Natamycin and voriconazole in Fusarium and Aspergillus keratitis: subgroup analysis of a randomised controlled trial

INTRODUCTION

Fusarium and Aspergillus species are aggressive corneal pathogens, and even with proper treatment, can lead to poor outcomes.¹ Voriconazole is effective in vitro against Aspergillus species, but may not perform as well against Fusarium species.² We undertook a clinical trial comparing topical voriconazole versus topical natamycin, with the overall results presented previously.³ Here, we perform a prespecified subgroup analyses of treatment within Fusarium cases, and separately, within Aspergillus cases, assessing the efficacy of voriconazole and natamycin.

METHODS

Complete methods of the clinical trial have been reported elsewhere.³ In brief, eligible patients had a KOH-positive fungal smear with filamentous fungal elements, and were randomised to receive either topical voriconazole 1% (Pfizer, New York, NY, USA) or natamycin 5% (Alcon, FtWorth, Texas, USA), and to repeat scraping or no rescraping. The primary outcome for the trial was best spectacle-corrected visual acuity (BSCVA) at 3 months from enrolment. Secondary outcomes included corneal perforation and/or therapeutic penetration keratoplasty (TPK) by treatment arm. Fungal cultures were performed for all patients enrolled in the trial. Fungal identification was performed using gross and microscopic characteristics. Ethical approval was obtained from the University of California, San Francisco and the Aravind Eye Care System.

Analyses were performed in the subset of patients who were diagnosed with Fusarium or Aspergillus keratitis. Wilcoxon rank sum and Fisher’s exact tests were used to analyse baseline characteristics. Linear regression was used to analyse the effect of voriconazole (vs natamycin) on BSCVA and scar size, controlling for baseline BSCVA and infiltrate size, respectively. Corneal perforation between the treatment groups was assessed using logistic regression, controlling for baseline infiltrate depth, using Firth’s correction. Analyses were performed in Stata V.10.0 (StataCorp, College Station, Texas, USA).

RESULTS

Of 120 smear-positive cases of fungal keratitis enrolled in the trial, 101 had positive growth on culture (84%). Forty-four were identified as Fusarium species (44%): 21 were randomised to natamycin (48%) and 23 to voriconazole (52%). Nineteen isolates were identified as Aspergillus species, two of which were mixed infections, so 17 (17%) were included in this analysis (11 (58%) A flavus, 5 (26%) A fumigatus and 1 (5%) A terreus); 10 were randomised to natamycin (59%) and 7 to voriconazole (41%). At baseline, median infiltrate/scar size in Fusarium cases was 3.2 mm (IQR 2.2–5.1) in the natamycin arm and 3.7 mm (IQR 2.7–4.3) in the voriconazole arm (p=0.68). Median infiltrate/scar size in Aspergillus cases was 4.1 mm (IQR 3.4–6.0) in the natamycin arm, and 4.4 mm (IQR 2.1–5.5) in the voriconazole arm (p=0.44). The majority of patients in each arm in both Fusarium and Aspergillus cases had an infiltrate depth of >0–33% at baseline (for Fusarium, 11 of 21, 52%, in the natamycin arm, and 16 of 23, 70%, in the voriconazole arm, p = 0.60; for Aspergillus, 6 of 10, 60%, in the natamycin arm, and 4 of 7, 57% in the voriconazole arm, p = 1.00).

In Fusarium cases, there was no difference in 3-month BSCVA with voriconazole versus natamycin (0.11 logMAR, 95% CI −0.24 to 0.46, p = 0.54). Voriconazole was associated with an increase in perforation in Fusarium cases (OR 33.4, 95% CI 1.16 to 962.9, p = 0.041, table 1). Of seven perforations total in Fusarium cases, six were in the voriconazole arm and one in the natamycin arm. Six TPKs were performed, all of which were in cases that had perforated, five in the voriconazole arm and one in the natamycin arm. In Aspergillus cases, there was no difference in 3-month BSCVA with voriconazole versus natamycin (−0.21 logMAR, 95% CI −0.71 to 0.29, p = 0.38) or perforation (OR 0.09, 95% CI 0.0002 to 40.6, p = 0.44, table 1). Of four perforations, three were in the natamycin arm and one was in the voriconazole arm. Of these, there were two TPKs in the natamycin arm, and one in the voriconazole arm. There was one additional TPK in a case in the natamycin arm that had not perforated. There was no difference in perforation between the cases randomised to rescraping or no rescraping in either the Fusarium species or Aspergillus species subgroups.

Table 1.

Perforation in Fusarium and Aspergillus keratitis patients

	Fusarium species (n=44)		Aspergillus species (n=17)
Covariate	OR (95% CI)*	p Value	OR (95% CI)*	p Value
Voriconazole (vs natamycin)	3.51 (0.15 to 6.87)	0.041	0.09 (0.00 to 40.6)	0.44
Rescraping (vs not rescraping)	0.19 (−1.87 to 2.24)	0.86	0.34 (0.00 to 50.0)	0.67
Depth at enrolment	2.59 (0.57 to 4.61)	0.012	5.8 (0.45 to 75.2)	0.18

^*Logistic regression predicting perforation with voriconazole (vs natamycin), rescraping (vs not rescraping), and infiltrate depth at baseline as covariates, using a Firth correction.

COMMENT

We found no difference in 3-month BSCVA or scar size between voriconazole- and natamycin-treated patients in Fusarium or Aspergillus keratitis. However, voriconazole-treated Fusarium cases were more likely to perforate than natamycin-treated patients. There are conflicting reports about the efficacy of voriconazole against Fusarium species in vitro, as well as reports of treatment failures with voriconazole despite a relatively low MIC₉₀.^2,4,5 Of the five perforations and/or TPKs in Aspergillus, four were in natamycin-treated cases. Overall in the trial, there was no difference between voriconazole and natamycin in 3-month BSCVA or in proportion of cases perforating.³ Even when prespecified, subgroup analyses need to be treated with caution. These numbers are small, and further research is warranted on differential effect of voriconazole and natamycin in Fusarium and Aspergillus keratitis.