Incidence of BK with Tacrolimus Versus Cyclosporine and Impact of Preemptive Immunosuppression Reduction


*Corresponding author: Daniel C. Brennan,


Our purposes were to determine the incidence of BK viruria, viremia or nephropathy with tacrolimus (FK506) versus cyclosporine (CyA) and whether intensive monitoring and discontinuation of mycophenolate (MMF) or azathioprine (AZA), upon detection of BK viremia, could prevent BK nephropathy.

We randomized 200 adult renal transplant recipients to FK506 (n = 134) or CyA (n = 66). Urine and blood were collected weekly for 16 weeks and at months 5, 6, 9 and 12 and analyzed for BK by polymerase chain reaction (PCR).

By 1 year, 70 patients (35%) developed viruria and 23 (11.5%) viremia; neither were affected independently by FK506, CyA, MMF or AZA. Viruria was highest with FK506-MMF (46%) and lowest with CyA-MMF (13%), p = 0.005. Viruria ≥ 9.5 log10 copies/mL was associated with a 3-fold increased risk of viremia and a 13-fold increased risk of sustained viremia. After reduction of immunosuppression, viremia resolved in 95%, without increased acute rejection, allograft dysfunction or graft loss. No BK nephropathy was observed.

Choice of calcineurin inhibitor or adjuvant immunosuppression, independently, did not affect BK viruria or viremia. Viruria was highest with FK506-MMF and lowest with CyA-MMF. Monitoring and preemptive withdrawal of immunosuppression were associated with resolution of viremia and absence of BK nephropathy without acute rejection or graft loss.


BK virus infection occurs early in life, appears to be asymptomatic and results in a seroprevalence of 65–90% in the general adult population (1). After transplantation, 30–60% of renal transplant recipients develop BK viruria, 10–20% develop BK viremia and 5–10% develop BK virus nephropathy (2–9). Up to 70% with BK nephropathy lose the transplant from the infection, and those that do not lose the transplant often remain with renal dysfunction (10,11). Thus BK virus has emerged as one of the most important problems in renal transplantation.

The BK virus was originally discovered and deemed to be the cause of obstructive uropathy in a patient treated with azathioprine (AZA) and prednisone (12). BK infection in transplant recipients occurred rarely throughout the 1970s and 1980s when only AZA and prednisone were available as maintenance immunosuppressive therapy. However, a single-center analysis of archived renal biopsy specimens showed an increase in the incidence of the disease contemporaneous with the introduction and use of more intense immunosuppression with tacrolimus (FK506), mycophenolate mofetil (MMF) (CellCept, Hoffman-La Roche, Nutley, NJ) or the combination compared to previous eras (7–9,11–14). BK viruria is the first sign of active virus replication and progression to BK viremia appears to be a pre-requisite for the development of BK nephropathy (5,6). There is no known safe and effective antiviral agent, and reduction of immunosuppression, in an attempt to clear the virus, risks precipitation of acute rejection and graft loss.

The primary purpose of this study was to determine the effect of the calcineurin inhibitors, tacrolimus (FK506) or modified cyclosporine A (CyA), on the incidence of BK virus infection in de novo renal transplant patients. The secondary purpose was to test the safety and efficacy of a management strategy consisting of intensive monitoring for BK viruria and viremia with preemptive reduction of adjuvant immunosuppression, MMF or AZA, upon detection of BK viremia to prevent progression to BK nephropathy.

Materials and Methods

Study design

This was an open-label prospective trial of de novo renal transplant recipients randomized, prior to transplantation, in a 2:1 block-design fashion stratified by race and gender to receive FK506 or CyA at a single center. The Washington University Human Studies Committee approved the study.

Inclusion and exclusion criteria

All adult patients (age >18 years) were considered for enrollment. Exclusion criteria were previous treatment with rabbit anti-T-cell polyclonal agents or allergy to rabbit proteins, malignancy, except skin cancers in the previous 2 years, serologic evidence of infection with HIV-1, HTLV-1 or hepatitis B, use of an investigational agent in the prior 4 weeks or inability to comply with the protocol in the opinion of the investigators. Pregnant or nursing females were not included.


All patients except 6 antigen-matched living-related allograft recipients were induced with rabbit anti-thymocyte globulin [Thymoglobulin (TMG), SangStat Medical Corp., Fremont, CA]. Four doses of TMG (1.5 mg/kg each) were administered; the first intra-operatively followed by doses on post-operative days 1–3. Dose adjustments were made for leukopenia or thrombocytopenia.

Calcineurin inhibitors were instituted upon a brisk diuresis but no later than 4 days post-operatively. Cyclosporine was started at 8 mg/kg per day in two divided doses targeting 12-h whole blood trough levels of 250–300 ng/mL, using monoclonal fluorescent polarization immunoassay (FPIA) (Abbott Laboratories, Abbott Park, IL) during the first 3 months and 100–250 ng/mL, subsequently. FK506 was initiated at 0.1 mg/kg per day divided in two doses aiming at trough levels of 5–10 ng/mL by microparticle enzyme immunoassay (MEIA) (IMx, Abbott Laboratories, Abbott Park, IL).

