Background
BK virus (BKV) is a ubiquitous polyoma virus that causes a clinically important pattern of renal allograft injury in approximately 5 % of renal transplant recipients (RTR). Although viral replication can occur in up to 60 % of recipients [
1], in its most severe form, invasive disease leads to tubulointerstitial nephritis and allograft dysfunction with a significant risk of allograft failure in 15–50 % of affected individuals. Recognised risk factors for BK virus nephropathy (BKVN) include the intensity of immunosuppression, treatment of acute rejection, male gender, ureteral trauma, and diabetes mellitus [
2]. Strategies involving early screening and pre-emptive management (mainly reduction in immunosuppression) can prevent long term graft dysfunction due to invasive disease [
2]. Once BKVN is established, treatment options apart from reduction in immunosuppression are unproven and largely untested by clinical trials. Measures for the early identification of risk in order to prevent progression of viral replication and progressive infection are therefore preferred.
Screening for BK virus can be performed by examination for urinary decoy cells, Haufen on electron microscopy of the urine, or more commonly by urine or plasma detection of virus by molecular methods. Recommendations on the suggested frequency of screening are largely opinion based but in general suggest 1–3 monthly in the first 6 months post transplantation followed by 3 monthly testing until 1 to 5 years post transplantation [
2‐
4]. Definitive evidence of BKVN requires histological evidence of viral injury and exclusion of other causes of graft dysfunction by renal biopsy. Typically, extensive tubulointerstitial nephritis with positive immunohistochemical staining of the SV40 large T antigen is noted, however disease can be focal and missed by renal biopsy if sample size is inadequate [
2]. Some studies suggest that BKVN can be presumptively diagnosed when the plasma viral load exceeds 10
4 copies/mL [
2,
4]. However in the absence of a quantitative international standard for BK virus assays, such cut-off values for individual assays need to be validated at an institutional level.
In Western Australia, the two renal transplant services independently introduced different screening strategies such that the renal transplant centre at Royal Perth Hospital performed fortnightly BKV screening of both urine and plasma for the first 6 months post transplantation, whilst Sir Charles Gairdner Hospital tested 3-monthly urine and plasma for the first 6 months. The introduction and adherence to these strategies varied over time within both centres. This “quasi-randomisation” by hospital allows unique comparison of the performance of different compartments and frequencies utilised in screening for BKVN.
Methods
We performed a retrospective cohort study of the 277 consecutive adult renal transplants performed over the 5 year period between January 2008 and December 2012 at Royal Perth and Sir Charles Gairdner Hospitals who had at least 12 months clinical follow up. We extracted laboratory data from the electronic pathology results and patient clinical records and recorded the histological diagnoses and scores assigned to renal biopsies (performed for protocol or investigation of renal dysfunction), specifically relating to the coding of acute rejection and BKVN. We recorded all BKV plasma and urine viral load results and screening frequency then determined the timing and level of the first positive and the peak BK viral load for urine and plasma. Data were evaluated to 31 December 2013.
In Western Australia patients receive Basiliximab induction and routine triple immunosuppression with Tacrolimus, Mycophenolate sodium or mofetil and oral Prednisolone. Target levels are usually 5–10 mcg/L for Tacrolimus for the first 3 months and Prednisolone is tapered to 5–10 mg daily by 3 months.
All renal histology was performed by histopathologists trained in renal histology and reported for the purposes of clinical management only. Histology diagnosed as acute rejection (borderline, grade I-III), acute humoral rejection or BKVN were coded dichotomously as biopsy-proven acute rejection (BPAR) rejection (yes/no) and BKVN (yes/no). All BKVN cases demonstrated positive immunostaining against SV40. Rejection episodes diagnosed by biopsy were managed with an increment in steroids as intravenous pulses, and for unresponsive cases additional treatment with anti-T cell therapies were commenced.
The four value MDRD study equation for creatinine values standardised to creatinine reference materials was used to calculate eGFR in mL/min per 1.73 m2 = 175 × SCr-1.154× Age0.203 × (0.742 if female) × (1.21 if black). We did not apply the correction factor for “black” race to Aboriginal and Torres Strait Islanders as there is insufficient evidence it applies to this specific indigenous group.
