When our clinical trial was initiated, several studies had already reported that corticosteroid therapy could potentially decrease the incidence or duration of postherpetic neuralgia [9-14]. Despite their small sample sizes and use of inadequate controls, these clinical trials resulted in the widespread adoption of prednisone as the standard therapy for herpes zoster. It was hoped that this agent would prevent persistent or chronic pain. More recent studies [15-18] did not confirm the validity of this approach. A large study that used acyclovir with or without prednisolone for as long as 21 days showed no beneficial effect of steroids on persistent pain [19]. Nevertheless, many physicians continue to use therapy with both antiviral drugs and corticosteroids to manage herpes zoster. We and other members of the National Institute of Allergy and Infectious Diseases Collaborative Antiviral Study Group did a clinical trial to determine the effect of adding corticosteroids to acyclovir for the treatment of herpes zoster. Our key study end points were measurements of pain and quality of life.

Methods

Top Methods Results Discussion Author & Article Info References

Patients

Eligible participants were immunocompetent adults older than 50 years of age who presented with a localized dermatomal vesicular rash that was clinically consistent with herpes zoster and had been present for less than 72 hours. Patients were referred to participating investigators by community physicians or from university-affiliated clinics. The diagnosis was confirmed by the isolation of varicellazoster virus in tissue culture and by serologic testing. We excluded patients who required immunosuppressive therapy; patients with cancer; women capable of conceiving and bearing a child; patients who had a history of hypertension (diastolic pressure > 100 mm Hg) or were receiving antihypertensive treatment; patients with osteoporosis or insulin-dependent diabetes mellitus; patients who had received other antiviral drugs or immune globulin products within the 4 weeks before the study began; and patients with a history of glycosuria or hyperglycemia. Other concurrent medications or illnesses did not warrant exclusion. All patients gave informed consent before randomization. The informed consent document was approved by all local institutional review boards in compliance with regulations of the U.S. Food and Drug Administration.

Interventions

We used a 2 x 2 factorial study design to randomly allocate the 208 eligible and consenting patients. Patients received one of the following regimens: 1) acyclovir plus prednisone, 2) acyclovir plus prednisone placebo, 3) prednisone plus acyclovir placebo, or 4) placebos for both prednisone and acyclovir. Acyclovir was administered orally, 800 mg five times daily, for 21 days. Prednisone was administered orally at 60 mg/d for days 1 to 7, 30 mg/d for days 8 to 14, and 15 mg/d for days 15 to 21. All research personnel remained blinded to drug assignments until the study was completed and the database was locked. All matched medications (acyclovir plus prednisone placebo and prednisone plus acyclovir placebo) were identical in taste and appearance.

With a 2 x 2 factorial design, two treatments are evaluated simultaneously in one study. This design allows the assessment of an individual intervention and of the combined effects of multiple interventions (in our study, acyclovir and prednisone). The main or individual effect of acyclovir represents the difference between the effect of acyclovir therapy and the effect of no acyclovir therapy. We compared the patients who received acyclovir (the acyclovir plus prednisone group and the acyclovir plus prednisone placebo group) with the patients who did not receive acyclovir (the prednisone plus acyclovir placebo group and the placebo group). Similarly, the main effect of prednisone represents the difference between the effect of prednisone therapy and the effect of no prednisone therapy. We compared the two groups that received prednisone with the two groups that did not. This study design also allowed us to compare the combined effect of acyclovir and prednisone therapy with the effect of either therapy alone or with the effect of placebo.

Outcome Measures and Surveillance

Our primary end points were the persistence of pain and assessments of quality of life (return to 100% usual activity, return to uninterrupted sleep, and cessation of use of analgesic agents) during a 6-month period. Secondary end points were cutaneous healing and clinical or laboratory evidence of toxicity. After enrollment, patients were evaluated daily until the skin was completely healed. The number of new vesicles formed, extent of healing (measured by the presence of vesiculation, pustulation, scabbing, and total healing), severity of pain, and analgesic requirements were recorded at each visit. Acute neuritis was assessed during the first month after enrollment. Each patient was asked to rate his or her pain as none, mild, moderate, or severe or incapacitating. The effect of herpes zoster on usual activity (measured as a percentage reduction in activity), continued analgesic requirements, and the ability to sleep without interruption was also recorded at each visit. After lesions had completely healed, patients were followed monthly for the 6 months after therapy began and were assessed for persistence of pain, analgesic requirements, and quality-of-life measures.