Patients were assigned to receive AZA at a dose of 5 mg/kg per day for 3 days then 2.5 mg/kg per day orally as the primary adjuvant agent. The dosage was adjusted per WBC count. Under certain circumstances (glomerulonephritis or other immunologic basis for end-stage renal disease (ESRD); second transplant; high sensitization with panel reactive antibody (PRA) >20%; or history of gout) MMF was substituted for AZA at 1000 mg orally twice a day. Methylprednisolone (7 mg/kg IV) was infused intra-operatively. Prednisone was started at 1 mg/kg per day, orally, post-operatively and tapered by month 3 to 5–7.5 mg daily.

Cytomegalovirus (CMV) prophylaxis

Recipients at risk for CMV received oral ganciclovir 1000-mg t.i.d. for 3 months if D−/R+, 6 months if D+/R+ and 12 months if D+/R−.


Undiluted whole blood, plasma and urine samples for BK virus polymerase chain reaction (PCR) were collected pre-transplant, weekly for 16 weeks, and at months 5, 6, 9 and 12. BK viruria was defined by presence of BK virus DNA in the urine. BK viremia was defined by detection of BK virus DNA in plasma. Sustained viruria and viremia were defined as two or more consecutive positive urine or plasma samples, respectively, spanning 3 or more weeks, and ‘transient’ otherwise.

An early internal analysis compared the results of PCR performed on DNA extracted from 419 blood samples: 26 were positive in both plasma and whole blood, 11 positive in plasma but negative in whole blood while 2 were positive in whole blood but negative in plasma and 380 negative in both plasma and whole blood (p = 0.03, McNemar test). Based on these results, all subsequent testing of blood samples was performed only on plasma. Plasma samples were tested on all dates that the urine sample from the same date was positive for BK virus. This was based on another internal analysis in which plasma PCR was performed and found to be uniformly negative on 150 samples for which the corresponding urine was negative for BK virus. A smaller internal analysis of six urine samples, positive for BK virus by PCR, showed that decoy cells could be detected in less than half and was deemed an unreliable method to follow patients at our institution. Patient contact, clinic and laboratory follow-up were maintained throughout the study, with at least bi-weekly review of all patient data.

Polymerase chain reaction methods

For purposes of this study, qualitative PCR for detection of BKV DNA was used for patient management. DNA was extracted and purified from urine and ethylene-diamine-tetra-acetic acid (EDTA) anti-coagulated plasma using QIAamp spin columns (QIAGEN Inc., Valencia, CA) according to the manufacturer's instructions. Extracted DNA was tested for BK virus DNA by real-time PCR using the LightCycler System (Roche Diagnostics Corporation, Indianapolis, IN) using the primers, Pep-1 AGT CTT TAG GGT CTT CTA CC and Pep-2 GGT GCC AAC CTA TGG AAC AG, which amplify a 176-base-pair segment of the BK virus large T-antigen gene (15).

BK virus-specific hybridization probes 5′TTG CCA TGA AGA TAT GTT TGC CAG TGA TGA FITC′3 and 5′LCRed640 GAA GCA ACA GCA GAT TCT CA′3 (TIB Molbiol LLC, Adelphia, NJ) were used for detection. These probes were designed to detect BK virus without detection of the related polyoma viruses, JC virus and SV-40.

Reactions were carried out in a 20-μL volume that included 2 μL of Roche LightCycler FastStart DNA Hybridization Probe 10× reaction mix and 2 μL of sample DNA. The final concentrations of other components were 3.5-mM MgCl2, 0.5 μM of each primer and 0.2 μM of each probe. The reaction program consisted of denaturation for 7 min at 95°C followed by 45 cycles of denaturation at 95°C instantaneous, annealing at 52°C for 10 s and extension at 75°C for 7 s. Each reaction included positive controls consisting of 1000 copies of BK virus plasmid DNA ( ATCC 45026) obtained from the American Type Culture Collection (Manassas, VA) and negative controls for amplification and DNA preparation. The sensitivity of the qualitative assay was 20–50 BK virus DNA copies per reaction. The validity and performance of our assay, including the lack of detection of JC virus and SV-40, have been confirmed in a multicenter international inter-laboratory comparison (manuscript in preparation).

Quantitative polymerase chain reaction

BK positive samples were quantified retrospectively by repeat real-time PCR analysis of aliquots of extracted DNA that had been frozen at −70°C, using four copy numbers of control BK DNA (2 × 108, 2 × 106, 2 × 104 and 2 × 102). The quantitative standard was prepared from a plasmid (pBKV[35-1]) that contains the entire linearized BK genome ( ATCC 45026) using standard techniques of plasmid growth and purification. The copy number of the standard was determined by spectrophotometry. The coefficient of variability for the quantitative BK LightCycler assay in our laboratory was less than 6% for results expressed as log10 copies/mL. The lower limit of sensitivity of the quantitative PCR assay is 10 000 copies/mL, but it reliably detects 25 000 copies/mL and absolutely detects 5 log10 copies/mL. Although the LightCycler was able to generate a number far below these levels through extrapolation from the standard curves, we chose to use the more conservative result from interpolation of the standard curves. Therefore, positive, but unable to quantitate, BK samples were thus assigned a median value of 2.5 log10 copies/mL.

