Reasons for voriconazole prophylaxis discontinuation in allogeneic hematopoietic cell transplant recipients: A real-life paradigm

Abstract

We sought to describe the clinical experience of voriconazole as primary antifungal prophylaxis (AFP) in allogeneic hematopoietic cell transplant recipients (allo-HCTr). This was a single-center retrospective study of adult allo-HCTr (1 January 2014 to 31 December 2016) who received ≥two doses of voriconazole-AFP. Voriconazole-AFP was started on day +7 post-HCT and continued at least through day +60 post-HCT, or longer as clinically indicated. We reviewed the rate, reasons, and risk factors of voriconazole-AFP discontinuation until day-100 post-HCT. A total of 327 patients were included. Voriconazole-AFP was continued for a median of 69 days (mean: 57.9; range 1, 100): for a median of 90 days (mean :84; range 2, 100) in 180/327 (55%) in the standard-of-care (SOC) group and 20 days (mean :25.6 ; range 1, 89; P-value < .001) in 147/327 (45%) patients in the early-discontinuation-group. Early-voriconazole-AFP discontinuation was due to adverse events, drug interactions, insurance coverage, and other reasons in 101/147 (68.7%), 27 (18.4%), 13 (8.8%), and 6 (4.1%) patients, respectively. Early-voriconazole-AFP discontinuation occurred in 73/327 (22.3%) patients due to hepatotoxicity. Important predictors for early-voriconazole-AFP discontinuation included: graft-versus-host disease grade ≥2 (odds ratio [OR]: 1.9, P-value: .02), alanine-aminotransferase ≥75 IU/ml on voriconazole-administration day-14 (OR: 5.6, P-value: .02) and total bilirubin ≥1.3 mg/dl on voriconazole-administration day-7 (OR: 3.0, P-value: .03). There were 13 proven/probable invasive fungal infections by day-180 post-HCT (8/147, 5.4%, and 5/180, 2.8% in the early-discontinuation and SOC-groups, respectively; log-rank:0.13). By day-180 post HCT, 23/147 (15.6%) and 14/180 (7.8%) patients in the early-discontinuation and SOC-groups had died, respectively (log-rank:0.03). Voriconazole-AFP was discontinued in up to 45% of allo-HCTr. Hepatotoxicity during the first 2 weeks post-HCT is a significant predictor of early-voriconazole-AFP discontinuation.

Introduction

Primary antifungal prophylaxis (AFP) with fluconazole, micafungin, or voriconazole is recommended during the first 75 days after a hematopoietic cell transplant (HCT), while posaconazole is the preferred AFP in cases of severe graft-versus-host disease (GvHD).^1–4 However, the use of voriconazole as AFP in allogeneic HCT recipients may be hindered due to variable pharmacokinetics, interpersonal variability and drug-drug interactions.^5–8 In addition, voriconazole-associated adverse events (AEs), including neurotoxicity, visual abnormalities and, most importantly, liver function impairment are frequently encountered in allogeneic HCT recipients.^9–13 Moreover, we have previously demonstrated that premature discontinuation of voriconazole AFP may occur during the early post-transplantation period and before the cessation of immunosuppression in up to one third of patients due to, predominately, liver toxicity.¹⁴ The aim of this study was to describe the clinical experience of voriconazole as primary AFP in allogeneic HCT recipients in a cancer center. More specifically, we sought to describe the rates and risk factors for discontinuation of voriconazole AFP during the first 75 days after an allogeneic HCT.

Methods

Study design

This was a single-center retrospective observational study of all adult (≥18 years) allogeneic HCT recipients between 1 January 2014 and 31 December 2016 at Memorial Sloan Kettering Cancer Center (MSKCC), who received a minimum of two doses of voriconazole as primary AFP post-HCT with available liver function tests at baseline and during voriconazole administration. Patients who did not receive voriconazole due to allergy, previous AEs or other reasons, or had a proven/probable invasive fungal infection (IFI) within 60 days prior to their HCT, were excluded. The study was approved by the MSKCC Institutional Review Board.