We used two analyses to measure chronic pain. First, postherpetic neuralgia was defined as pain that was present 30 days after disease onset and that continued until complete resolution or 6 months. Because we sought to maintain the principles of the intention-to-treat analysis and include patients with no pain at 30 days after disease onset, we assigned such patients a value of zero for the time to resolution of postherpetic neuralgia. Second, pain associated with zoster was defined as the continuum of pain extending from study enrollment until final resolution. This second analysis included all patients from the time of enrollment. For all pain analyses, resolution of pain was defined as the last occurrence of any pain attributed to herpes zoster. Time to return to uninterrupted sleep was assessed by the patient as the time when he or she was no longer awakened by pain or altered sensations attributable to herpes zoster. Similarly, times to cessation of the use of analgesic agents and return to 100% usual activity were defined as the first recorded time to that event with no subsequent documented relapse.

Blood and urine samples were tested for evidence of treatment toxicity every week for the first 4 weeks. Assessments included a complete blood count with differential and platelet determinations; urinalysis; and quantitation of serum and urine levels of urea, creatinine, bilirubin, glucose, and aspartate glutamyltransferase. Adverse clinical events, regardless of their probability of being associated with drug administration, were reported immediately by telephone or facsimile to the central unit at the University of Alabama at Birmingham.

Randomization

The Biostatistical Unit of the National Institute of Allergy and Infectious Diseases Collaborative Antiviral Study Group at the University of Alabama at Birmingham used a computer-generated randomization code to randomly assign patients to one of the four treatment groups. Randomization was done in blocks of four.

Estimation of Sample Size

The sample size was calculated according to assumptions of the main effect of acyclovir that were based on data available when the trial was designed. The null hypothesis was that patients receiving acyclovir and those not receiving acyclovir would have equal rates of persistent pain at 6 months. On the basis of a one-sided test procedure that had a significance level of 0.025 and of a projected 20% detected difference between treatment groups, we calculated that a sample size of 256 patients was needed to achieve a power of 90%. Because of slow accrual during the latter stages of the trial (which was probably caused by the widespread acceptance and availability of oral acyclovir), the study was terminated with a total enrollment of 208 patients. Two hundred one of the 208 patients were included in the final analysis, yielding a power of 82%.

Interim Analysis

A Data Safety and Monitoring Board was established by the National Institute of Allergy and Infectious Diseases through the Division of Microbiology and Infectious Diseases to monitor the progress of the trial and any adverse events. However, no interim analyses of efficacy end points were done.

Statistical Analysis

The clinical response measurements for time-to-event variables were estimated using the Kaplan-Meier method and were compared using the log-rank test. We used a Cox regression model to analyze the main effects of acyclovir and prednisone and their interaction, with adjustments for other covariates. The main effects of acyclovir and prednisone were each represented in the Cox model as a dummy variable coded as 0 or 1. The interaction of these drugs was represented as the product of the two dummy variables. We also used a Cox model to compare each treatment group with the placebo group, with adjustment for all specified covariates (sex, race, age, duration of lesions before enrollment, number of lesions at entry, severity of pain at presentation, and surface area of lesion involvement). Risk ratios and 95% CIs were calculated from the model. The associated P values for the risk ratios are based on the Wald chi-square statistic. All P values are two sided.

Results

Top Methods Results Discussion Author & Article Info References

Patient Characteristics

We enrolled 208 patients from 15 participating National Institute of Allergy and Infectious Diseases Collaborative Antiviral Study Group centers. Five randomly assigned patients were not included in these analyses because they never received study medication; no case record forms were submitted. Two other patients were proven to have herpes simplex virus infection and thus were not included in the intention-to-treat analysis. Of the 201 patients included in the intention-to-treat analysis, 51 (25%) received acyclovir plus prednisone, 48 (24%) received acyclovir plus prednisone placebo, 50 (25%) received prednisone plus acyclovir placebo, and 52 (26%) received acyclovir and prednisone placebos. Of these 201 patients, 1 died and 32 (16%) were lost to follow-up after the acute phase (the first 30 days after disease onset). These 33 patients were equally distributed among the treatment groups: Fourteen percent of the acyclovir plus prednisone group, 13% of the acyclovir plus prednisone placebo group, 20% of the prednisone plus acyclovir placebo group, and 17% of the placebo group were lost to follow-up after the acute phase.

