Safety and efficacy of intravenously administered cidofovir in adult haematopoietic cell transplant recipients: a retrospective multicentre cohort study

Abstract

Objectives

To evaluate the safety and efficacy of cidofovir for the treatment of double-stranded DNA (dsDNA) viral infections following allogeneic haematopoietic cell transplant (HCT).

Methods

This was a retrospective multicentre cohort study including adult HCT recipients who received ≥1 dose of IV-administered cidofovir for any dsDNA viral infection from 2006 to 2019. The objectives were to describe the rate of and risk factors for nephrotoxicity and virological response by the end of treatment (EOT).

Results

We included 165 patients from nine centres. Cidofovir was administered at 5 mg/kg/week (N = 115; 69.7%), 1 mg/kg/week (18; 10.9%), 3 mg/kg/week (12; 7.3%) or 1 mg/kg three times/week (11; 6.7%). Cidofovir was administered for adenovirus, cytomegalovirus (CMV) and BK virus infection in 75 (45.5%), 64 (38.8%) and 51 (30.9%) patients, respectively. Among 158 patients with renal function data at baseline and EOT, 40 (25.3%) developed nephrotoxicity. In multivariable analyses, age (OR 1.04; P = 0.05), weight (OR 1.05; P = 0.01), CMV infection (OR 3.6; P = 0.02), liposomal amphotericin B (OR 8.06; P = 0.05) and IV voriconazole/posaconazole (OR 13.0; P = 0.003) were predictors of nephrotoxicity. Creatinine concentration was significantly higher at EOT (1.16 ± 0.95 mg/dL) compared with baseline (0.91 ± 0.39 mg/dL; P < 0.001), but improved by 2 weeks (0.91 ± 0.84 mg/dL; P = 0.007) and 4 weeks (0.96 ± 0.89 mg/dL; P = 0.03) post-EOT. Median viral load significantly declined for patients with adenovirus DNAaemia by EOT (P < 0.0001) and for patients with CMV DNAaemia by EOT + 4 weeks (P = 0.003), but not for patients with BK virus DNAaemia.

Conclusions

One in four HCT recipients treated with IV cidofovir developed largely reversible nephrotoxicity. Careful selection of patients and close follow-up of renal function may minimize toxicity.

Introduction

Double-stranded DNA (dsDNA) viral infections contribute to substantial morbidity and mortality in allogeneic haematopoietic cell transplant (HCT) recipients.¹ Clinically relevant dsDNA viruses include herpesviruses, adenovirus and polyomaviruses (BK virus and JC virus). Cidofovir is a monophosphate nucleotide analogue of cytosine that inhibits viral DNA polymerase and exhibits in vitro activity against various dsDNA viruses.² Cidofovir has been used for the treatment of adenovirus, BK virus and cytomegalovirus (CMV) infections in high-risk allogeneic HCT recipients.^3,4

The major dose-limiting toxicity of cidofovir is nephrotoxicity and the emergence of a Fanconi-type syndrome.^5–7 The reported rates of cidofovir-associated nephrotoxicity in adult allogeneic HCT recipients have ranged between 25% and 33%.^8–10 In this retrospective, multicentre study, we sought to describe the use of cidofovir in adult allogeneic HCT recipients, including indication for administration, timing and dosing, the frequency of nephrotoxicity and virological response associated with cidofovir administration.

Materials and methods

We performed a retrospective multicentre international cohort study including all adult (≥18-year-old) allogeneic HCT recipients who received ≥1 dose of IV cidofovir at any time post-HCT for any dsDNA viral infection from 2006 to 2019. Patients who received intravesicular cidofovir were not included. Nine centres participated: four from the USA; four from Europe; and one from Australia.

Ethics

The local Ethics Committees at each institution approved the study.

Data collection

Patients were identified through the HCT and pharmacy databases at each participating centre. Infectious disease or haematology specialists at each centre reviewed charts from identified patients. Data were initially entered in case-report forms (CRFs) at each centre in a de-identified manner and subsequently into a master database for further analyses. The following data were collected: (a) patient demographics, including age, gender, ethnicity and weight; (b) HCT-related variables, including type of transplant, conditioning regimen, graft-versus-host disease (GvHD) at cidofovir initiation and associated treatments; (c) cidofovir-related variables, including indication, dose, frequency and probenecid administration; (d) renal function (creatinine and CL_CR) between baseline and 4 weeks post-cidofovir end of treatment (EOT); (e) prior and concomitant administration of potentially nephrotoxic medications and concomitant bacteraemia; and (f) outcome-related variables, including virological response, relapse of viral disease and mortality within 4 weeks post-EOT. For patients with DNAaemia at baseline, viral loads in plasma or whole blood were collected, at baseline and up to 4 weeks post-EOT.

