Presentation of BK polyomavirus-associated hemorrhagic cystitis after allogeneic hematopoietic cell transplantation

Abstract

BK polyomavirus (BKPyV) has been associated with hemorrhagic cystitis (HC) after allogeneic hematopoietic cell transplantation (HCT), but the natural history of HC and factors associated with the clinical course are incompletely understood. We retrospectively analyzed allogeneic HCT patients transplanted from 2007-2017 who presented after platelet engraftment or after day 28 post-HCT with BKPyV-associated HC (BKPyV-HC), which was defined as a positive urine BKPyV PCR, ≥1 plasma BKPyV viral load result, and macroscopic hematuria (Bedi grade ≥2). Factors associated with resolution of macroscopic hematuria and resolution of all cystitis symptoms within 90 days after HC diagnosis were investigated in multivariable models. In 128 patients with BKPyV-HC, the median times from diagnosis to resolution of all symptoms, macroscopic hematuria, and urinary clots (present in 55% [71/128]) were 24 days (15-44), 17 days (10-30), and 14 days (5-26), respectively. Ninety percent of patients had BKPyV viremia at the onset of HC with a median viral load of 1850 copies/mL (interquartile range, 240-8550). In multivariable models, high plasma viral load (≥10 000 copies/mL) and cytopenias at the beginning of BKPyV-HC were significantly associated with longer macroscopic hematuria and cystitis symptoms. Use of cidofovir was not associated with shorter duration of illness. In conclusion, BKPyV-HC after allogeneic HCT is characterized by prolonged and severe symptoms and requires improved management strategies. High-grade viremia and cytopenias were associated with a longer duration of BKPyV-associated HC. Accurate descriptions of disease and factors associated with prolonged recovery will inform end points of future clinical trials.

Graphical abstract

Figure 1.

Clinical presentation of BKPyV-HC on each study day by reported symptom. (A) Macroscopic hematuria. (B) Urinary clots. (C) Dysuria. (D) Frequency. (E) Urgency. (F) Abdominal pain. (G) Flank pain. Day 0 represents the diagnosis of BKPyV-HC, defined as the later of a positive urine BKPyV sample or the start of macroscopic hematuria. (H) A summary of proportion of patients reporting each symptom and median duration. If a patient died, then the symptom was only counted for the days they were alive.

Figure 2.

Characteristics of patients at time of diagnosis of BKPyV-HC. (A) Laboratory data, comorbidities, and medication use. (B) Presence of copathogens detected at time of diagnosis. (C) Specific clinical outcomes observed during their course. ^aCell counts are from the closest test within 14 days before or after beginning of BKPyV-HC and are measured in cells/µL. ^bIncluding diagnoses up to 90 days before or 7 days after beginning of BKPyV-HC. Grade is the peak grade of the episode. ^cIncludes cidofovir use up to 7 days before or after baseline date. Among the 19 patients on cidofovir, 4 (21%) had a concomitant diagnosis of adenovirus. ^dBedi score 2 indicates macroscopic hematuria, and 3 indicates presence of clots. ^ePain medication use includes opiates (n = 74), oxybutynin (n = 46), or phenazopyridine (n = 72). ^fTransfusion includes red blood cell (n = 92) or platelet (n = 79) transfusions. ALC, absolute lymphocyte count; AMC, absolute monocyte count; ANC, absolute neutrophil count.

Figure 3.

Incidence of disease resolution and mortality over the study period. Cumulative incidence (red) of symptom resolution (A) and hematuria resolution (B) plotted with the proportion of the population who died with ongoing symptoms or hematuria (blue). Among patients who died with ongoing BKPyV-HC, causes of death in this cohort included bleeding events (n = 2), hepatic failure due to vanishing bile duct syndrome (n = 1), Clostridioides difficile colitis (n = 1), GVHD (n = 1), pneumonia (n = 3), acute heart failure (n = 1), and sepsis related to cholecystitis (n = 1). No autopsies were done among this group of patients. At any time point, the height over the blue line represents the proportion of patients who have died during their BKPyV-HC episode (a), the height between blue and red lines represents the proportion of patients who are alive with ongoing symptoms/hematuria (b), and the height under the red line represents the proportion of patients whose symptoms/hematuria have resolved (c).

Figure 4.

Results of multivariable models. For the models describing factors associated with time to macroscopic hematuria resolution or time to resolution of all cystitis symptoms, an HR <1 is associated with lower likelihood of resolution of disease at any given time point and thus a longer duration of disease. All cell counts reported in cells/µL. ^aMultivariable model is adjusted for CMV serostatus and duration of cystitis symptoms prior to beginning of BKPyV-HC. Although CMV serostatus was included in the final model, detection of CMV viremia was not significantly associated with resolution of macroscopic hematuria (supplemental Table 1A). ^bMultivariable model is adjusted for duration of cystitis symptoms prior to beginning of BKPyV-HC. ^cMultivariable model is adjusted for duration of cystitis symptoms prior to beginning of BKPyV-HC, HLA match, and diagnosis of chronic GVHD.

Figure 5.

Multivariable GEE model examining the relationship between evolution of symptoms and viral load in the subset of patients who had regular BKPyV viral load testing. (A) Multivariable GEE models examining difference in mean viral loads at different symptomatic stages of BKPyV-HC. (B) Multivariable and univariable GEE models examining differences in mean viral loads. ^aMultivariable models were adjusted for transplant cell source, underlying disease, baseline absolute lymphocyte count, baseline absolute monocyte count, race, and diagnosis of chronic GVHD.