As shown in Table 1, the treatment groups had similar demographic characteristics. Slightly more than 50% of patients were women, and 30% of patients were from minority groups. The median age of patients was 61 years, and the age by decade was well distributed among treatment groups. Most patients were randomly assigned within 2 days of the onset of lesions. The treatment groups had similar median numbers of lesions, but patients in the acyclovir plus prednisone group tended to have more lesions than did patients in the other groups. At baseline, we found that non-narcotic analgesic agents were the most common concurrent medication. At randomization, patients significantly differed only in extent of pain. The group that received two placebos contained the most patients with severe or incapacitating pain, followed by the acyclovir plus prednisone placebo group and then by the prednisone plus acyclovir placebo group. The acyclovir plus prednisone placebo group contained the fewest patients with moderate, severe, or incapacitating pain. We adjusted for these differences among treatment groups by including covariates in the Cox regression model. Inclusion of these "prognostic" variables in the model estimated treatment effect while keeping all other prognostic variables constant.

Evaluation of Efficacy

As shown in Table 2, the acyclovir plus prednisone group had significantly accelerated times to total crusting (risk ratio, 2.27 [95% CI, 1.46 to 3.55]) and total healing (risk ratio, 2.07 [CI, 1.26 to 3.38]) compared with the group that received two placebos. As shown in Figure 1, acyclovir plus prednisone caused faster healing than did the two placebos or prednisone plus acyclovir placebo. Healing was not accelerated in patients receiving prednisone plus acyclovir placebo compared with patients receiving two placebos. In almost all patients (including the recipients of two placebos), lesions completely healed by 37 days after treatment began.

View larger version (23K): [in this window] [in a new window]   Figure 1. Time to total healing of localized herpes zoster in the normal host. P = 0.008 for acyclovir plus prednisone compared with prednisone plus acyclovir placebo; P = 0.06 for acyclovir plus prednisone compared with placebos; P = 0.05 for acyclovir plus prednisone placebo compared with prednisone plus acyclovir placebo. * = number of patients at risk on study days 20 and 30; ACV = acyclovir.

Evaluation of Pain

Acyclovir plus prednisone had the greatest effect on the resolution of acute neuritis, as measured during the first month after disease onset. The group receiving acyclovir plus prednisone was 3.02 times (CI, 1.42 to 6.41 times) more likely to have resolution of acute pain than was the placebo group (Table 2). Patients who received prednisone, regardless of whether they also received acyclovir, were 2.28 times (CI, 1.35 to 3.86 times) more likely to have resolution of acute pain during the first month after disease onset than were patients who did not receive prednisone.

Zoster-associated pain assessed during the 6 months after disease onset was not significantly reduced in the acyclovir plus prednisone group compared with the placebo group (P = 0.10). Similarly, results of the Cox regression analysis showed that the resolution of postherpetic neuralgia was not significantly accelerated in any active-treatment group compared with the placebo group. The proportion of patients with pain at 3 or 6 months did not significantly differ among the four groups (P > 0.05).

Quality-of-Life Measurements

During the first month after disease onset, the times to return to uninterrupted sleep, return to 100% usual activity, and total cessation of the use of analgesic agents were accelerated in the acyclovir plus prednisone group compared with the placebo group (P < 0.01 for all comparisons) (Table 2). The acyclovir plus prednisone group was 2.12 times (CI, 1.25 to 3.58 times) more likely to return to uninterrupted sleep, 3.22 times (CI, 1.92 to 5.40 times) more likely to return to 100% usual activity, and 3.15 times (CI, 1.69 to 5.89 times) more likely to discontinue use of analgesic agents than was the placebo group. The main effects of acyclovir and prednisone for return to 100% usual activity were statistically significant. Prednisone recipients were 1.74 times (CI, 1.21 to 2.51 times) more likely to return to 100% usual activity than were patients who did not receive prednisone, and acyclovir recipients were 1.90 times (CI, 1.33 to 2.71 times) more likely to return to 100% usual activity than were patients who did not receive acyclovir.

Safety Evaluation

Rates of Discontinuation of Therapy

In each group, the percentages of patients that discontinued the four treatments were similar: 14% (7 of 51) of acyclovir plus prednisone recipients, 13% (6 of 48) of acyclovir plus prednisone placebo recipients, 18% (9 of 50) of prednisone plus acyclovir placebo recipients, and 19% (10 of 52) of patients receiving two placebos (P > 0.2). Patients discontinued therapy because of influenza, conjunctivitis or iritis, nausea and vomiting, complete resolution of disease, cutaneous dissemination, hyperglycemia, and bacterial pneumonia.