Antifungal prophylaxis strategy

Per institutional standard of care (SOC), an echinocandin (e.g., micafungin) was administered as primary AFP to all allogeneic HCT recipients starting 2 days (day -2) before until day +7 post-HCT or steady state levels of calcineurin inhibitors, when AFP was switched to voriconazole.⁷ Voriconazole was continued until at least day +75 after HCT or cessation of immunosuppression. Of note, voriconazole therapeutic drug monitoring was not routinely performed at our institution during the study period.

Data collection

Demographics, underlying hematologic malignancy, type of transplant (unmodified vs T-cell depleted vs umbilical cord), stem cell source (bone marrow [BM] vs peripheral blood stem cells [PBSC] vs cord blood), donor-recipient matching, and conditioning regimen were recorded. Incidence of probable or proven IFI from day 0 until day +180 from HCT and voriconazole administration details (voriconazole initiation day, duration of AFP, mode of administration, loading and maintenance dose, AEs, and reasons of voriconazole discontinuation) were recorded. Liver function tests were recorded, including alanine transaminase (ALT), alkaline phosphatase and total bilirubin, before the start of voriconazole (baseline), during the period of voriconazole AFP, and 2 and 4 weeks after the end of voriconazole administration were recorded. All-cause mortality was also recorded.

Definitions

IFIs were defined based on the European Organization for Research and Treatment of Cancer/Mycoses Study Group (EORTC/MSG) consensus guidelines.¹⁵Ex vivo T-cell depletion was accomplished by positive selection of CD34⁺ stem cells by using a CliniMACS CD34 Reagent system.¹⁶ Acute graft-versus-host disease (GvHD) ≥ grade 2 was defined based on established guidelines.¹⁷

Statistical analysis

Categorical variables were compared by the chi-square test while continuous variables were compared by a t-test. Risk factors for voriconazole discontinuation (i) at any time, (ii) by day +14 and (iii) by day +28 post-HCT were identified using logistic regression models. The following independent variables were evaluated in univariable analyses: demographics, underlying disease, HCT characteristics, GvHD prophylaxis, GvHD grade ≥2, cytomegalovirus (CMV) serostatus, voriconazole administration variables, and ALT and total bilirubin at baseline and by days 7, 14, and 28 post-voriconazole administration. Variables with a P-value <.10 in univariable analyses were subsequently introduced into multivariable logistic regression models in a stepwise fashion. P-values of <.05 were considered significant. All analyses were performed by Stata 12.

Results

Patient population

A total of 435 patients received an allogeneic HCT during the study period, of which 108 patients were excluded due to previous AEs with voriconazole (9/108, 8.3%), baseline liver impairment precluding voriconazole use (41/108, 38%), proven or probable IFI within 60 days prior to HCT (13, 12%) or after HCT before the intended date of voriconazole initiation (4, 3.7%), recommendation to use antifungal agents other than voriconazole by the infectious disease service (8, 7.4%), fear of voriconazole toxicity (5, 4.6%), potential drug-drug interaction (3, 2.8%), death before the intended date of voriconazole initiation (3, 2.8%), and other reasons (22, 20.4%).

A total of 327 patients were included in this study (Table 1). The median age was 55 years, and almost 80% of the patients were white. The most common underlying disease prior to HCT was acute myeloid leukemia (112/327, 34.3%), followed by lymphoma (66, 20.2%) and myelodysplastic syndrome (47, 14.4%). A total of 143 (43.7%) HCT were from a matched unrelated donor, while haploidentical HCT accounted for 18 (5.5%) patients. The majority (176, 53.8%) of patients received a reduced intensity conditioning regimen, while others received a myeloablative conditioning regimen (135, 41.3%).

Table 1.

Baseline patient characteristics.