Study objectives

The primary objective was to determine the proportion of patients treated with IV cidofovir with acute kidney injury (AKI) at EOT. As secondary objectives, we described the proportion of patients with AKI at 2 and 4 weeks post-EOT, predictors of AKI, virological response and overall survival of patients treated with cidofovir.

Definitions

Renal function

Renal function changes were defined based on the RIFLE (risk, injury, failure, loss and end-stage kidney disease) criteria.¹¹ More specifically, AKI was defined as an increase in serum creatinine of ≥1.5 times baseline and/or a decrease in CL_CR by >25%.¹¹ Creatinine was reported in mg/dL and μmol/L in seven and two centres, respectively. All μmol/L creatinine values were converted to mg/dL values for the purposes of this study. CL_CR was recorded as estimated glomerular filtration rate (eGFR) in mL/min/m² at all centres. The eGFR range was reported as a full-range absolute number in four centres and as an absolute number from 0–60 and 0–90 mL/min/m² and then as >60 and >90 mL/min/m² in three and two centres, respectively. Frequency of IV cidofovir administration was reported as once weekly, three times weekly and once every other week. For patients who received one dose of cidofovir, treatment frequency was considered as one weekly dose. Due to large variability, doses were rounded up to the closest absolute value as follows: 0.5–1.5 rounded to 1 mg/kg; 2.5–3.9 rounded to 3 mg/kg; and 4.0–5.9 rounded to 5 mg/kg. The average weekly cidofovir dose per patient during the study period was calculated by adding all cidofovir doses administered per patient divided by the number of weeks of treatment. The cumulative cidofovir dose was calculated by adding the weekly doses of cidofovir administered per patient.

Infection diagnosis and treatment response

Viral infection and disease were defined based on established guidelines.^12–14 DNAaemia was reported as copies/mL in whole blood or plasma for all viruses based on individual laboratory practices. CMV DNAaemia was reported in IU/mL after 2012 in most centres. DNAaemia measurement parameters (whole blood versus plasma and copies/mL versus IU/mL) for individual patients included in this study remained the same during the study follow-up. Virological response was defined as ≥1 log decrease in viral load at EOT compared with baseline in patients with documented DNAaemia.

Statistical analysis

Standard descriptive statistics were used to summarize population characteristics. We used the chi-squared test for categorical variables, paired Student’s t-test for normally distributed continuous variables and Wilcoxon signed-rank test for non-parametric variables. Continuous variables were presented as mean with SD and range, and as median with IQR, as deemed appropriate. We performed logistic regression to identify risk factors for nephrotoxicity at EOT. Results were presented as ORs with 95% CI. Variables with P < 0.10 in univariable analyses were included in the multivariable models using forward stepwise logistic regression. We used the Pearson correlation coefficient to determine possible correlations between independent variables. The overall 4 week mortality post-EOT was analysed using Kaplan–Meier survival curves. A two-sided P < 0.05 was considered statistically significant for all tests. Statistical analysis was performed using STATA 16.0 (StataCorp, College Station, TX, USA).

Results

We included 165 adult allogeneic HCT recipients treated with IV cidofovir (Table 1). The median age was 50 years (IQR 39–60) and 66 patients (40%) were female. Patients received an allogeneic HCT from an HLA-matched related donor (28; 17.1%), an HLA-matched/mismatched unrelated donor (107; 65.2%) or a haploidentical donor (29; 17.7%). In 23 (14%) patients, HCT source was cord blood and 89 (56.3%) were T cell depleted. The most common indication for cidofovir treatment was adenovirus infection in 75 (45.5%) patients, followed by CMV (64; 38.8%), BK virus (51; 30.9%), human herpesvirus 6 (HHV-6) (6; 3.6%), herpes simplex virus (HSV) (6; 3.6%) and JC virus (1; 0.6%) infections. Thirty-two (19.4%) patients had more than one viral infection. Details on viral infections are presented in Table S1, available as Supplementary data at JAC Online.

