A Controlled Trial Comparing Foscarnet with Vidarabine for Acyclovir-Resistant Mucocutaneous Herpes Simplex in the Acquired Immunodeficiency Syndrome

Patients with HIV infection were eligible if they had herpetic lesions that were unresponsive to 2 weeks or more of oral therapy and to 10 days or more of intravenous therapy with acyclovir (≥5 mg per kilogram every eight hours) and if resistance to acyclovir and susceptibility to foscarnet and vidarabine were shown in vitro. Patients were excluded if they had a Karnofsky score below 40, a serum creatinine level above 177 μmol per liter, an absolute neutrophil count of less than 0.5×10⁹ per liter, a platelet count below 25×10⁹ per liter, more than a 2.5-fold increase in alkaline phosphatase or bilirubin levels, more than a 5-fold increase in the serum aspartate aminotransferase level, a serum calcium level <1.7 or ≥3 mmol per liter, a serum phosphorus level below 0.3 or above 2.6 mmol per liter, or were pregnant or nursing. Treatment with zidovudine, ganciclovir, or immunomodulators was not permitted during the study, in order to avoid confounding in the assessments of efficacy and toxicity. All study participants gave written informed consent.

The study patients were randomly assigned according to a centralized, computer-generated permuted-block scheme to receive either foscarnet (40 mg per kilogram intravenously every eight hours) or vidarabine (15 mg per kilogram per day intravenously). The dose of foscarnet was adjusted according to serial determinations of serum creatinine and weight.¹⁴ Therapy ended on day 10 if all lesions had healed completely and continued for up to 42 days if the response was partial (defined as a decrease in the total surface area of the two largest lesions of ≥25 percent). Patients in whom therapy failed (defined as <25 percent healing) were allowed to cross over to the alternative study drug. Eligible patients with a previous failure of vidarabine therapy or intolerance to the drug were exempt from randomization and received foscarnet on a noncomparative basis.

The primary end point of the study was the complete healing (epithelialization) of all lesions. Progress was assessed by serial measurement of the lesions with photographic documentation, the determination of a pain score (possible score, 0 to 10), and the collection of specimens for viral cultures every five days. Patients were monitored for the recurrence of herpes simplex virus for three months after they completed therapy with the study drug; the use of antiviral agents active against herpes simplex virus was not permitted during this period.

Virologic Studies

Viral culture and typing were performed at local laboratories.¹⁵ The Virology Research laboratories of San Francisco General Hospital and Beth Israel Hospital tested antiviral susceptibility with a plaque-reduction assay.¹⁶ Our definitions of resistance were as previously described.⁶ A biochemical assay of thymidine kinase activity measured [³H]thymidine phosphorylation with an extract of infected cells.^{9 , 17}

Statistical Analysis

We chose a sample of 26 randomized patients to enable us to detect a 50 percent difference in the proportion of patients with healing (90 vs. 40 percent) with a power of 80 percent and a two-tailed level of significance of 0.05. Interim data were presented to the Data and Safety Monitoring Board of the National Institute of Allergy and Infectious Diseases when 11 randomized patients had been enrolled. The study was terminated after the enrollment of 14 randomized patients, because of evidence of severe toxicity and a lack of efficacy of vidarabine.

A reviewer who was blinded to treatment assignment confirmed all classifications of measurements. Comparisons of efficacy involved the randomized patients only, on an intention-to-treat basis. At the suggestion of the monitoring board, treatment comparisons were stratified according to lesion size at entry (defined as the area of the two largest lesions combined). Patients who were exempt from randomization were analyzed for toxicity and recurrences only.

Randomized Patients

Base-Line Characteristics

Table 1. Table 1. Base-Line Characteristics of 25 Patients with Acyclovir-Resistant Herpes Simplex Infection.

Fourteen patients underwent randomization, eight to the foscarnet group and six to the vidarabine group. The two groups were comparable, except that patients in the vidarabine group had larger herpetic lesions (P = 0.07) that had been present for a longer period (P = 0.009) (Table 1).

Clinical Response

Time to healing was shorter in the patients assigned to receive foscarnet (median, 13.5 days) than in those receiving vidarabine (median, 38.5 days; P = 0.01 by the log-rank test, with stratification according to lesion size). The lesions in all eight patients assigned to foscarnet healed completely after 10 to 24 days of therapy. In contrast, in all six patients receiving vidarabine, the drug was discontinued because of therapeutic failure, often in combination with toxicity.