	Total number of patients
Patient characteristic	N: 327 (%)
Demographics
Age, median years (IQR)	55 (42.7–64.5)
Sex, female	145 (44.3)
Ethnicity
White	259 (79.2)
Black	27 (8.3)
Asian	15 (4.6)
Other/Unknown	26 (7.9)
Underlying disease
Acute myeloid leukemia	112 (34.3)
Acute lymphocytic leukemia	39 (11.9)
Myelodysplastic syndrome	47 (14.4)
Multiple myeloma	30 (9.2)
Lymphoma	66 (20.2)
Otherα	33 (10)
HCT characteristics
Donor type
Matched related	82 (25.1)
Matched unrelated	143 (43.7)
Mismatched unrelated	84 (25.7)
Haploidentical	18 (5.5)
HCT source
Bone marrow	33 (10.1)
Peripheral blood stem cells	242 (74.0)
Cord blood	31 (9.5)
Cord blood / peripheral blood stem cells	21 (6.4)
HCT manipulation
Ex vivo T-cell depletion (CD34 + selected)	110 (33.6)
Conditioning
Myeloablative	135 (41.3)
Reduced intensity	176 (53.8)
Nonablative	16 (4.9)
GvHD prophylaxis
Tacrolimus + methotrexateβ	120 (36.7)
Cyclosporine + mycophenolate mofetil	52 (15.9)
Tacrolimus + mycophenolate mofetilγ	28 (8.6)
Tacrolimus + sethotrexate + sirolimus	12 (3.7)
Othersδ	5 (1.5)
Ex vivo T-cell depletion (CD34 + selected)	110 (33.6)
CMV serostatus
D+/R+	105 (32.1)
D+/R−	39 (11.9)
D−/R+	79 (24.2)
D−/R−	104 (31.8)

D, donor; CMV, cytomegalovirus; GvHD, graft versus host disease; HCT, hematopoietic cell transplant; IFI, invasive fungal infection; R: recipient.

^αOther underlying disease included: chronic leukemia (N: 16), myeloproliferative syndrome (N: 11), aplastic anemia (N: 2) and nonmalignant hematologic disorder (N: 4).

^βIn addition to tacrolimus + methotrexate, nine patients also received bortezomib, six patients also received maraviroc.

^γIn addition to tacrolimus + mycophenolate mofeti, three patients also received methotrexate, 22 patients also received cyclophosphamide.

^δIncluded three cyclophosphamide, one cyclophosphamide/mycophenolate mofetil/sirolimus, and one cyclophosphamide/sirolimus.

Voriconazole administration data

Voriconazole was started at a median of 7 days (mean: 8.6; range 1, 43) post-HCT and continued for a median of 69 days (mean: 57.9; range 1, 100) (Table 2). A loading dose was administered in 292 (89.3%) patients (administered intravenously in 270/292 patients, 92.5%) and maintenance dose was weight-based in 279 (85.3%) patients. Mode of administration varied in the cohort, with most patients treated with intravenous (IV) voriconazole during the first 3 days of administration, with transition to oral (PO) after the first week of treatment. A total of 180 (55%) patients continued voriconazole prophylaxis as per institutional SOC (SOC group) for a median of 90 days (mean: 84; range 2, 100). In 147 (45%) patients, voriconazole prophylaxis was discontinued prematurely (early-discontinuation group) at a median of 20 days (mean: 25.6; range 1, 89; t-test P-value <.001) (Fig. 1A). Among the 147 patients in the early-discontinuation group, 137 (93.2%) patients were started on another antifungal agent for primary antifungal prophylaxis: 65/137 (47.4%) on posaconazole, 54 (39.4%) on micafungin, 14 (10.2%) on fluconazole, 3 (2.2%) on isavuconazole, and 1 (0.8%) patient on liposomal amphotericin B. In 10/147 (6.8%) patients no additional antifungal prophylaxis was administered after discontinuation of voriconazole. Voriconazole was resumed in 27/147 (18.4%) of patients. Voriconazole trough blood levels were measured in only 34 (10.4%) patients once during the first 10 days of voriconazole administration. Voriconazole trough level was at a mean of 2.2 mg/l (range: 0.3, 7.8): 2.1 mg/l (range: 0.3, 4.9) in 19 and 2.3 mg/l (range: 0.3, 7.8) in 15 patients in the SOC and early-discontinuation groups, respectively (P-value: .65).

Table 2.

Voriconazole administration characteristics.