Table 1.

Baseline patient characteristics

Variables	n (%)
	N = 165
Demographics
Age (years), median (IQR)a	50 (39–60)
Gender, male	99 (60)
Ethnicity
White	95 (57.6)
Black or African-American	13 (7.9)
Asian	10 (6)
Hispanic	15 (9.1)
Other/unknown	32 (19.4)
Weight (kg), median (IQR)b	68 (60–75)
HCT characteristics
HCT yearc
2006–12	52 (31.7)
2013–19	112 (68.3)
HLA matchingc
Matched related donor	28 (17.1)
Matched/mismatched unrelated donor	107 (65.2)
Haploidentical donor	29 (17.7)
HCT sourced
Bone marrow	15 (9.3)
Peripheral blood stem cells	124 (76.5)
Cord blood	23 (14.2)
T cell depletione	89 (56.3)
T cell depletion type
Alemtuzumab	1 (1.1)
CD34 selection	49 (55.1)
Anti-thymocyte globulin	39 (43.8)
Conditioningc,f
Cyclophosphamide	76 (46.3)
Melphalan	59 (36)
Busulfan	49 (29.9)
Fludarabine	124 (75.6)
Total body irradiation	70 (42.4)
GvHD characteristicsg
Acute, ≥grade 2	63 (38.9)
Chronic	22 (13.6)
Corticosteroid treatment ≥1 mg/kg	55 (33.9)
Indication for CDV treatmenth
>1 viral infection	32 (19.4)
Adenovirus infection	75 (45.5)
BK virus infection	51 (30.9)
CMV infection	64 (38.8)
HHV-6 infection	6 (3.6)
HSV infection	6 (3.6)
JC virus infection	1 (0.6)

CDV, cidofovir.

^aAge was available for 153 patients.

^bWeight was available for 161 patients.

^cHCT year, HLA-matching and conditioning data were available for 164 patients.

^dHCT source was known for 162 patients.

^eT cell depletion was known for 158 patients.

^fMore than one agent was used concomitantly or sequentially.

^gData on GvHD were available for 162 patients. GvHD data are recorded at the time of cidofovir initiation.

^h30 patients had two viral infections and two patients had three and four viral infections each. Detailed information on viral infections is presented in Table S1.

Cidofovir administration

IV-administered cidofovir was initiated at a median of 91 days (IQR 60–163) after HCT at an average weekly dose of 3.5 mg/kg for a median of 3 doses (IQR 2–5) and an overall duration of 3 weeks (IQR 2–5; Table 2). Almost one-third of patients (53; 32%) received cidofovir for longer than 4 weeks. Cidofovir was initiated at 5 mg/kg once weekly (115; 69.7%), 1 mg/kg once weekly (18; 10.9%), 3 mg/kg once weekly (12; 7.3%) or 1 mg/kg three times weekly (11; 6.7%). Thirty-three (20%) patients received a single dose of cidofovir. Dose changes were required in 40 (24.2%) patients. The median cumulative dose was 10 mg/kg (IQR 5–20). Cumulative dose was significantly higher for patients who received cidofovir for adenovirus (10.2 mg/kg; IQR 6–20) and/or CMV infection (12 mg/kg; IQR 9.9–20) compared with those whose sole indication was BK virus infection (5 mg/kg; IQR 2.7–10; P < 0.001 and P < 0.0001, respectively). Probenecid was administered concomitantly with cidofovir in 147 (89.1%) patients. Out of the 18 patients who did not receive probenecid, the indication for cidofovir was BK virus cystitis in 14 patients and adenovirus infection in 4 patients.

Table 2.

