Introduction

Polyoma BK virus (BKV) is a nonenveloped DNA virus containing a 5-kb circular double-stranded DNA.1 Primary infection occurs during childhood and is largely asymptomatic. The virus subsequently remains latent, and may undergo reactivation during immunosuppression.

Two distinct diseases due to BKV reactivation are recognized: polyomavirus-associated nephropathy in renal allografting and severe postengraftment hemorrhagic cystitis (HC) in hematopoietic SCT (HSCT).2, 3 As both diseases involve the urinary tract, the uroepithelium has been proposed to be the main site of BKV latency.

Recently, data from another closely related polyoma virus, JC virus, showed that the gastrointestinal tract may be a latency site.4 Interestingly, BKV has also been detected by PCR in the stool of about 40% of hospitalized children.5 Nonetheless, these data collected at single time points did not distinguish primary infection from reactivation of latent infection. Moreover, quantification of viral load was lacking, precluding kinetic analysis of the viral reactivation process. As a result, it remains unclear whether the gastrointestinal tract could be a latency site for BKV.

In this study, we hypothesized that if the gastrointestinal tract was a site of BKV latency, viral reactivation might occur there during immunosuppression. To test this proposition, patient populations with frequent BKV reactivation would need to be studied. Taking advantage of the high incidence of BKV reactivation after HSCT, we quantified prospectively the fecal BKV load in consecutive HSCT patients. Urine BKV load was quantified in parallel to indicate the possible concomitant uroepithelial viral reactivation.

Materials and methods

Patients and HSCT protocols

A consecutive unselected cohort of 40 patients undergoing HSCT at Queen Mary Hospital, Hong Kong, was studied. The transplantation protocol, antimicrobial prophylaxis and prevention and treatment of GVHD were also described.6 The informed consent was obtained from the patients and the investigation was approved by the institution review board in accordance with the Declaration of Helsinki.

Stool sample collection

Stool samples were collected prospectively before conditioning, on the day of HSC infusion, and weekly thereafter until day 50 or discharge. Methods of stool collection for viral studies have been reported.7 Briefly, stool was passed into a bedpan and from the center, approximately 0.5 g of stool was collected and swirled into 5 μl of viral transport medium. DNA was extracted from an aliquot of 200 μl of stool suspension (QIAamp Blood Minikit, Qiagen, Basel, Switzerland) and eluted with 200 μl of buffer.

Urinary sample collection

Serial spot urine samples (50 ml) were collected at weekly intervals as aforementioned. The sample was spun at 2000 g for 10 min. DNA was extracted from 200 μl of free urine (QIAamp Blood Minikit) and eluted with 200 μl of buffer.

BKV quantification

Quantification of BKV from fecal and urinary samples was performed by Q-PCR with the ABI Prism 7700 Sequence Detector (PE Biosystems, Foster City, CA, USA). Sequences of the Q-PCR primers and TaqMan probe (targeting the BKV VP1 gene), amplification protocols, PCR precautions and quality assurance have been described.6 All samples were tested in triplicates. Viral excretion was expressed in BKV genome copies per fecal sample (0.5 g stool) or per milliliter of urine.

Controls for PCR interference

For stool samples found negative for BKV by Q-PCR, fresh aliquots were obtained from the original stock stool suspensions. Known amounts of the BKV VP1 plasmid standard6 were spiked into them. DNA was then extracted from these samples and Q-PCR performed. In all experiments, negative stool samples spiked with control BKV VP1 plasmid showed Q-PCR results corresponding to the amount of input BKV plasmid, thereby excluding false negatives due to PCR failure arising from possible nonspecific interference in the stool (data not shown).

Types of BKV excretion

For this study, stringent criteria were used. BKV excretion, both fecal and urinary was defined as positive only if two or more samples were positive. Furthermore, we surmised that three or more consecutively positive samples (representing a continuous duration of 3 or more weeks) might be more biologically relevant. Therefore, positive samples were defined as transient when positive for <3 consecutive samples, and persistent when positive for ⩾3 consecutive samples. Persistent positive samples with significant peaking were defined as a peak BKV viral load with a ⩾103-fold increase over baseline.6, 8

Statistical analysis

Comparisons between groups of numerical data were evaluated by Kruskal–Wallis, and categoric data by χ2-tests. Risk factors for fecal BKV excretion were evaluated by binary logistic regression (SPSS, Chicago, IL, USA). The occurrence of fecal BKV reactivation was the dependent variable, with age, gender, types of HSCT (autologous vs sibling donors vs matched unrelated donors), sources of HSC (BM vs peripheral blood), conditioning regimens, occurrence of acute GVHD and HC, and the types of urinary BKV reactivation as covariates. P-values of <0.05 were considered statistically significant.