	Total number of patients
Voriconazole characteristics	N: 327 (%)
Time of voriconazole administration
Initiation, median days post-HCT (mean; range)	7 (8.6 ; 1, 43)
Total days of voriconazole administration, median (mean; range)	69 (57.9 ; 1, 100)
Voriconazole dosing
Loading dose given	292 (89.3)
Maintenance dose
200 mg twice daily	48 (14.7)
4 mg/kg twice daily	279 (85.3)
Voriconazole mode of administration
Loading dose, IV	270 (82.6)
Voriconazole administration, day 1–3α
IV (only)	237 (73.4)
PO (only)	51 (15.8)
IV/PO (administration changed from IV to PO or vice-versa)	35 (10.8)
Voriconazole administration, day 1–7β
IV (only)	95 (31.8)
PO (only)	48 (16.0)
IV/PO (administration changed from IV to PO or vice versa)	156 (52.2)
Voriconazole administration, day 1–14γ
IV (only)	37 (13.2)
PO (only)	42 (15)
IV/PO (administration changed from IV to PO or vice versa)	202 (71.8)

HCT, hematopoietic cell transplant, IV, intravenous, PO, oral; SD, standard deviation.

^αData were available for 323 patients overall.

^βData were available for 299 patients overall.

^γData were available for 281 patients overall.

Figure 1.

(A) Rate of voriconazole discontinuation in a cohort of allogeneic hematopoietic cell transplant (HCT) recipients. (B) Presentation of proportions of patients with discontinuation of voriconazole prophylaxis and the reasons for discontinuation. (C) Rate of voriconazole discontinuation based on the alanine aminotransferase (ALT) level by post-HCT day 14.

Voriconazole discontinuation

Among the 147 patients in the early-discontinuation group, voriconazole was prematurely discontinued in 101 (68.7%) patients due to AEs, followed by drug-drug interactions in 27 (18.4%) patients and other reasons in six (4.1%) patients: three patients had suboptimal serum voriconazole levels, two patients were changed to empirical antifungal agent treatment, and one patient could not receive voriconazole for other reasons (Fig. 1B). In 13 (8.8%) patients, voriconazole was discontinued due to insurance coverage issues. Among the 101 patients with premature voriconazole prophylaxis discontinuation due to AEs, liver function test abnormalities were the most frequent AE observed in 73/101 (72.3%) patients, followed by visual hallucinations and central nervous system symptoms (16, 15.8%), skin rash (6, 5.9%), renal dysfunction (3, 3%) and other (3, 3%: 2 patients for QTc prolongation and one patient with gastrointestinal symptoms).

Liver function tests

Overall, voriconazole was discontinued in 73 of 327 (22.3%) patients due to liver function abnormalities. Patients with abnormal liver function tests were more likely to have stage ≥2 liver GvHD (4/73, 5.5%) when compared to the remaining 254 patients in this cohort, amongst whom none developed liver GvHD (P-value >.002). At the time of voriconazole discontinuation, ALT and AST were ≥100 IU/ml in 39 (12.1%) and 27 (8.4%) of 321 patients with available values, respectively. The mean ALT at voriconazole initiation was similar between the two groups: 36.7 IU/ml (standard deviation [SD]: 122) in the early-discontinuation and 24.8IU/ml (SD: 16) for the SOC patient group (P-value = .2; Fig. 2B). However, the mean ALT at the end of voriconazole administration was significantly higher in the early-discontinuation (76.9 IU/ml, SD: 167.9) than the SOC-patient group (38.6 IU/ml, SD: 130; P-value: .02). Similar observations were made for total bilirubin (Fig. 2B). Within 7 days after voriconazole discontinuation, the mean ALT was similar between the two groups (early-discontinuation: 44.9 IU/ml vs SOC: 35.6 IU/ml, P-value = .30). However, the mean total bilirubin remained significantly higher in the early-discontinuation (1.1 mg/dl, SD: 2.4) compared to the SOC group (0.46 mg/dl, SD: 0.5; P-value = .002) by 7 days post-voriconazole discontinuation.

Figure 2.

(A) Boxplots of alanine aminotransferase (ALT) between patients who continued to receive voriconazole prophylaxis as per standard of care (“standard of care group”) and those patients who discontinued voriconazole prophylaxis due to variable reasons (“early discontinuation group”) at different time-points post-voriconazole prophylaxis initiation. (B) Boxplots of total bilirubin (TBil) between patients who continued to receive voriconazole prophylaxis as per standard of care (“standard of care group”) and those patients who discontinued voriconazole prophylaxis due to variable reasons (“early discontinuation group”) at different time-points post-voriconazole prophylaxis initiation. D, day; EOT, end of treatment; ALT, alanine aminotransferase; TBil, total bilirubin.