Management protocol

Identification of BK viremia triggered discontinuation of AZA or MMF. If viremia failed to clear within 4 weeks, the calcineurin inhibitor dose was tapered to trough CyA levels of 100–200 ng/mL or trough FK506 levels of 3–5 ng/mL.

Allograft biopsy

All patients with unexplained elevation of the serum creatinine underwent a kidney biopsy interpreted by a pathologist blinded to the treatment arms and BK virus status. At least two cores per biopsy were obtained with an 18-gauge biopsy needle. All specimens were examined with a dissecting microscope by the pathology technician in attendance for the presence of medullary and cortical tissue and evaluated with light microscopy as well as immunoperoxidase staining using a monoclonal Ab specific for BK virus (MAB8505; Chemicon, Temecula, CA). Acute rejection was defined by the Banff '97 criteria (16). All biopsy specimens were also evaluated for BK nephropathy by an independent pathologist outside the institution and blinded to the treatment regimen and BK virus status of the patients during and after the study.

Endpoints and outcomes

Primary endpoints were incidence of BK viruria, viremia and nephropathy in each arm at 1 year. Also studied were the rates of acute rejection, patient and graft survival and renal function at 3, 6, 9 and 12 months.

Sample size and statistics

Enrollment was based on an incidence of BK viruria of 45%. To find a difference of 25% with an alpha of 0.05 and power of 0.80, 45 patients in the CyA and 90 in the FK506 arm were to be enrolled. An interim analysis suggested the number enrolled should be increased to 200, as the observed difference in the incidence of viruria in the first 36 patients was 15%. Differences in the characteristics of patients were tested with Student's t-test for continuous variables, Fisher's exact test for binary categorical variables and an analysis of variance (ANOVA) for multiple categorical variables. Incidence comparisons were made with the Kaplan–Meier method. Multivariate analysis of incidence was performed with Cox proportional hazards regression. Subjects were censored from incidence calculations at the time of collection of their last specimen.

Multivariate models were constructed using stepwise selection over the following donor, recipient and transplant characteristics and intermediate outcomes: donor—race, gender, age, body mass index, cold ischemia time, warm ischemia time, expanded criteria; recipient—race, gender, age, body mass index, cause of end-stage renal disease and pre-existing conditions (anemia, cardiovascular disease, hypertension and diabetes); transplant— individually HLA-A, B or DR mismatches, total HLA-A, B and DR mismatches, donor and recipient HLA identical, donor and recipient CMV seropairing, ureteral stent placement and immunosuppression; and intermediate outcomes: lymphocele, acute rejection and CMV disease measured as time-varying covariates beginning at the time of diagnosis. Forward and backward stepwise selection were used to construct all multivariate models due to the large number of factors considered relative to the sample size.


Randomization and immunosuppressive regimens

From December 27, 2000 to October 29, 2002, 226 patients were considered for enrollment and 26 patients (11%) were ineligible (n = 6) or refused consent (n = 20). Of 200 enrolled subjects, 134 (67%) were randomized to receive FK506 and 66 (33%) to receive CyA and followed for 1 year. Demographic variables were similar between groups (Table 1). In the FK506 group, 69 (52%) received AZA and 65 (49%) received MMF. In the CyA group, 43 (65%) received AZA and 23 (35%) received MMF. There were 3 crossovers from CyA to FK506 in 3 women for hirsutism. One patient switched for interstitial nephritis and one for rejection. Three patients in the CyA arm were switched to sirolimus. Two were for CyA toxicity and 1 was for rejection. In the FK506 group, 2 patients were switched to sirolimus for FK506 toxicity and one for rejection. There were two crossovers from AZA to MMF for rejection. Patients were analyzed on an intent-to-treat basis. Analyzing them after the conversion did not impact the overall results.

Table 1.  Patient baseline demographics
 FK n = 134CyA n = 66p-values
  1. FK = tacrolimus; CyA = cyclosporine; ESRD = end-stage renal disease; GN = glomerulonephritis; PCKD = polycystic kidney disease; BMI = body mass index; HLA = human leukocyte antigen; CMV = cytomegalovirus seropositivity; D = donor; R = recipient; CIT = cold ischemia time.

Age (years)44 ± 1346 ± 130.30
 Male86 (64%)40 (61%)0.73
Race 0.99
 Caucasian106 (79%)52 (79%) 
 African American24 (18%)12 (18%) 
 Others4 (3%)2 (3%) 
ESRD 0.41
 Diabetes30 (22)18 (27) 
 Hypertension24 (17)15 (22) 
 GN46 (34)14 (21) 
 PCKD13 (9)6 (9) 
 Others21 (15)13 (19) 
Donor 0.59
 Deceased79 (58%)34 (51%) 
 Living55 (41%)32 (48%) 
BMI25.9 ± 4.627 ± 50.13
HLA mismatch2.282.480.41
CMV serostatus 0.52
 D−/R−26 (19%)19 (28%) 
 D−/R+30 (22%)13 (20%) 
 D+/R−31 (23%)14 (21%) 
 D+/R+47 (35%)20 (30%) 
CIT13 ± 4.812 ± 4.30.38

Incidence and timing of BK virus

Of the 8400 total possible samples from recipients, 7177 (85%) were collected. Missing samples were explained primarily by the following: total non-compliance in 3 patients who were censored; 5 patients thrombosed their kidneys, early, and were censored; 1 elderly non-compliant patient died early; 1 patient had a severe rejection and lost the graft, early, and no further samples were collected, and finally most patients were unable to provide a urine sample when they presented for transplantation because of anuria. Thus we collected >90% of samples from analyzable patients.