Molecular methods
Samples were EDTA plasma or sterile urine. Extraction, polymerase chain reaction (PCR) and detection was performed as previously described with some modifications [
5]. Briefly, 200 μL of sample was extracted with the MagNaPure LC instrument using the Total Nucleic Acid Isolation Kit. A 50 μL volume of elution buffer was used. A total of 5 μL of eluted nucleic acid was added to 15 μL of PCR mix comprising 2 μL FastStart 10X Reaction Mix Hybridisation probes, 2 μL MgCl
2 at 25 μmol/L, 1 μL each of forward and reverse primer at 20 μmol/L, 0.2 μL BKV probe at 20 μmol/L and 8.8 μL of water. PCR Amplification of the BK virus capsid protein-1 (VP-1) gene was performed using the LightCycler 1.5 instrument with 45 cycles amplification at 95°C 10 s, 60°C 20 s, 72°C 15 s. Oligonucleotide sequences used for the BKV forward primer were 5’ GCA GCT CCC AAA AAG CCA AA 3’, for the BK reverse primer were 5’ CTG GGT TTA GGA AGC ATT CTA 3’, and for the BKV probe were 5’ 6FAM--ACC CGT GCA AGT GCC AAA ACT AC--BHQ1 3’. An entire genome of Dunlop strain BKV in a pBR322 plasmid served as a quantitative standard of 1 × 10
6 copies/reaction with a crossing point of 22 cycles. Positive and negative controls were utilised for each PCR run. The lower and upper limits of quantification are 2.7 log
10 and 9.87 log
10 copies/mL.
Statistical methods
Categorical variables were compared with Fisher’s exact test, and continuous variables by Mann Whitney U test (of log
10 results in the case of viral load measurements). Urine and plasma viral load association were determined by simple linear regression and determination of Pearson’s correlation coefficient after log-transformation. Receiver operator characteristic analysis was performed according to DeLong [
6]. MedCalc version 12.4.0.0 was used for statistical analysis. Statistical significance was defined by p-value <0.05 with two-tailed tests.
Ethical considerations
The study was approved by the Human Research Ethics Committee of Royal Perth Hospital and was conducted according to the principles of the National Statement on Ethical Conduct in Human Research [
7].
Discussion
We examined and compared the frequency and compartment of BK viral load testing in relation to BKVN diagnosed by renal histopathology and defined potential receiver operator characteristic derived cut-offs for diagnostic accuracy. In our subset of 53 cases where fortnightly testing of both urine and plasma were performed, BKV was detected 29 days sooner in urine than plasma. Therefore monitoring of urine allows the earliest identification of viral replication, in agreement with other studies [
8,
9]. Subgroup analysis of 14 cases by Funk et al. found BKV in the plasma became detectable 50 days after the urine compartment, however the frequency of testing and time from transplant were not recorded.
8 Babel et al. recorded viremia 6 weeks after viruria from 233 patients tested monthly for BKV in blood and urine [
9].
It was anticipated that more frequent screening (fortnightly versus 3-monthly) would detect BKV in the urine earlier and at a lower viral load as shown in our study, however we demonstrate that there was no statistical difference in the time to first detection or the initial viral load for the plasma compartment between the two screening frequencies. We reason that this discrepancy relates to BKV appearing in the plasma much later than in urine such that less advantage is found by more frequent plasma screening early post transplantation. We could not detect a difference in BKVN prevalence between these two screening frequencies although this may be due to insufficient sample size. We cannot explain the low prevalence of BKVN demonstrated for those transplants in whom no virological screening was performed. Although renal biopsy was performed for a similar percentage of these transplants, it is possible there was less active case-finding associated with unmeasured variables in this group. We do not advocate cessation of BK viral load screening on account of this finding of uncertain significance.
