; range, 3–14 months) and required enteral or total parenteral nutrition because of malnutrition and severe weight loss. This involved lengthy hospitalizations, and the median number of inpatient days per patient, incurred during their gastroenteritis, was 73 (range, 26–108).
Eleven of the 12 patients were receiving immunosuppression (including 8 taking oral or parenteral steroids) at the time of diagnosis of norovirus infection (Table 1); 8 patients were being treated for GVHD in an organ other than gut, 1 was being treated for previous gut GVHD, and 2 patients were being treated for presumed gut GVHD that was in fact norovirus gastroenteritis. Immunosuppression was stopped or reduced in 8 patients (Figure 1) after diagnosis of norovirus infection.
Epidemiologic and virologic features of norovirusgastroenteritis. The first patient was identified at the end of 2004 and the remaining 11 after June 2006 (Table 2), coinciding with the introduction of norovirus RT-PCR. In July and August 2006, 5 patients received diagnoses of norovirus gastroenteritis, and a nosocomial outbreak was suspected; however, reference to the patients' hospital records showed that, in 3 patients, the diarrhea had started several months previously and that, in all 5 patients, the initial presentation had been from home. In addition, screening of fecal samples from all other hematology inpatients in August 2006, regardless of whether they had gastrointestinal symptoms, did not reveal any unsuspected cases of norovirus gastroenteritis, and no asymptomatic individuals were found to be shedding virus (data not shown).
Eleven of the 12 patients had community-acquired norovirus gastroenteritis, and 4 patients reported a family outbreak (family members were not tested). Another patient (a nurse) developed symptoms after returning to work on a ward involved in a diarrheal outbreak (organism responsible not identified). Several different genotypes were found among the patients (Table 2). The most common was norovirus GII genotype 4 (GII-4), but the variants differed, being 3, 4, 6, and 8. In most cases the same variant was found in a particular patient no matter how many times fecal samples were tested. However, in patient 2 a different variant was detected transiently at a low level after cessation of diarrhea, and in patient 7, the variant found in the first sample was thereafter replaced by a different variant. Phylogenetic analysis (Figure 2) showed that all of the 11 GII-4 noroviruses from 9 of the patients differed from each other (ie, none of the infections had a common source).
To confirm infection with norovirus and assess persistent carriage or clearance of the virus, serial fecal specimens were tested by RT-PCR; in each patient in whom multiple samples had been taken, high-level virus RNA was repeatedly detected with ongoing diarrhea (Table 2and Figure 1). Although the diarrhea settled in the 10 patients who survived the episode, in only 4 patients (Figure 1) were samples sent after recovery. In 3 patients, there was evidence of virus clearance at 4, 6, and 14 months after onset of illness. In the fourth patient (patient 7), low-level viral loads (Ct, 39) were found in 2 stool samples in the absence of symptoms, but gastroenteritis recurred 27 months after first onset; this coincided with increased immunosuppression because of pulmonary GVHD and the original norovirus strain was again detected in stool samples at a high level (Ct, 23).
Noroviruses are well established as a cause of gastroenteritis and now, because of the adoption of sensitive molecular techniques , are emerging as a significant cause of sporadic cases. This first description of adult HSCT recipients with norovirus gastroenteritis is a direct result of our use of RT-PCR for virus detection; before this, electron microscopy was used to screen feces, but this rather insensitive method is not suited to detect sporadic cases as opposed to outbreaks of infection. This difference in sensitivity between electron microscopy and RT-PCR is well illustrated in the present report in which at diagnosis norovirus was found by electron microscopy in only 2 of the 9 patients tested, whereas RT-PCR found virus in every instance, confirming previous findings .
Although the much greater sensitivity of RT-PCR is important for the reliable diagnosis of norovirus gastroenteritis, its use has led to the realization that norovirus can be shed in the feces in both symptomatic and asymptomatic individuals. Thus, although detection of norovirus by electron microscopy in a patient with gastroenteritis confirms the diagnosis , detection by RT-PCR does not necessarily prove that it is the cause of the illness. The site of primary replication of norovirus has not been established, but replication in the small intestine can be assumed in view of histopathologic studies showing blunting of the villi . Indeed, symptoms of vomiting and diarrhea correlate with high viral load levels of norovirus in the feces of pediatric patients with cancer, whereas asymptomatic shedding is associated with lower levels . For the present study, Ct values were used as a semiquantitative measure of the amount of norovirus in feces because a viral load cutoff has been established for the RT-PCR used by us in the reference laboratory ; a Ct <30 indicates a high viral load and correlates with disease, whereas a Ct ⩾30 indicates a low viral load and is consistent with asymptomatic shedding. In the 12 HSCT recipients described herein, norovirus was shed at a high level in 11 while symptomatic, confirming its etiologic role in the gastrointestinal illness. The remaining patient had resolving gut GVHD when he presented with vomiting, and a low viral load of norovirus was detected in feces. Although this finding is consistent with asymptomatic carriage, norovirus was the only enteric pathogen detected, and it is probable that the virus caused the vomiting and diarrhea but was decreasing as symptoms were rapidly resolving.