Results

Samples

A total of 283 fecal and 342 urinary samples were quantified for BKV DNA (referred herewith BKV for short, median number of samples evaluated per patient: fecal=8.0; urine=9.0).

Fecal BKV excretion

Fecal BKV excretion was detectable in 16 patients (40%) (transient, n=10; persistent, n=3; persistent with peaking, n=3) (Figure 1). Interestingly, in eight patients, fecal BKV was already detectable before marrow conditioning, although at low levels (102–103 copies/stool sample). While the fecal BKV load might fluctuate in patients with persistent excretion, three patients showed significant peaking, reaching 5.11 × 106, 4.68 × 107 and 2.75 × 108 copies/sample at 14, 14 and 21 days post-HSCT respectively, before declining to baseline levels. Fecal BKV was negative in the other 24/40 patients.

Figure 1
figure 1

Fecal BK virus (BKV) excretion in 15 patients undergoing hematopoietic SCT (HSCT). The number at each time point represented the logarithm of the BKV copies/stool sample (∼0.5 g of fecal material). For example, 3+ represented 103 copies and 5+ represented 105 copies. (–) denoted negativity for BKV. T=transiently positive (<3 consecutive positive samples; note that there might be ⩾nonconsecutive positive samples); P=persistently positive (⩾3 consecutive positive samples); PP=persistent positive with peaking (⩾3-log increase over baseline).

Full size image

Urinary BKV excretion

Urinary BKV excretion was detectable in 25 patients (65.0%) (transient, n=4; persistent, n=2; persistent with peaking, n=19).

Patterns of fecal and urinary BKV excretion

Three different patterns of fecal and urinary BKV excretion could be distinguished (Figure 2). Pattern a patients (n=3) showed concomitant and parallel increases and declines of fecal and urinary BKV load, with peaking of fecal and urinary BKV excretion coinciding temporally (fecal: 5.11 × 106 copies/sample, urinary: 5.47 × 109 copies/ml, day 14; fecal: 4.68 × 107 copies/sample, urinary: 4.67 × 108 copies/ml, day 14; fecal: 2.75 × 108 copies/sample, urinary: 1.54 × 109 copies/ml, day 21). All three patients underwent allogeneic HSCT, each having significant GVHD. The first patient, who received HSCT from HLA-identical sibling, developed postengraftment HC with gross hematuria and dysuria on day 42, and subsequent liver and skin GVHD at 6 months. The second patient, who received HSCT from matched unrelated donor, developed postengraftment HC on day 35. The third patient received HSCT from HLA-identical sibling and developed acute skin GVHD on day 26. Pattern b patients (n=16) showed significant peaking of urinary BKV excretion, at a median of 2.28 × 108 (6.33 × 104–3.09 × 109) copies/ml, occurring at a median of 24.5 (14–56) days after HSCT. However, none of these patients had significant peaking in fecal BKV excretion. Eleven patients received HSCT from HLA-identical siblings and two from matched unrelated donors. Three patients received autologous HSCT. In these patients, the types of fecal BKV excretion were persistent in two cases, transient in eight cases and negative in six cases. Pattern c patients (n=21) showed no detectable peaking in urinary BKV excretion. Eleven patients received HSCT from HLA-identical siblings and three from matched unrelated donors. Seven patients received autologous HSCT. In these patients, fecal BKV excretion was persistent in one, transient in two, and negative in 18 patients.

Figure 2
figure 2

Patterns of fecal and urinary BKV excretion. (a) Three patients developed concomitant fecal and urinary BKV peaking. (b) Sixteen patients developed urinary but not fecal BKV peaking. (c) Twenty-one patients developed neither urinary nor fecal BKV peaking. Each data point represented mean results of BKV excretion at each time point for all the patients in that group. Therefore, although some patients in patterns b and c never had quantifiable fecal BKV, the stool line did not touch baseline. Each error bar represented one s.d. BKV, BK virus.

Full size image

Correlation between fecal and urinary BKV excretion

The types (negative, transient, persistent and persistent with peaking) of urinary and fecal BKV excretion were significantly correlated (Kendall's Tau-b correlation coefficient 0.488, P=<0.01; Figure 3). It was noteworthy that all but one patient with positive fecal BKV excretion have had urinary BKV excretion. On the other hand, 10 patients with positive urinary BKV excretion, including six patients with viral peaking, showed no fecal BKV (encircled points, Figure 3).