Adverse Clinical Events

Forty-two patients (21%) had one or more adverse events. The rates of adverse events did not significantly differ among the four treatment groups, but the placebo group tended to have the lowest rate (12%). The rates of adverse events in the other three groups were 22% in the acyclovir plus prednisone group, 27% in the acyclovir plus prednisone placebo group, and 24% in the prednisone plus acyclovir placebo group. The most frequently reported clinical adverse events were gastrointestinal symptoms, especially nausea and vomiting (the latter occurred in about 21% of patients). Other reported adverse events included edema; increased leukocyte counts; and altered platelet counts, bilirubin levels, or hepatic function test results. These events occurred in less than 2% of all patients.

Laboratory Abnormalities

Blood urea nitrogen, creatinine, and total bilirubin levels; platelet counts; hematocrit; and leukocyte counts did not significantly change during the study. Of note, serum aspartate glutamyltransferase levels exceeded 50 U/L in 16 patients (9%). Hyperglycemia developed in 7 patients (glucose level >9.99 mmol/L); 5 of these patients had been randomly assigned to one of the prednisone groups.

Complications of Treatment

One patient died during the 6-month study. This patient, who had been assigned to receive acyclovir plus prednisone, died of myocardial infarction on study day 26. Three patients (one receiving acyclovir plus prednisone placebo and two receiving prednisone plus acyclovir placebo) developed cutaneous dissemination, which was defined as more than 20 vesicles beyond the primary dermatome. These three patients recovered without developing complications or requiring intravenous acyclovir therapy. One patient in the placebo group was hospitalized for bacterial pneumonia that required intravenous antibiotic therapy. No other complications were reported.

Natural History

On the basis of a stepwise Cox regression model, two variables predicted the times to resolution of pain, return to uninterrupted sleep, return to usual activity, and cessation of use of analgesic agents during the first month after therapy began. As shown in Table 3, symptoms (especially zoster-associated pain) resolved significantly faster in patients with no or mild pain than in those with severe or incapacitating pain. Similarly, patients who had fewer lesions at enrollment had accelerated rates of resolution of the above variables during the first month. Age did not predict acute pain or resolution of any quality-of-life variable.

Correlation between Severity of Pain and Use of Analgesic Agents

As expected, severity of pain and use of analgesic agents during the first month were significantly correlated (Spearman correlation coefficient, 0.60; P < 0.01) (Table 4). Ninety-six percent of patients without pain and only 3% of patients with severe or incapacitating pain did not use analgesic agents. Further, as the severity of pain increased, the percentage of patients using more potent forms of analgesic agents (oral, parenteral, or combination) correspondingly increased. This trend persisted during the 6 months of the study (Spearman correlation coefficient, 0.68; P < 0.01). This finding indicates that the patients' self-report of pain was a valid representation of the degree of discomfort.

Discussion

Top Methods Results Discussion Author & Article Info References

Our study unequivocally showed that the combination of prednisone and acyclovir had a beneficial effect on the management of herpes zoster in relatively healthy persons older than 50 years of age. Although the addition of prednisone to acyclovir significantly accelerated the early events of cutaneous healing compared with placebo, the more important effect was on the time to return to usual activities, cessation of use of analgesic agents, and time to return to uninterrupted sleep. To our knowledge, ours is the first clinical trial of therapy for herpes zoster to use quality-of-life measures as primary end points. When the main effects of these therapies were evaluated during the first month after disease onset, prednisone had the greatest effect on sleep and use of analgesic agents, whereas acyclovir plus prednisone affected return to usual activity, especially for patients with severe or incapacitating pain at presentation. The clinical benefit of each of these variables can best be summarized as a 1.5-fold to 3-fold acceleration in the resolution of these debilitating complications during the study. These benefits occurred without such laboratory or clinical toxicities as elevated blood pressure and glycosuria. The addition of prednisone to acyclovir did not significantly affect the persistence of zoster-associated pain or postherpetic neuralgia. Because of the clear effect on quality-of-life measures, we recommend that combination antiviral and prednisone therapy be used for herpes zoster, especially in patients with severe or incapacitating pain.