Cidofovir treatment administration data in the overall patient population

Variables	n (%)
	N = 165
Time from HCT to CDV start (days), median (IQR)	91 (60–163)
Duration of CDV treatment (weeks), median (IQR)	3 (2–5)
1–4a	112 (67.9)
5–8	37 (22.4)
>8	16 (9.7)
CDV dosing
Number of doses, median (IQR)	3 (2–5)
Dose per week (mg/kg), median (IQR)b	3.5 (3–5)
Cumulative dose (mg/kg), median (IQR)b	10 (5–20)
CDV administration
Dose and frequency (first CDV dose)
5 mg/kg once weekly	115 (69.7)
3 mg/kg once weekly	12 (7.3)
1 mg/kg once weekly	18 (10.9)
1 mg/kg three times weekly	11 (6.7)
5 mg/kg once every other week	9 (5.4)
Single dose	33 (20)
5 mg/kg once weekly	27 (81.8)
3 mg/kg once weekly	3 (9.1)
1 mg/kg once weekly	3 (9.1)
Multiple doses	132 (80)
At least one change in dose/frequency	40 (30.3)
Initial dose if ≥1 change in dose/frequency
5 mg/kg once weekly	33 (82.5)
3 mg/kg once weekly	2 (5)
1 mg/kg once weekly	2 (5)
5 mg/kg once every other week	3 (7.5)
No changes in dose/frequency	92 (69.7)
Dose if no changes in dose/frequency
5 mg/kg once weekly	55 (59.8)
3 mg/kg once weekly	8 (8.7)
1 mg/kg once weekly	11 (12)
1 mg/kg three times weekly	12 (13)
5 mg/kg once every other week	6 (6.5)
Probenecid	147 (89.1)

CDV, cidofovir.

^aIn the category of 1–4 weeks of CDV treatment, 29 patients who received a single dose of CDV were also included, considered to have received 1 week treatment.

^bThe average weekly cidofovir dose per patient during the study period was calculated by adding all cidofovir doses administered per patient during the study period divided by the number of weeks of treatment. For patients who received cidofovir every other week, 2 weeks of treatment were included. Cumulative cidofovir dose represents the overall mg/kg of cidofovir administered per patient during the study period.

AKI during cidofovir administration

There were 158 patients with available renal function data at baseline and EOT. Therefore, for all renal function data analyses, only those 158 patients were included. Among those patients, cidofovir was administered as a single dose in 29 patients (18.4%) at 5, 3 and 1 mg/kg/week in 24, 3 and 2 patients, respectively. In the remaining 129 (81.6%) patients, the initial cidofovir dose was 5, 3 and 1 mg/kg/week in 85, 38 and 16 patients, respectively, for a median of 3 doses (IQR 2–5; Table S2).

Forty of 158 (25.3%) patients developed AKI using the RIFLE criteria. Creatinine concentration increased from a mean of 0.91 mg/dL (SD 0.39; range 0.3–2.9) at baseline to 1.06 mg/dL (SD 0.75; range 0.2–6.2) by Week 1 (P = 0.003); 1.09 mg/dL (SD 0.73; range 0.3–4.5) by Week 2 (P = 0.003); and 1.16 mg/dL (SD 0.95; range 0.1–5.2) by EOT (P < 0.001) (Figure 1a and b). Compared with EOT, creatinine concentration significantly improved, returning close to baseline levels by 2 weeks post-EOT, with a mean of 0.91 mg/dL (SD 0.84; range 0.38–5.7; P = 0.007), and by 4 weeks post-EOT, with a mean of 0.96 mg/dL (SD 0.89; range 0.4–6.7; P = 0.03).

Figure 1.

Renal function changes during cidofovir treatment administration. (a) Serum creatinine between baseline and Week 8 of cidofovir administration; (b) serum creatinine between baseline, EOT and 2 and 4 weeks after the end of cidofovir administration; (c) eGFR between baseline and Week 4 of cidofovir administration; and (d) eGFR values between baseline, EOT and 2 and 4 weeks after the end of cidofovir administration. Only significant P values are shown; P values were calculated using paired t-test. Wk, Week.

Sixty-four patients had available eGFR values at baseline and EOT. Overall, mean eGFR significantly decreased from 82.2 mL/min/m² (SD 30.9; range 34–171) at baseline to 70.6 mL/min/m² (SD 35.6; range 9–216; P = 0.01) at Week 1, 72.6 mL/min/m² (SD 34.8; range 13–176; P = 0.01) at Week 2 and 68.8 mL/min/m² (SD 37.3; range 12–216; P = 0.005) at EOT (Figure 1c and d). Compared with EOT, eGFR levels remained approximately the same at 2 weeks post-EOT, with a mean of 64.5 mL/min/m² (SD 36.7; range 13–154; P = 0.21) and at 4 weeks post-EOT, with a mean of 66.9 mL/min/m² (SD 35.8; range 13–168; P = 0.52). Among the 40 patients with AKI by EOT, data on kidney function by 4 weeks post-EOT were available for only 17 (42.5%) patients. Among those 17 patients, 12 (70.6%) had ongoing impaired renal function 4 weeks after discontinuing cidofovir.