The estimated incidence of BK viruria through 365 days post-transplant was 35% (Figure 1A). The baseline demographic characteristics did not affect incidence of BK viruria. BK viremia was detected in 23 patients. BK viremia never occurred in the absence of viruria and almost always followed BK viruria. The overall estimated incidence of viremia was 11.5% through 365 days (Figure 1A). Sustained BK viremia was detected in 11 (6%) patients. No BK nephropathy was seen at 12 months, and none had been seen at the last date of observation with a mean observation period of 32 ± 6 months.

Figure 1.

Overall cumulative incidence of BK viruria (yellow line) and viremia (red line). (A) Cumulative incidence of BK viruria (FK506 = teal line; CyA = magenta line) (B). Cumulative incidence of BK viremia (FK506 = teal line; CyA = magenta line) (C).

The median onsets of viruria and viremia were 40.5 (range: 0–415 days) and 60 days (range: 18–276 days) after transplantation, respectively.

The median interval from the onset of viruria to the onset of viremia was 27 days (range: 0–147 days). Viruria always preceded viremia, with the exception of 1 patient with simultaneous onset at month 12. After onset of viruria, levels of BK viral DNA in the urine often increased rapidly, with a median interval from onset to peak level of 21 days (range: 0–325 days).

BK viral levels in urine varied over a wide range. The median peak levels in recipients were 8.98 log10 copies/mL (range: <5–12.5 log10 copies/mL) in urine and 5.04 log10 copies/mL (range: <5–6.0 log10 copies/mL) in plasma. Sustained viremia was also more likely to develop in recipients with earlier onset, higher peak urine titers and longer duration of viruria compared to recipients with viruria alone. (Table 2 and Figure 2A,B). Compared to recipients with transient viremia, sustained viremia was more likely to develop in recipients with longer durations of viruria (p = 0.05), longer durations of viremia (p < 0.001) and higher plasma titers (p < 0.001).

Table 2.  Viral parameters from recipients with BK viruria alone, transient viremia and sustained viremia*
ParametersType of BK virus infectionp-values
Viruria aloneViremiaViruria alone versus:
Transient (n = 11)Sustained (n = 36)Transient (n = 12)Sustained (n = 11)Transient viremiaSustained viremia
  1. *All values are medians. Statistical analysis was performed with the Mann-Whitney U-test.

  2. Duration of longest interval of continuous BK virus infection.

  3. An isolated positive sample was considered to span 7 days.

Time from transplantation to onset of viruria (days)1534330.5190.210.02
Time from transplantation to onset of viremia (days)90.548
Interval from onset of viruria to onset of viremia (days)3027
Longest period of viruria (days)7213.51393550.24<0.001
Longest period of viremia (days)753
Peak urine titer (log10 genomes/mL)6.228.579.6510.810.06<0.001
Peak plasma titer (log10 genomes/mL)2.505.60
Figure 2.

Relationships between types of BK virus infections and viral parameters. Viruria and viremia include all recipients with a positive urine and plasma BK sample, respectively. Viruria alone includes only viruric recipients who never developed viremia. Viremia was either transient or sustained, defined as ≥2 consecutive positive plasma samples spanning ≥3 weeks. (A) Relationship between the onset of BK viruria and the development of BK virus infections. (B) Relationship between the peak urine titers of BK virus and types of BK virus infection. (C) Urine peak viral titers in recipients with different types of BK virus infections. Minimum and maximum titers in each group are noted with whiskers. Titers from the 10th to 90th percentiles are boxed. The median is indicated by the dash (-). The threshold titer is indicated by the horizontal line at 9.5 log10 copies/mL.

Viruria threshold for sustained viremia

Viruria alone occurred in 47 patients. Transient viruria occurred in 11 and sustained viruria in 36. As above, viremia occurred in 23 patients. Transient viremia occurred in 12 patients and sustained viremia in 11 patients. Retrospective inspection of the data suggested that a peak urine level of 9.5 log10 copies/mL was a threshold above which patients were at high risk of sustained viremia. (Figure 2C). Thirty subjects had a peak urine level ≥9.5 log10 copies/mL: 1 of 11 (9%) with transient viruria, 13 of 36 (36%) with sustained viruria, 6 of 12 (50%) with transient viremia and 10 of 11 (91%) with sustained viremia. In comparison, of the 40 subjects whose peak urine levels were <9.5 log10 copies/mL, 6 (15%) had transient viremia (p = 0.002) and only 1 (2.5%) had sustained viremia (p < 0.001). Thus, a urine level of ≥9.5 log10 copies/mL was associated with a 3-fold increased risk of viremia and a 13-fold increased risk of sustained viremia. The sensitivity, specificity, positive predictive value and negative predictive values of a urine level of >9.5 log10 copies/mL were 70%, 70%, 53% and 83% for any viremia, and 91%, 66%, 33% and 98% for sustained viremia.