Risk factors for voriconazole discontinuation

Three different analyses were performed to identify predictors of voriconazole discontinuation: (a) at any time, (b) by day 14 post-HCT, and (c) by day 28 post-HCT (Supplement Tables 1 and 2; Supplement Fig. 1). Important predictors for early voriconazole discontinuation at any time in multivariable analyses included: GvHD ≥2 grade (odds ratio [OR]: 1.9, P-value: .02), ALT on day 14 of voriconazole administration ≥75 IU/ml (OR: 5.6, P-value: .02) and total bilirubin on day 7 of voriconazole administration ≥1.3 mg/dl (OR: 3.0, P-value: .03) (Fig. 1C). Risk factors for discontinuation of voriconazole by day 14 post-HCT in multivariable analyses included: age ≥65 years (OR: 3.2, P-value: .001) and baseline ALT ≥75 IU/ml (OR: 5.4, P-value: .04). Elevated ALT by day 14 of voriconazole administration ≥75 IU/dl was the only significant risk factor for discontinuation of voriconazole by day 28 post-HCT in multivariable analyses (OR: 7.7, P-value: .01).

Clinical outcomes

There were 13 proven and probable IFI observed during the first 180 days in this cohort: 8/147 (5.4%) and 5/180 (2.8%) in the early-discontinuation and the SOC-group, respectively (log-rank: 0.13; Fig. 3A). In the SOC group, there was one breakthrough probable invasive mould infection (IMI) during prophylaxis with voriconazole, while three additional IMI and one case of coccidiomycosis occurred after discontinuation of primary voriconazole prophylaxis. In the early-discontinuation group, there were four breakthrough IFI during voriconazole administration (two cases of invasive candidiasis, one infection due to Saccharomyces spp., and one probable IMI), while four cases of probable IMI occurred after discontinuation of voriconazole administration. By day-180 post-HCT, 23 (15.6%) of 147 patients in the early-discontinuation and 14 (7.8%) of 180 patients in the SOC patient groups had died (log-rank: 0.03; Fig. 3B). In mortality predictor analysis, only age (OR: 1.03, P-value: .04) and early voriconazole discontinuation (OR: 2.3, P-value: .02) were significant predictors of mortality by day-180 post-allogeneic HCT (Supplement Table 3).

Figure 3.

(A) Incidence of invasive fungal infections between allogeneic hematopoietic cell transplant recipients whose voriconazole prophylaxis was continued as per standard of care (“standard of care group”) and those patients who discontinued voriconazole prophylaxis due to variable reasons (“early discontinuation group”). (B) Overall 180-day mortality between allogeneic hematopoietic cell transplant recipients whose voriconazole prophylaxis was continued as per standard of care (“standard of care group”) and those patients who discontinued voriconazole prophylaxis due to various reasons (“early discontinuation group”).

Discussion

In this retrospective single-center study we report a high discontinuation rate of primary AFP with voriconazole in up to 45% of allogeneic HCT recipients within the first 3 months post-transplant. In almost one in five patients, voriconazole AFP was discontinued due to hepatotoxicity, with liver test abnormalities during the first 2 weeks of voriconazole AFP administration predicting early voriconazole discontinuation. There was a trend for lower rates of proven and probable IFI and a survival benefit in patients who continued to receive AFP as per institutional SOC.

Almost half of the patients had to stop voriconazole in this series. This rate of voriconazole discontinuation is higher than previously reported, and highlights the difficulties associated with the administration of this agent in clinical practice.¹⁸ Voriconazole-associated AEs and drug-drug interactions were the main reasons leading to drug discontinuation. In addition to liver function test abnormalities, visual hallucinations, other neurological symptoms, renal dysfunction, and prolonged QTc were among the most commonly observed AEs. It is likely that in a proportion of patients, voriconazole-associated AEs could have been due to elevated voriconazole blood levels. Voriconazole therapeutic drug monitoring (TDM) has been associated with a significant reduction in voriconazole-associated AEs and discontinuation rates.¹⁸ Lack of routine TDM in this series could have, in part, contributed to more frequent toxic levels and therefore higher rates of discontinuation.