The time to halving of the pain score (median, 4 vs. 19 days; P = 0.004 by the log-rank test, with stratification according to lesion size), time to resolution of pain (median, 9 vs. 23 days; P = 0.12), and time to a 50 percent decrease in the size of the initial lesions (median, 9 vs. 28 days; P = 0.01) were shorter in patients assigned to receive foscarnet than in those assigned to receive vidarabine.

Five patients crossed over from vidarabine to foscarnet; the sixth did not because of a failure to regain base-line mental status. Two of the five had complete healing after 15 and 20 days of therapy with foscarnet. Three stopped receiving foscarnet after 93, 96, and 98 percent of their lesions had healed, on days 24, 32, and 38 of therapy, respectively.

Virologic Response

Table 2. Table 2. Susceptibility to Antiviral Drugs and Thymidine Kinase Activity of Herpes Simplex Virus Isolates from the Patients Who Underwent Randomization.

Every isolate was found to be resistant to acyclovir on the basis of marked deficiency of thymidine kinase activity, and each was susceptible to both foscarnet and vidarabine at the time the patients entered the study (Table 2). The time to virologic cure was shorter in the patients assigned to foscarnet than in those assigned to vidarabine (median, 6 vs. 17 days; P = 0.006 by the log-rank test, with stratification according to lesion size). All seven patients treated with foscarnet who were shedding herpex simplex virus at entry had a virologic cure, as compared with only one of the five patients randomly assigned to receive vidarabine. Virologic cure was confirmed by a minimum of two negative viral cultures of consecutive specimens collected at least 48 hours apart.

Table 3. Table 3. Patterns of Antiviral Susceptibility in Herpes Simplex Virus Isolates from All Study Patients.

The four patients with viral shedding at the time of crossover had negative cultures after a median of 5.5 days of treatment with foscarnet (range, 5 to 13). A comparison of the drug concentrations that inhibited plaque formation by 50 percent (ID₅₀) in isolates at study entry and at crossover showed no substantial changes in susceptibility to vidarabine (P = 0.8 by paired t-test), and the median ID₅₀ was unchanged (Table 3).

Patients Exempt from Randomization

Of the 11 patients who were exempt from randomization, 8 had complete healing after a median of 12.7 days of treatment with foscarnet (range, 10 to 42). Three patients stopped therapy when 35, 52, and 58 percent of their lesions had healed; cryptococcal meningitis developed in the first, and the other two had received the maximal duration of therapy (lesion size at entry, 238 and 110 cm², respectively). Virologic cure was documented in all 11 patients after a median of 6 days of therapy (range, 3 to 42).

Toxicity

Table 4. Table 4. Adverse Reactions in Patients Receiving Foscarnet or Vidarabine.

No patient receiving foscarnet had any dose-limiting toxicity, although side effects were frequent (Table 4). One patient had a transient exacerbation of preexisting neuropathic pain in the lower legs and feet during the administration of foscarnet.

Neurologic abnormalities developed in three patients receiving vidarabine. One had acute clumsiness and disorientation on day 9 of therapy, which resolved completely when vidarabine was discontinued and did not recur after crossover to foscarnet. Another had fever and nausea on day 9, followed by tremors, marked confusion, agitation, and myoclonic jerking 12 hours later. Neither myoclonus nor confusion fully resolved despite the prompt discontinuation of vidarabine, and the patient died 10 weeks later. In the third patient, symptoms of lightheadedness, insomnia, and mild confusion during the last days of vidarabine therapy progressed to disorientation, agitation, and paranoia in the ensuing three days; these resolved completely with the administration of clonazepam and did not recur after foscarnet therapy began. Extensive investigation revealed no cause of the neurologic abnormalities other than vidarabine treatment in any of the three patients.