When evaluating all patients with available eGFR values (including absolute values and values reported as >60 or >90 mL/min/m²), 123 patients had available eGFR data at baseline (64 patients with absolute eGFR numbers and 59 with values reported as >60 or >90 mL/min/m²) and 107 patients at EOT (66 with absolute numbers and 41 with >60 or >90 mL/min/m²). At baseline, 98 (79.7%) patients had an eGFR of ≥60 mL/min/m², 24 patients (19.5%) had an eGFR between 30 and 59 mL/min/m² and one patient (0.8%) had an eGFR of <30 mL/min/m². At EOT, 74 (69.2%) patients had an eGFR of ≥60 mL/min/m², followed by 26 patients (24.3%) with an eGFR between 30 and 59 mL/min/m² and 7 (6.5%) patients with an eGFR of <30 mL/min/m² (P = 0.003). Among the seven patients with an eGFR of <30 mL/min/m² at EOT, the mean eGFR was 21.1 mL/min/m² (range 12–29). No patients required haemodialysis by EOT.

There was no significant difference in the baseline creatinine concentration between patients who did (median 0.90 mg/dL; IQR 0.68–1.1) and those who did not (median 0.82 mg/dL; IQR 0.62–1.2) develop AKI (P = 0.49). Similarly, there was no difference between the baseline eGFR for patients with (median 71 mL/min/m²; IQR: 55–102) and without (median 71 mL/min/m²; IQR 34–150) AKI by EOT (P = 0.51). Renal function changes during cidofovir administration in patients with and without AKI are presented in Figure 2. For patients without AKI at EOT, renal function did not significantly change during the first 4 weeks of cidofovir administration (Figure 2a and c). In contrast, in patients with AKI at EOT, creatinine concentration significantly increased from baseline (mean 0.95 mg/dL; SD 0.44; range 0.3–2.9) to Week 1 (mean 1.45 mg/dL; SD 1.17, range 0.4–6.2; P = 0.002), Week 2 (mean 1.72 mg/dL; SD 1.12; range 0.5–4.5; P = 0.0005) and Week 4 (mean 1.37 mg/dL; SD 0.71; range 0.5–2.75; P = 0.002) (Figure 2a). Creatinine concentration significantly increased from baseline to EOT (mean 2.15 mg/dL; SD 1.27; range: 0.5–5.16; P < 0.0001) (Figure 2b). Similar patterns were observed in eGFR between baseline and Weeks 1–4 and EOT for the two groups (Figure 2c and d).

Figure 2.

Renal function changes during cidofovir treatment administration presented separately for patients with and without AKI at EOT. (a) Serum creatinine between baseline and Week 8 of cidofovir administration; (b) serum creatinine between baseline, EOT and 2 and 4 weeks after the end of cidofovir administration; (c) eGFR between baseline and Week 4 of cidofovir administration; and (d) eGFR values between baseline, EOT and 2 and 4 weeks after the end of cidofovir administration. Only significant P values are shown; P values were calculated using paired t-test. Wk, Week.

Risk factor analyses for AKI in patients treated with cidofovir

To better define the patient population who develop AKI during cidofovir treatment, risk factor analyses for AKI at EOT were performed (Table S3). Multivariable analyses demonstrated the following independent variables as significant risk factors for AKI in allogeneic HCT recipients treated with cidofovir: age (OR: 1.04; 95% CI 1.02–1.1; P = 0.05), weight (OR: 1.05; 95% CI 1.02–1.1; P = 0.01), CMV infection (OR: 3.6; 95% CI 1.3–10; P = 0.02) and administration of liposomal amphotericin B (OR: 8.06; 95% CI 1–63; P = 0.05) or IV voriconazole/posaconazole (OR: 13.0; 95% CI 2.4–69; P = 0.003) at the time of cidofovir initiation.