Liver function abnormalities are common in allogeneic HCT recipients due to a large variety of potential causes, including, but not limited to, viral and other infections, GvHD, veno-occlusive disease, and concomitant drug administration, including azoles. Voriconazole-associated liver toxicity has been described in up to 16% of patient in the setting of prospective randomized clinical trials.^9,11,19,20 Our observations are consistent with those prior reports, with only 8–12% of patients in this series having elevated transaminases at ≥100 IU/ml at the time of voriconazole discontinuation. In fact, liver function test abnormalities in this series were rather modest, with the mean ALT value at the end of voriconazole administration being only twice the ALT upper limit of normal in the early-discontinuation group. Despite this moderate degree of hepatotoxicity, voriconazole was discontinued in almost one in five patients due to liver test abnormalities. This is consistent with prior observations at our institution.¹⁴ In fact, moderate elevations in ALT (≥75 IU/ml) and/or total bilirubin (≥1.3 mg/dl) during the first 2 weeks of voriconazole AFP administration were significant predictors of early discontinuation of voriconazole. The above suggests that, although a causal relationship between voriconazole and hepatotoxicity cannot always be definitively demonstrated, even mild liver test abnormalities may lead clinicians to stop the administration of this agent in up to 20–25% of patients due to fear or suspicion that voriconazole may be contributing to the observed test abnormalities.

The overall effect of voriconazole discontinuation on allogeneic HCT recipients is not well described. We observed a trend toward more IFI in patients who had to discontinue voriconazole AFP. This retrospective study was not powered to detect the effect of voriconazole AFP on the incidence of IFI. However, the incidence of IFI in the SOC and early discontinuation groups, respectively, was similar to that reported in a clinical trial that compared voriconazole to fluconazole as primary AFP in allogeneic HCT recipients.¹⁰ Early voriconazole discontinuation was associated with higher all-cause mortality by day 180 post-transplant. It is not possible to assess the direct effect of voriconazole discontinuation on mortality, and it is possible that voriconazole was more likely to be stopped in patients with more comorbidities and worse overall prognosis. However, these data taken together suggest that patients in whom clinicians are more likely to discontinue voriconazole AFP may be at higher risk for complications. These patients could potentially benefit from more intensive management and closer follow-up to optimize clinical outcomes.

This study has several limitations, including its retrospective nature and lack of data on voriconazole levels and concomitantly administered drugs. Attribution of hepatotoxicity to voriconazole or other potential causes was not feasible but reflects the real life limitations of clinical practice in allogeneic HCT recipients early post-transplant.

In conclusion, this study describes the clinical experience with voriconazole AFP among unselected allogeneic HCT recipients in a single cancer center. Adherence to voriconazole AFP among HCT recipients is challenging with early discontinuation observed in a substantial proportion of HCT recipients, mostly due to anticipated AEs. Moderate abnormalities in liver function tests during the first 2 weeks of administration may predict early voriconazole discontinuation. Strategies for monitoring toxicities and managing intolerance of voriconazole are essential for HCT recipients receiving voriconazole as AFP.

Conflicts of interest

M.P. reports honoraria from Abbvie, Bellicum, Celgene, Bristol-Myers Squibb, Incyte, Merck, Novartis, Nektar Therapeutics, Omeros, and Takeda. He serves on DSMBs for Cidara Therapeutics, Servier and Medigene, and the scientific advisory boards of MolMed and NexImmune. He has received research support for clinical trials from Incyte, Kite/Gilead and Miltenyi Biotec. He serves in a volunteer capacity as a member of the Board of Directors of American Society for Transplantation and Cellular Therapy (ASTCT) and Be The Match (National Marrow Donor Program, NMDP), as well as on the CIBMTR Cellular Immunotherapy Data Resource (CIDR) Committee. D.N. has received research support from MSD and has served as has received consulting fees from Roche Diagnostics, MSD, Pfizer, Basilea, and Gilead. G.A.P. has been an investigator for Shire, Merck, Chimerix, and Astellas, and has received consulting fees from Shire, Merck, Chimerix, and Astellas. All other authors report no conflicts of interest.