Recurrence of Herpes Simplex Infection after Therapy

Eighteen of the 25 patients had complete healing during the study; data on 1 of the 18, who received suppressive acyclovir therapy after healing, were excluded from the analysis of recurrence. In the other 17, mucocutaneous herpes simplex infection recurred a median of 14 days after foscarnet was discontinued (range, 2 to 117). Of the 17 first recurrences, 10 were at the site of the healed acyclovir-resistant lesion (7 of them were susceptible to acyclovir, and 3 resistant). Eight patients had second recurrences at the site of the initial resistant lesion; all were resistant to acyclovir.

The median time to the recurrence of acyclovir-resistant herpes simplex virus was 41 days (range, 2 to 173); 16 patients were treated again with foscarnet, and 1 patient died of other causes. A comparison of paired ID₅₀ values for foscarnet and vidarabine at the time of the first (n = 15) and second (n = 8) recurrence with values at study entry showed no significant differences (for foscarnet, P = 0.8 and 0.6, respectively; for vidarabine, P = 0.6 and 0.3). In contrast, ID₅₀ values for acyclovir were significantly lower at the time of first recurrence than at study entry (P = 0.003 by the paired t-test); the median ID₅₀ was 0.8 as compared with 27 μg per milliliter (Table 3). A comparison of susceptibility to acyclovir at entry with susceptibility at second recurrence showed no significant difference (P = 0.1).

This randomized prospective trial shows that there is better efficacy and lower toxicity with foscarnet than with vidarabine for the treatment of acyclovir-resistant herpes simplex infection in patients with HIV infection. All 8 patients randomly assigned to foscarnet and 63 percent of the 16 patients who received foscarnet after the failure of vidarabine therapy had complete healing. In addition, 23 patients had a documented virologic cure during therapy with foscarnet. Our study thus confirms preliminary reports of successful foscarnet treatment in uncontrolled case studies among patients with acyclovir-resistant herpes simplex infection.^{1 , 3 , 6 , 9 , 13}

The imbalance in the size and duration of lesions between treatment groups that occurred despite randomization may have biased our assessment of the rate of healing. For this reason, all analyses of efficacy were statistically controlled for the size of the lesions. Also of note is that one patient in the vidarabine group had a relatively small lesion (<33 cm²) that did not heal and that each of the three patients with large lesions who were assigned to foscarnet had healing.

Although vidarabine is effective for the treatment of neonatal and central nervous system herpes simplex infection,^{10 , 11} all of our patients randomly assigned to vidarabine had no response despite in vitro susceptibility at study entry and the lack of evidence that resistance developed during therapy. Possible explanations include poor delivery of the drug to the skin or inadequate intracellular concentrations in target skin cells. Alternatively, the dose of vidarabine required for optimal mucocutaneous antiviral efficacy against herpes simplex virus may be higher than patients with HIV infection can tolerate.

Neurotoxicity occurred in three patients receiving vidarabine and contributed to the premature termination of the study. Although neurologic abnormalities have previously been described in association with vidarabine therapy,^{12 , 18 19 20 21 22 23 24 25 26 27} the frequency and severity of these events in our study were unexpected. It is conceivable that patients with HIV infection are particularly prone to neurologic side effects of vidarabine, for unexplained reasons.

Although the majority of first recurrences at the site of the index resistant lesion were susceptible to acyclovir, all second recurrences at that site had become resistant to the drug. Thus, although patterns of susceptibility in recurrences of acyclovir-resistant lesions are unpredictable, further treatment with foscarnet was eventually required in virtually all the patients. Of concern is the possibility that repeated administration of foscarnet may facilitate the development of foscarnet resistance. In vitro resistance to foscarnet has been described in clinical isolates of herpes simplex virus from three severely immunocompromised patients,^{8 , 28} and two other reports have documented alterations in the DNA polymerase enzyme in isolates of herpes simplex virus recovered from a patient who had received a bone marrow transplant^{29 , 30} — alterations that would be likely to confer resistance to foscarnet.

The frequency and severity of the neurologic toxicity, in addition to the poor efficacy, observed in this study indicate that vidarabine should not be administered to patients with HIV infection who have acyclovir-resistant herpes simplex infection and that foscarnet is the best alternative therapy at the current time. It is clearly important to identify additional antiviral agents that are effective against acyclovir-resistant herpes simplex virus, are well tolerated, and do not require parenteral administration. In addition, future studies must focus on optimal methods of suppressing the recurrence of successfully treated acyclovir-resistant herpes simplex infection.