Virological response and mortality

We evaluated virological response in patients who had ≥1 type of DNAaemia with quantifiable PCR measurements at baseline. We included 128 DNAaemia episodes from 113 patients, including 53 episodes of adenovirus, 48 episodes of CMV, 22 episodes of BK virus and 5 episodes of HHV-6 DNAaemia. Virological response to cidofovir, defined as ≥1 log reduction of viral load from baseline to EOT, was achieved in 51 (40%) DNAaemia episodes including 38 (72%) episodes of adenovirus infection, 8 (17%) episodes of CMV infection and 5 (23%) episodes of BK virus infection. None of the five patients with HHV-6 DNAaemia had virological response at EOT. For patients with adenovirus DNAaemia, the median viral load at EOT (891 copies/mL) was significantly lower compared with baseline (48 978 copies/mL; P < 0.0001). No significant differences were noted in median viral loads for patients with CMV and BK virus DNAaemia between baseline and EOT (Figure 3). However, 4 weeks after EOT, the median viral load for CMV was significantly lower compared with baseline (137 versus 5700 copies/mL, respectively; P = 0.003).

Figure 3:

Viral load (log) values at baseline (first cidofovir dose), EOT and 2 and 4 weeks after EOT for: (a) all viraemia episodes (N = 128); (b) adenovirus (N = 53); (c) BK virus (N = 22); and (d) CMV viraemia episodes (N = 48). P values were calculated using the Wilcoxon signed-rank test. Data on HHV-6 DNAaemia are not presented, due to the small number of patients.

There were 43/165 (26.1%) patients who died within 28 days after administration of the last dose of cidofovir. Mortality was significantly higher among patients with AKI at EOT (15/40; 37.5%) compared with those without AKI (28/118; 23.7%; log-rank test: 0.02; Figure 4).

Figure 4.

Kaplan–Meier survival curve for the first 56 days after discontinuation of cidofovir administration. This figure appears in colour in the online version of JAC and in black and white in the print version of JAC.

Discussion

In this retrospective multicentre cohort study, one in four adult allogeneic HCT recipients treated with IV cidofovir developed mostly reversible AKI at EOT, consistent with prior reported rates of nephrotoxicity associated with cidofovir.^7,9,15 A multicentre survey by the Infectious Diseases Working Party of the European Group for Blood and Marrow Transplantation reported some degree of renal toxicity in 31% of adult and paediatric HCT recipients who received cidofovir for adenovirus infection and 25% in HCT recipients treated with cidofovir for CMV infection.⁹ In a recently published case series from Johns Hopkins University, nephrotoxicity was reported in 37.5% of adult HCT and solid organ transplant recipients with resistant or refractory CMV infection treated with IV cidofovir.¹⁵ Comparing rates of nephrotoxicity across different studies is hindered by different patient populations, types and time of HCT, and the definitions of renal toxicity used. Furthermore, assessing the nephrotoxicity of cidofovir retrospectively may underestimate the real effect of this agent on the kidneys, as clinicians might have discontinued its administration based on mild changes in renal function parameters.

The latter may, in part, explain the mild to moderate AKI observed in our study, with an EOT median creatinine concentration of 1.16 mg/dL and mean eGFR of 68.8 mL/min/m². Based on creatinine measurements, renal function was back to baseline by 2 and 4 weeks after cidofovir EOT. This observation is in contrast to foscarnet administration in HCT or solid organ transplant recipients, which has been associated with an eGFR decrease of >20% in over 50% of patients, with almost one-third of patients maintaining renal dysfunction after 6 months from EOT and 8% requiring renal replacement therapy.¹⁶ Only seven (6.5%) patients had an eGFR <30 mL/min/m² at EOT in our series and no patient required haemodialysis. It is possible that some of the patients who developed significant nephrotoxicity either had cidofovir prematurely discontinued or died before they required dialysis. This information was not available and hence no further conclusions could be drawn.