The interval between onset of viruria and the first urine specimen with a BK virus level exceeding 9.5 log10 copies/mL was usually 1–2 weeks, and exceeded 4 weeks in only 4 recipients and 1 with sustained viremia. The interval between the first occurrence of a urine level exceeding 9.5 log10 copies/mL and the detection of BK virus in plasma was typically 2 weeks or less and was less than 4 weeks in all 10 recipients with sustained viremia. By 4 weeks after onset of viruria, 44 recipients, including 2 with sustained viremia, had not exceeded the threshold value of 9.5 log10 copies/mL of urine. Thus, monitoring urine beyond 4 weeks in recipients with titers below 9.5 log10 copies/mL would identify <5% of recipients who would progress to sustained viremia.

Risk factors for BK viruria, viremia and sustained viremia

Multivariate analysis revealed several associations with BK viruria, viremia and sustained viremia (Table 3). An increased hazard of BK viruria was associated with the use of FK506-MMF (HR 2.5, p = 0.002), and CyA-AZA (HR 2.0, p = 0.038) relative to either FK506-AZA or CyA-MMF. An increased hazard of BK viremia was associated with placement of ureteral stents (HR 3.0, p = 0.018). An increased hazard of sustained BK viremia was also associated with the placement of ureteral stents (HR 4.3, p = 0.044). Stents were placed in 54 patients: 22 of 70 (31%) with and 32 of 128 (25%) without viruria, 10 of 23 (43%) with and 44 of 175 (25%) without viremia and in 5 of 11 (46%) with and 49 of 187 (26%) without sustained viremia. There were no significant differences in the use of stents among the immunosuppressive combination groups: 8 of 42 (18%) in the CyA-AZA group, 5 of 22 (23%) in the CyA-MMF group, 22 of 71 (31%) in the FK506-AZA group and 19 of 63 (30%) in the FK506-MMF group. All other factors analyzed in the models were not associated with BK viremia, viruria or sustained viremia including the presence of HLA-DR mismatches between the donor and recipient, age, race, warm or cold ischemia time.

Table 3.  Multivariate analysis of factors related to BKV viruria, viremia and sustained viremia*
 Viruria parameter (HR; p-values)Viremia parameter (HR; p-values)Sustained viremia parameter (HR; p-values)
  1. *Forward and backward stepwise multivariate Cox regression were used with entry and exit criteria set at 0.05.

  2. Hazard ratios (HR) were calculated by the exponential transformation of the parameter estimates.

FK506-MMF0.40 (2.5; 0.002)
CyA-AZA0.30 (2.0; 0.034)
Urethral Stent0.48 (3.0; 0.018)0.63 (4.3; 0.045)

Effect of immunosuppression

There was no difference in the observed occurrence of BK viruria in subjects randomized to FK506 compared to CyA (36% vs. 31%), independently, or in subjects who received AZA compared to MMF (33% vs. 38%), independently, as shown in Figure 1B and Table 4. There was also no difference in the occurrence of BK viremia in patients randomized to FK506 or CyA (12% vs. 11%, p = 1.0), independently, (Figure 1C), or AZA compared to MMF (13% vs. 9%, p = 0.46), independently. Sustained viremia was observed in 10 patients in the FK506 group and 1 patient in the CyA group (p = 0.10) and 4 patients receiving MMF and 7 patients receiving AZA, p = 0.76. CyA and FK506 levels were not different in subjects with or without BK viruria early after transplant (Figure 3). Thereafter, the FK506 or CyA levels in the BK viremic patients decreased due to the protocol-driven immunosuppression reduction. Mean AZA doses decreased in the CyA group and mean MMF doses decreased in the FK506 group.

Table 4.  Effect of demographic factors and immunosuppression on the occurrence of BK viruria
 All (n = 200)BKV + (n = 70)BKV − (n = 130)p-values
  1. ESRD = end-stage renal disease; GN = glomerulonephritis; PCKD = polycystic kidney disease; HLA = human leukocyte antigen; CMV = cytomegalovirus seropositivity; D = donor; R = recipient; CIT = cold ischemia time, FK = tacrolimus; CyA = cyclosporine; AZA = azathioprine; MMF = mycophenolate mofetil.