We confirmed that our in-house BKV quantitative assay is able to predict the risk of histologically proven BKVN and show that the ROC analysis identifies an optimal viral load cut-off for BKVN of 4.10 log
10 c/mL for the first positive urine, 3.79 log
10 c/mL for first positive plasma, 9.24 log
10 c/mL for peak urine and 4.53 log
10 c/mL for peak plasma. Kudose et al. obtained similar optimal BK viral load cutoffs of 7.2 and 3.7 log
10 copies/mL in urine and plasma respectively for the detection of BKVN in concurrent biopsy [
10], another study identifying plasma BK viral load of 4.1 log
10 copies/mL as the best discriminator of BKVN [
11]. Therefore viral load cut-offs proposed in the literature of 7 log
10 c/mL in urine and 4 log
10 c/mL in plasma are similar to our results [
2,
3,
12,
13], after accounting for our difference between initial and peak cut-offs.
However since the object of screening is to prevent the development of BKVN, based on our data, we deduce that the urine must be utilised if regular screening is performed early post transplantation to provide an advantage in earlier detection of BKV at a low viral load. There is no advantage in regular screening of plasma because there was no significant difference found in the time to first detection or initial viral load between fortnightly and 3-monthly testing. Additionally in contrast to the urine compartment, there is little difference in initial and peak ROC derived plasma viral load cut-offs for BKVN, suggesting the moment BKV is detected in the plasma the patient is at imminent risk for BKVN. If only plasma is screened, there is limited lead time for intervention to prevent BKVN. Adding to these concerns of relying on testing of the plasma compartment to screen or presumptively diagnose BKVN is a single study showing 11 of 31 patients with BKVN had plasma BK viral loads consistently <4 log
10 copies/mL [
14].
Accordingly, our adopted practice is regular urine BK viral load testing as the primary screening strategy due to its high sensitivity and earlier detection than the plasma compartment, with restriction of plasma testing to patients with a urine viral load exceeding 7 log
10 c/mL. We suggest that utilising a urine first strategy provides an early warning to clinicians to take measures to modify the risk of progressive BKV replication. We confirm that a plasma BK viral load greater than 4 log
10 c/mL is predictive of a risk of BK nephropathy and should prompt diagnostic biopsy and significant modification of immunosuppression or additional therapy according to the clinical situation. This study validates the approach previously proposed by Hirsch et al. [
2].
The major limitation of the study relates to the reliance on histopathology for the diagnosis because histopathological changes occur late in the disease and disease can be missed by biopsy due to focal involvement. Biopsies were performed at the discretion of the treating physician rather than as part of a protocol and the decision to biopsy the patient may have been influenced by the BK viral loads leading to ascertainment bias. Due to logistical constraints, we lack information about risk factors for the development of BKVN and the management of presumptive or confirmed BKVN, however this was not the aim of this study. The transplants screened 3-monthly and fortnightly were largely from two different transplant centres, and BKV management was not protocolised across the centres. Bias according to centre could have potentially contributed to the outcomes of the two groups.
As the presence of BKV antigen-specific T-cell responses are associated with control of BKV replication [
15‐
18], and there are scant reports of the association of BKV genotype with BKVN [
19,
20], with further study it would be interesting to evaluate BKV specific T-cell immunity and BKV genotype for differences in cases with low or undetectable BK viral loads, high BK viral loads without BKVN, and those cases with BKVN.
Conclusions
In this retrospective observational study, we have not been able to demonstrate that better clinical outcomes are associated with more frequent BK viral load testing or testing of a particular compartment. However we have shown if frequent BK viral load testing is employed early post transplantation, the advantage of earlier detection is found for urine but not plasma testing. Additionally, the initial detection of BKV in the plasma occurs at a level already predicting a risk of BKV nephropathy. For these reasons we suggest urine rather than plasma as the optimal compartment for BK viral load screening.
Abbreviations
ATSI, aboriginal or Torres Strait Islander; BKV, BK virus; BKVN, BK virus nephropathy; c/mL, copies per millilitre; CI, confidence interval; IQR, interquartile range; PCR, polymerase chain reaction
Acknowledgements
There was no funding source for this study.