Patients who did not develop AKI by cidofovir EOT in our study had stable renal function parameters throughout their treatment course. In contrast, patients who developed AKI had a significant rise in creatinine and/or drop in eGFR as early as 1 week after cidofovir initiation. The above suggest that detection of early renal function impairment after cidofovir treatment initiation may alert clinicians to potential development of AKI and the need for closer renal function monitoring. We tried to better describe the group of patients who were more likely to develop AKI during cidofovir administration. Multivariable analyses suggested that nephrotoxicity was directly dependent on patients’ age and weight. Both older age and higher body mass have been previously associated with drug-induced nephrotoxicity from several other known nephrotoxic medications, including vancomycin, polymyxin B and cyclosporine.^17–21 For instance, weight over 100 kg and age over 52 years have been reported as independent risk factors for vancomycin-associated nephrotoxicity.²⁰ Similarly, age over 60 years and obesity were independently associated with increased creatinine levels in patients with psoriasis treated with cyclosporine.²¹ These results may have clinical implications as administration of cidofovir in younger patients with lower body weight may be safer, while closer monitoring may be required for older patients with higher weights requiring higher doses. CMV infection was associated with a higher risk for AKI. Renal toxicity has been reported among patients treated with ganciclovir and/or foscarnet for CMV.²² Furthermore, CMV infection has been associated with poor clinical outcomes in HCT recipients.²³ Considering that cidofovir is commonly a third-line treatment for CMV infection, it is likely that this association is the cumulative effect of the overall direct and indirect effect of CMV infection and its associated treatments on renal function. Administration of liposomal amphotericin B and IV voriconazole/posaconazole was identified as a significant predictor of AKI in our study. This may be explained, in part, by the direct nephrotoxicity of amphotericin B and/or possible nephrotoxicity associated with cyclodextrin-containing formulations of IV voriconazole/posaconazole, but also the overall status of patients with an invasive mould infection. Notably, none of the HCT- or cidofovir-associated variables, baseline renal function or prior administration of foscarnet were associated with nephrotoxicity in this series.

Virological response differed according to the indication of cidofovir administration. While 72% of adenovirus DNAaemia episodes resulted in virological response, only 17% of CMV episodes and 23% of BK DNAaemia episodes achieved ≥1 log reduction of viral load at EOT. These results are consistent with prior studies showing favourable results with cidofovir for adenovirus infection following HCT.^24–26 Limited data exist for adenovirus in adult HCT recipients, with response rates ranging between 68% and 80%, with immune reconstitution being critical for sustained responses.^4,7 A significant virological response was not demonstrated by EOT for patients with CMV infection, most likely due to lack of relevant data for a significant number of patients. In contrast, virological response was demonstrated by 4 weeks after EOT in patients with CMV infection, perhaps reflecting the long-term effect associated with the long half-life of this agent. Similarly, a recent study evaluating cidofovir treatment for ganciclovir-resistant/refractory CMV showed clearance of DNAaemia in 50%.¹⁵ Cidofovir has been used for the treatment of BK virus haemorrhagic cystitis after HCT, with reported clinical and virological efficacy in the range of 66%–86%.^8,10,27–29 Due to the small number of patients with BK virus and HHV-6 infection included in our study and lack of consistent DNAaemia measurements, further conclusions cannot be drawn on the efficacy of this agent for the treatment of these infections.

Our study has several limitations associated with its retrospective design and lack of controls, precluding assessment of the true contribution of cidofovir to renal toxicity. Our cohort was highly heterogeneous in terms of transplant characteristics, cidofovir administration and types and severity of viral infection. Moreover, renal function measurement and reporting and DNAaemia testing (plasma versus blood, molecular assays used, copies versus IU/mL) varied quite significantly across centres. Data on other adverse events associated with cidofovir, namely uveitis, were not included. Finally, the large variability of doses of cidofovir administered, including rather lower dosing regimens, such as 1 mg/kg/week, might have biased the observed outcomes. However, risk factor analyses failed to show any definitive statistically significant associations between cidofovir administration variables and AKI by EOT. While acknowledging these limitations, the major strength of this study lies in the large sample size and systematic and detailed data collection, which provides the largest real-world assessment of nephrotoxicity of cidofovir across diverse centres and standards of care.

In conclusion, our data suggest that cidofovir treatment was associated with reversible nephrotoxicity in 25% of HCT recipients and that careful selection of patients based on their age, weight and concomitant administration of mould-acting antifungal treatments with close renal function monitoring may be important factors to consider in minimizing associated toxicity. Notably, patients with a significant renal dysfunction development after 1–2 weeks of cidofovir administration may be more likely to develop AKI and warrant closer follow-up and monitoring. As cidofovir is currently the only readily available pharmacological option for multiple dsDNA virus infections, including adenovirus and refractory CMV infections, our study highlights the unmet need for optimal management of high-risk patients with dsDNA virus infection addressing safety, efficacy and also pharmacokinetics.