It is now clear from our experience that norovirus infection can cause severe disease with significant morbidity and mortality after allogeneic HSCT, both before and after engraftment. Instead of recovery after a few days—as expected in immunocompetent patients—many of our patients had diarrhea for months. It was not possible to identify predisposing factors for protracted diarrhea, but it is noteworthy that 11 of the 12 affected patients were receiving systemic immunosuppression, with 9 having GVHD at norovirus diagnosis. In addition, 9 had received alemtuzumab in their conditioning regimen and were therefore more prone to infection due to T and B lymphocyte depletion. Of the 3 patients whose diarrhea lasted <2 months, 1 was the only patient not receiving immunosuppressive drugs at the start of his illness, and the other 2 had received grafts without T cell depletion. Of the remaining 9 patients with more prolonged diarrhea, 6 patients required invasive feeding (hospital based with lengthy stays) for associated weight loss and malnutrition, but despite this, 1 patientdied as a direct consequence of the infection.
The norovirus strains found in the patients were genotyped to exclude nosocomial transmission. Fourteen different strains of noroviruses were identified, 11 being different variants of GII-4, the most common strain encountered worldwide . All but 1 of the patients acquired their infection in the community, and the strains acquired followed the pattern of variants causing outbreaks at the time. In 2004, the predominant strain of GII-4 in England and Wales was variant 3 , and this was found in patient 1 in that year. Variant 4 emerged late in 2005  and was found in patients 2, 5, and 7–9 in 2006, variant 6 first appeared in 2006  and was detected in patients 2, 10, and 11 in 2007, and variant 8 emerged in 2006 (unpublished data) and was found that same year in patient 7.
Norovirus infection outbreaks commonly occur both in the community and in institutional settings, such as hospitals, where transmission is usually from person to person. It is therefore remarkable that, although several of our patients had lengthy stays in the hospital because of the debilitating effects of their as yet undiagnosed norovirus infection, there was no evidence of an associated hospital outbreak. The explanation lies in the use of single rooms with en suite toilet facilities together with adherence to a high standard of hygiene by medical and nursing staff. In addition, almost all patients presented from the community with diarrhea but after the early phase of nausea and vomiting had passed. This means that the point in the illness with the greatest potential for transmission  had passed because projectile vomiting in norovirus gastroenteritis may be the key to the establishment of episodes through airborne transmission and widespread contamination of environmental surfaces.
In allogeneic HSCT recipients, diarrhea is a common symptom often ascribed to GVHD, an immunologic disorder that results from donor lymphocytes reacting against host tissues. The manifestations of acute gut GVHD usually include protracted nausea and profuse watery diarrhea, whereas vomiting is uncommon. Severe abdominal pain, bloody diarrhea, and massive enteral fluid loss accompany advanced disease. The differential diagnosis is usually between GVHD and infection; norovirus should be suspected in the absence of other enteric pathogens. Norovirus gastroenteritis is characteristically accompanied by sudden onset of projectile vomiting. In our norovirus cohort, vomiting was a prominent but transient presenting symptom. Other factors that point to a diagnosis of norovirus infection included a family outbreak of gastroenteritis infection coinciding with the start of the patient's illness and apparently steroid-resistant gut GVHD. In addition, noroviruses cause secretory nonbloody diarrhea, whereas fecal mucus, blood, and leukocytes are sometimes encountered in gut GVHD. Finally, although cytomegalovirus colitis was excluded as a cause of diarrhea in our patients on the basis of endoscopy and biopsy results (data not shown), differentiation between norovirus gastroenteritis and gut GVHD was not possible because both are associated with the relatively nonspecific features of increased enterocyte apoptosis and inflammation. The characteristic histologic feature of GVHD, namely, crypt apoptosis, is also sometimes seen in norovirus infection , although it should be limited to the small intestine as opposed to gut GVHD, which usually affects both the small and large bowel.
A major clinical concern is that misdiagnosis of norovirus gastroenteritis as GVHD will lead to inappropriate augmented immunosuppression with steroids, cyclosporin, or mycophenolate mofetil and, in some severe cases, antithymocyte globulin. There is no literature available on the likely detrimental impact of these treatments on allogeneic HSCT patients with norovirus gastroenteritis. However, after intestinal transplantation , it has been shown that increased immunosuppression (for suspected graft rejection) in the presence of norovirus infection can significantly worsen symptoms and adversely affect outcome; conversely, a reduction in immunosuppression may be critical in resolving infection. Among our allogeneic HSCT recipients, 8 of 12 had immunosuppressive therapy decreased or stopped, but diarrhea persisted for several months after this change in some cases, although there was often a reduction in volume and frequency of stool.
In summary, norovirus infection can cause life-threatening gastroenteritis after HSCT and should always be considered in the differential diagnosis of gut GVHD so that a sensitive, specific test for norovirus infection can be requested. This allows early diagnosis in these vulnerable patients and prevents escalation of immunosuppression for wrongly suspected GVHD. Where norovirus infection is diagnosed, the appropriate supportive care must be given and meticulous infection control precautions taken to avoid a hospital outbreak.
Financial support. K.N.W. acknowledges funding from the University College London Hospitals/University College London Comprehensive Biomedical Research Centre.
Potential conflicts of interest. All authors: no conflicts.
- polymerase chain reaction
- hematopoietic stem cell transplantation
- graft-versus-host disease
- base sequence
- rna-directed dna polymerase
- viral load result
- therapeutic immunosuppression
- allogeneic hematopoietic stem cell transplant
- norwalk viral gastroenteritis
- phylogenetic analysis
- symptom onset
- norovirus infections