Age (years)45 ± 1447 ± 1344 ± 130.12
Gender 0.90
 Male12645 (35%)81 (65%) 
 Female7425 (33%)49 (66%) 
Race 0.44
 Caucasian15854 (34%)104 (65%) 
 African American3615 (41%)21 (58%) 
 Others61 (16%)5 (83%) 
ESRD 0.72
 DM4819 (40%)29 (60%) 
 HTN3914 (35%)25 (64%) 
 GN6020 (33%)40 (66%) 
 PCKD198 (42%)11 (57%) 
 Misc/others349 (26%)25 (74%) 
Donor 0.61
 Deceased11342 (37%)71 (62%) 
 Living8728 (32%)59 (83%) 
CMV serostatus 0.85
 D−/R−4515 (33%)30 (66%) 
 D−/R+4313 (30%)30 (69%) 
 D+/R−4517 (37%)28 (62%) 
 D+/R+6725 (37%)42 (62%) 
CIT (deceased donors only)12.7 ± 4.712.16 ± 4.412.5 ± 5.10.75
Immunosuppressive agents
 FK13449 (36%)85 (63%)0.61
 CyA6621 (31%)45 (68%) 
 AZA11237 (33%)75 (66%)0.52
 MMF8833 (38%)55 (62%) 
 FK-AZA6919 (27%)50 (72%)0.02
 FK-MMF6530 (46%)35 (53%) 
 CyA-AZA4318 (41%)25 (58%) 
 CyA-MMF233 (13%)20 (87%) 
Figure 3.

Immunosuppression in the study population. (A,B) Mean FK506 and CyA levels (ng/mL) in the BK viremic (+) and BK non-viremic (−) groups. (C,D) Mean MMF and AZA doses (mg) in the FK506 and CyA groups.

The incidence of BK viruria was lower at 1 year in subjects who received CyA-MMF (13%) compared to FK506-MMF (46%) (p = 0.005) (Figure 4A). The incidence of BK viruria was also lower in patients who received FK506-AZA (27%) compared to FK506-MMF (46%) (p = 0.03). The incidence of BK viremia tended to be lowest in the CyA-MMF (4%) group and highest in the CyA-AZA group (15%) (p = 0.27) (Figure 4B). The incidence of BK viremia was nearly identical in patients randomized to FK506 receiving either AZA (12%) or MMF (13%) (p = 0.80). The incidence of sustained viremia did not differ among the groups (p = 0.28) (Figure 4C).

Figure 4.

Cumulative incidence of BK viruria in the immunosuppression subgroups: CyA-Aza = magenta; CyA-MMF = yellow; FK506-Aza = teal; FK506-MMF = maroon. (A) Cumulative incidence of BK viremia in the immunosuppression subgroups: CyA-Aza = magenta; CyA-MMF = yellow; FK506-Aza = teal; FK506-MMF = maroon (B). Cumulative incidence of sustained BK viremia in the immunosuppression subgroups: CyA-Aza = magenta; CyA-MMF = yellow; FK506-Aza = teal; FK506-MMF = maroon (C).

Twelve subjects did not receive Thymoglobulin. Of these, BK viruria was detected in 5 (42%). The relationships between BK viruria or viremia and immunosuppressive regimen were not changed if these subjects were excluded.

Clinical signs and symptoms

Viruria and viremia remained silent, clinically. We did not observe any indication of development of BK nephropathy in the viremic or viruric patients. Ureteral stenosis or hemorrhagic cystitis was not seen.

Renal function

At 12 months, serum creatinine levels were lower in the FK506 group overall compared to the CyA group (1.3 ± 0.3 vs. 1.6 ± 0.7 mg/dL, p = 0.03). There was no difference in the serum creatinine levels between the BK viruria positive and negative groups at 1, 3, 6 and 12 months after transplant. The serum creatinine level at 12 months in the BK viruria positive group was 1.5 ± 0.4 mg/dL and 1.4 ± 0.5 mg/dL in the BK viruria negative group (p = 0.15). At 12 months the serum creatinine level of patients who developed viremia (1.5 ± 0.42 mg/dL) or sustained viremia (1.6 ± 0.52 mg/dL) was not different from those who never became viremic (1.4 ± 0.5 mg/dL, p = 0.38 and 0.21, respectively).

Effect of reduction of immunosuppression

Reduction of immunosuppression was associated with clearance of viremia in 22 of 23 viremic patients (95%) by 1 year after transplant. The mean time to clearance of viremia was 54 days (range: 7–213 days). Only 5 of 23 (21%) viremic patients cleared the urine during the observation period of the study. In all 5 cases, viremia resolved before viruria, consistent with the concept, and our internal analysis that viremia is unusual or non-existent in the absence of simultaneous viruria. One patient remained viremic during the course of follow-up. In 7 patients, viremia cleared after cessation of the adjuvant agent alone, in 2 patients a decrease in the calcineurin dose alone was made. Six patients required cessation of the adjuvant agent and a decrease in the calcineurin dose. Viremia cleared in 7 patients with a single positive plasma (transient) BK with standard immunosuppressive tapering and before protocol-driven discontinuation of AZA or MMF was implemented.

Acute rejection

Forty (20%) patients, underwent a total of 52 allograft biopsies. Ten cases of acute rejection (5%) occurred overall. Six (4%) occurred in the FK506 group and 4 (6%) in the CyA group (p = 0.72). Only one rejection episode was directly related to protocol-directed immunosuppression reduction. This patient developed viremia and was asked to decrease his FK506 dose as his qualitative viremia persisted despite discontinuing his adjuvant agent. He suffered a Banff IB rejection but responded to pulsed steroids and treatment with intravenous immunoglobulin. He stabilized on an FK506-sirolimus combination. His serum creatinine level was 1.8 mg/dL at the end of the study, and was 2.1 mg/dL at last report 36 months post-transplant. Another patient, received an expanded criteria donor allograft, and had a nadir serum creatinine of 1.8 mg/dL. He was asked to reduce the dose of FK506 because of persistent viremia. He stopped taking immunosuppression altogether and suffered a Banff II rejection. He was successfully treated with intravenous immunoglobulin and a steroid pulse. His serum creatinine stabilized at 2.5 mg/dL, and was 3.2 mg/dL at last report, 40 months post-transplant. Significantly, both patients had cleared BK-viremia before the rejection episode and remained viremia-free at the end of the study. The 6 other rejection episodes were not related to any intentional or unintentional immunosuppression changes and occurred in the absence of BK viruria or viremia.