Figure 3
figure 3

Correlation between fecal and urinary BKV excretion. The patterns were divided into transient (<3 consecutively positive samples during HSCT), persistent (⩾3 consecutively positive samples) and persistent with peaking (⩾103-fold increase from baseline). Each point represented data from an individual patient. The types of BKV excretion in fecal and urinary samples were significantly correlated (Kendall's Tau-b correlation coefficient 0.488, P<0.01). The encircled points denoted 10 patients with positive BKV viruria, but no detectable fecal BKV excretion. BKV, BK virus; HSCT, hematopoietic SCT.

Full size image

Clinicopathologic correlation of fecal BKV excretion

To examine the risk factors and clinical correlation of fecal BKV excretion, the clinicopathologic parameters of patients with or without fecal BKV excretion were compared (Table 1). Positive stool BKV excretion was significantly associated with allogeneic HSCT (P=0.037) as well as persistent and peaking of urinary BKV excretion (P<0.001). Binary logistic regression showed that BKV viruria was the only significant risk factor for fecal BKV excretion (P=0.021, odds ratio: 0.101, 95% confidence interval: 0.015–0.705).

Table 1 Correlation between clinicopathologic characteristics and fecal BKV excretion
Full size table

Discussion

Although BKV reactivation has been described in solid organ allografting, mostly in renal transplantation9 and occasionally in liver and heart/lung transplantation,10 the actual frequency of reactivation is lower than that in HSCT, because of the much more intense immunosuppression involved in the latter.3, 6, 8, 11 Therefore, HSCT represents a unique opportunity to address the issue of whether BKV might be latent in the gastrointestinal tract, by studying fecal BKV load.

This is the first prospective and quantitative analysis of fecal BKV load during HSCT. We showed positive fecal BKV excretion in 16 of 40 HSCT patients. Our protocol of fecal viral quantification has been validated for coronavirus associated with severe acute respiratory syndrome.7 A significant peaking in fecal BKV excretion was detected in 3/40 patients (pattern a cases), which showed a consistent pattern of increase, peaking and decline. It paralleled concomitant urinary BKV reactivation, with peaks coinciding on the same days. In fact, all three patients had undergone allogeneic HSCT and developed GVHD, which were typical risk factors for urinary BKV reactivation.3, 6, 8 Therefore, fecal and urinary BKV excretions probably reflected viral reactivation consequent to severe immunosuppression. Definitive proof of BKV reactivation from the gastrointestinal tract would require detection of BKV viral antigen in colonic tissues, direct demonstration of viral particles in stool by electron microscopy or detection of BKV DNA in the gastrointestinal tract by in situ hybridization in the absence of late gene expression in seropositive patients. These issues warrant further investigations.

Previous studies of isolated colonic biopsies12 or ad hoc stool specimens showed the detection of BKV in the gastrointestinal tract.5 As these studies examined specimens at single time points, the actual kinetics,13 and the significance of the presence of BKV remain obscure. Our findings offered much more rigorous results and provided kinetic data that were more meaningful and robust that isolated measurements. Indeed, the increase and decline of fecal BKV load is typical of what might be expected of in reactivation of latent viruses after organ transplantation.13

Although the proposition of the gastrointestinal tract as a BKV latency site would need further validation in patients and normal people, it may provide valuable insights into BKV epidemiology. The seroprevalence of BKV increases from about 50% in children to 70–90% in adults.14 The high BKV seroprevalence has never been adequately explained, assuming urine to be the only means of transmission. A fecal-oral route, however, offers an additional epidemiologic explanation for the high viral prevalence.

Data on urinary BKV excretion in this study confirmed once again the consistent link between allogeneic HSCT, GVHD, significant BKV viruria, and severe postengraftment HC.3, 6, 8, 15 We have previously postulated that immunosuppression allowing unchecked viral replication, in combination with the subsequent allogeneic reactivity against viral antigens upon hematopoietic engraftment (which might be heightened by GVHD) leads to severe HC. In this study, however, we were unable to observe any apparent association between fecal BKV excretion and gut GVHD. Nevertheless, this study contained a relatively small number of patients, so that the potential relationship between fecal BKV excretion and gut GVHD would need further examination in larger number of HSCT recipients.

Our observations provide two positive leads for future studies. Firstly, the factors predisposing to, and the biologic and clinical significance of, BKV latency in the gastrointestinal tract should be evaluated. In particular, recent study has identified BKV DNA sequence from colorectal cancers, suggesting that this virus may have a function in the pathogenesis of these diseases.12 Secondly, 8 out of 40 patients (20%) demonstrated significant fecal BKV excretion before BM conditioning. This was lower than that of 40% reported from stool and rectal swab samples in hospitalized children, based on the detection of the BKV T-antigen gene.5 Whether this reflects an age-related difference in BKV fecal excretion and hence differences in viral epidemiology in health and disease or a technical difference in BKV DNA detection remains to be determined.