Patient survival and graft loss

There were no significant differences in patient survival or graft loss between the two groups. Two patients in the FK506 arm died. One patient, who thrombosed the kidney, died after returning to dialysis and 1 patient died from hepatic carcinoma. Thus, patient survival was 99% in the FK506 group and 100% in the CyA group. Five patients, all in the FK506 arm, lost their graft due to thrombosis from technical complications in the first few days after transplant necessitating return to dialysis. One patient in the FK506 arm lost the allograft from severe rejection. Death-censored graft survival was 95% in the FK506 group and 100% in the CyA group.


BK nephropathy has emerged as an important cause of allograft loss and renal dysfunction in renal transplant recipients. Use of FK506 compared to CyA, or MMF compared to AZA, in maintenance immunosuppression has been implicated as an important determinant of BK viruria, viremia and nephropathy (7–9,11–14). We sought to determine if this were true in a randomized prospective trial of de novo renal transplant recipients. BK viremia has also been reported to be a prerequisite for progression to BK nephropathy. Thus, BK viremia may be an indicator of excessive immunosuppression. We hypothesized that BK viremia could be cleared safely by reduction of immunosuppression and specifically by withdrawal of the anti-metabolite component of the immunosuppressive regimen, thus preventing progression to BK nephropathy without a significant risk for acute rejection, renal dysfunction or graft loss.

Our study revealed no difference in the rate of BK viruria or viremia among those receiving FK506 compared to CyA, independently. Because of our immunosuppression protocol in which we use either AZA or MMF, as described in the Methods section, we were uniquely situated to explore the role of AZA compared to MMF.

We saw no differences in the rate of BK viruria or viremia with AZA compared to MMF, independently. Retrospective power calculations revealed that we had a power = 1. Thus, we can say with 100% statistical certainty that we had a sufficient number of patients to detect a difference between the AZA and MMF groups, if there was a difference to be detected. Our data support that although AZA is not a modern immunosuppressant, it is a potent immunosuppressant, as we use it. A recent study has shown that AZA is as effective as MMF in preventing rejection and less expensive (17).

The incidence of sustained viremia, however, tended to be higher with FK506 than CyA when the effect of each calcineurin inhibitor was analyzed, independently. This may be a Type 2 error. That is, we did not show a true difference when one may have existed because of too few numbers of observations. However, of the four possible combinations of calcineurin inhibitor and anti-metabolite, the CyA-MMF combination was associated with the lowest incidence of viruria and viremia and FK506-MMF the highest. This finding may be a Type 1 error. That is, we may have seen a difference when one did not truly exist because the number of observations was small. An alternative and plausible reason is that CyA lowers MMF levels and FK506 does not (18). The use of and interpretation of MMF levels remain to be defined, and we did not measure MMF or mycophenolate acid (MPA) levels. MMF doses on average, however, did not differ between the FK506 and CyA groups but tended to be higher in the CyA groups. Nevertheless, taken together, our findings support the idea that an FK506-MMF combination is the most permissive regimen for BK reactivation.

Besides the immunosuppressive regimen, our multivariate analysis revealed that use of ureteral stents was associated with development of BK viremia and sustained viremia. We do not know the precise reasons for the associations or lack thereof. Ureteral stents were placed at the time of transplantation and upon surgeon discretion. These were placed routinely by 2 of the 5 surgeons and uncommonly by 3 of 5 during the study period. Stents were routinely removed by 4 weeks after transplantation. This was before the median onset of viruria (40 days) or viremia (60 days). The small number of stents placed prevented any determination of placement or removal of stents with timing of viruria or viremia. Because only 54 patients had stents (of whom 22 had viruria), this can at best be only a partial explanation for BK after renal transplantation.

The most important finding of this randomized prospective study may be that intense monitoring early after transplantation for BK viruria and viremia, using a qualitative assay, coupled with prompt reduction in immunosuppression upon detection of persistent BK viremia was associated with resolution of viremia in almost all patients and the absence of progression to clinically evident BK nephropathy at 1 year, and with a mean follow-up observation period of almost 3 years.

Retrospective power calculations revealed that we had a power = 1 assuming an incidence of BK nephropathy of 3%. The lack of clinically evident BK nephropathy occurred despite the use of modern, potent quadruple immunosuppressive therapy, including Thymoglobulin. Indeed, there was no difference in the incidence of viruria or viremia among those who did or did not receive Thymoglobulin, confirming a recent report (9).

Although reduction of immunosuppression has the potential to precipitate acute rejection, only 10 patients (5%) suffered an acute rejection episode. Only 1 of these 10 episodes was directly related to lowering of immunosuppression in response to BK viremia. These results support the hypothesis that BK viremia may serve as an ‘in-vivo bioassay’ of excessive immunosuppression and resolves without major consequence if the immunosuppression is carefully reduced upon the detection of viremia.

We performed biopsies for cause only. We may have failed to detect subclinical BK nephropathy in some of the viremic patients. However, we believe this is of limited clinical significance as almost all of our patients cleared viremia and serum creatinines did not differ in the viremic and non-viremic patients.

In another prospective study in which serial surveillance biopsies were performed, BK virus in renal tissue was identified in two clinical settings: on routine surveillance biopsies, when it was not clinically associated with allograft dysfunction, and in association with allograft dysfunction (9). The patients diagnosed when clinically silent did well compared to patients who had allograft dysfunction leading to diagnosis.

It is not entirely surprising that BK virus might be found in surveillance biopsies as viral shedding arises from the renal tubules or uroepithelium, as detected by the presence of decoy cells in urine or by BK-PCR in approximately 30–35% of renal transplant recipients (5,6). Our success in prevention of BK nephropathy was not at the cost of higher rejection rates or allograft dysfunction. The overall rejection rates were extremely low (5%), and serum creatinines in the BK viruric, viremic or sustained viremic patients were excellent and did not differ.

Our data show strong relationships between the onset, duration and titer of virus in the urine and the development of viremia, suggesting that viremia reflects the intensity of infection in the allograft. It appears that BK virus begins to replicate, and is poorly controlled in some recipients. In this subset, an early, intense viral infection ensues, with a 1000-fold or greater increase in the level of urinary virus within a relatively short time frame of 2–3 weeks. In this subgroup, the increase in viral replication resulted in detectable viremia. The reason for this observation remains to be elucidated. The practical insight from this observation is that within a short time after the onset of viruria, a subgroup of patients can be identified whose high levels of BK viruria (greater than 9.5 log10 copies/mL) indicate that they are at high risk of viremia and probably of nephropathy if no intervention is undertaken to bring the viral replication under control. It may be desirable to focus future control efforts on this critical level and time period. However, while the level and early onset of viremia appear to be critical, the positive predictive value was low, and it is unlikely that risk is restricted to this level or period, especially in recipients whose immunosuppression is increased at a later time point because of allograft rejection or whose immunosuppression is effectively increased because of deteriorating renal function and reduced clearance of MMF.

Other investigators have found a strong relationship between BK viremia and transplant nephropathy. Hirsch et al., have suggested that a plasma viral titer >10 000 copies/mL be termed ‘presumptive’ BKV nephropathy, even in the absence of histological or biochemical evidence of nephropathy (5). In their study, BKV viremia had a sensitivity of 100% and a specificity of 88% for BK nephropathy, and all recipients with BKV nephropathy had plasma titer >7700 copies/mL. However, in our study 14 recipients, (representing 61% of all recipients with viremia) including 10 with sustained viremia reached plasma titers >100 000 copies/mL with no deterioration of renal function or evidence of BK nephropathy. Thus, our results suggest that intense monitoring and early reduction of immunosuppression upon detection of BK viremia, with qualitative monitoring, can prevent the development of clinically significant BK nephropathy.

In conclusion, this large, prospective, randomized study demonstrated an overall incidence of BK viruria of 35% and BK viremia of 12%, which is similar to rates reported in several smaller studies. We saw no differences in the incidence of BK viruria or viremia among those receiving FK506 or CyA, MMF or AZA, independently. There was a trend toward more sustained viremia with FK506 compared to CyA. There was a higher incidence of BK viruria and viremia with use of FK506-MMF and CyA-AZA and a lower incidence with CyA-MMF. Thymoglobulin did not increase the incidence of viruria or viremia. Our strategy of intensive monitoring and preemptive withdrawal of the antimetabolite upon qualitative detection of persistent BK viremia was associated with resolution of viremia and absence of progression to clinically evident BK nephropathy, without an unacceptable risk of precipitating acute rejection or allograft dysfunction.

Screening for BK viremia provides direct information about the risk of BK nephropathy and a rationale for immunosuppression reduction. Our data suggest that a proactive, preemptive approach can avoid progression from BK viremia to BK nephropathy.


We appreciate the clinical follow-up provided by Drs. Brent Miller, Matt J. Koch and Decha Enkvetchakul, and the surgical contribution of Drs. Jeffrey Lowell, Surendra Shenoy, Martin Jendrisak and Niraj Desai. We appreciate all the long-term care and diligence of the Barnes-Jewish Hospital 6500 personnel, the transplant coordinators, residents and fellows involved in the treatment and tracking of these patients. We especially thank the assistance of Alejandro Alvarez, Adeel Ansari, David Beffa, Connie Ceriotti, Pawel Dyk, Lissa Lopez-Rocafort, Mitch Mahon, Jana Smith and Hiwot Woldu.

This work was supported in part by NIH 1 K24–02886 (DCB), NIH K25-DK-02916-03 (MAS) and Fujisawa, Inc.