The rationale for using prednisone is based on the results of small clinical trials with low power that were reported between the 1950s and the early 1980s [9-14]. These studies suggested that administering prednisone to patients with acute herpes zoster had a significant anti-inflammatory effect and thus reduced the risk for persistent pain. Our results can be compared with those of another recent clinical trial [19]. Both of these studies showed that the addition of steroids to acyclovir accelerated resolution of acute pain but had no effect on long-term pain. Wood and colleagues [19] concluded that the benefit of steroids was not clinically significant and therefore did not recommend that it be routinely used to treat herpes zoster. Nevertheless, the distinguishing characteristic of our study was the striking evidence of improved quality-of-life measures with the use of corticosteroids. The mechanisms by which the administration of prednisone produced this benefit are unknown. In addition to having anti-inflammatory effects, prednisone can induce an improved sense of well-being that may also contribute to the beneficial effects we observed.

The use of a 2 x 2 factorial study design that included a placebo group enhanced the ability to evaluate effects of acyclovir plus prednisone compared with effects of either drug alone. Of note, we continued administering acyclovir placebo even though acyclovir had been licensed for treating herpes zoster in 1985 and even though some investigators believed that withholding acyclovir was unethical [6, 8]. This decision was based on the lack of published, peer-reviewed data unequivocally showing that acyclovir has a beneficial effect on persistence of pain over time.

Our findings on the natural history of herpes zoster should allow researchers to improve the design of future clinical trials for the evaluation of herpes zoster. First, detailed assessments of events of healing in the primary dermatome did not correlate with quality-of-life end points or persistence of pain. Future trials should thus focus on issues most relevant to persons with herpes zoster—return to usual activity and resolution of pain. Second, the definition of pain for clinical trials must be uniform. We evaluated pain as a continuum (zoster-associated pain) and used a common definition of postherpetic neuralgia; however, we recognize that doing so creates a potential for a censoring bias not well appreciated in earlier studies, including our own [20].

We administered acyclovir and prednisone for 21 days for two reasons. First, we wished to provide optimal benefit from prednisone therapy and allow proper tapering after initiation of high-dose treatment. Second, acyclovir was administered for the entire course of prednisone therapy so that potential risks for cutaneous dissemination of varicella-zoster virus would be minimized. Data from well-designed clinical trials have shown the value of acyclovir therapy given for 7 or 10 days [2-5, 19]. We cannot use data from our trial to infer that a shorter period of acyclovir therapy would produce the same clinical benefits seen with a 21-day course. However, analysis of the one end point best reflecting antiviral effect—formation of new vesicles—showed that all patients who received acyclovir plus prednisone had ceased forming new lesions by day 8. New lesions continued to form on day 11 in only 5% of patients who received acyclovir plus prednisone placebo. Similarly, a recent clinical trial of patients with herpes zoster receiving acyclovir for 7 or 21 days found no differences in events of cutaneous healing [19]. Thus, it seems reasonable to administer acyclovir for no more than 10 days, as currently recommended by other groups. As with acyclovir, we cannot use our data to recommend shorter prednisone therapy.

Two new antiviral drugs, famciclovir and valaciclovir, have been licensed for the treatment of localized herpes zoster in immunocompetent hosts [6, 8]. The controlled investigations of these drugs did not address quality-of-life issues or use adjunctive corticosteroid therapy [6]. Because valaciclovir and famciclovir have a mechanism of action similar to that of acyclovir, it is reasonable to speculate that combination therapy with corticosteroids would produce similar benefits. However, this hypothesis should be confirmed in appropriate controlled clinical trials.

We conclude that the concomitant administration of acyclovir and prednisone provides substantial benefit for persons with herpes zoster, as measured by quality-of-life assessments. Importantly, the administration of high-dose prednisone is not appropriate for all patients. This regimen should be considered only for relatively healthy elderly persons who have moderate or severe pain. Patients with osteoporosis, diabetes mellitus, glycosuria, or hypertension were intentionally excluded from our clinical trial and may not be appropriate candidates for corticosteroid therapy. We also do not recommend that corticosteroids be used without concomitant antiviral therapy. A therapeutic approach to herpes zoster that improves quality-of-life and reduces the frequency and severity of persistent pain is still needed.

Appendix

The following are members of the National Institute of Allergy and Infectious Diseases Collaborative Antiviral Study Group: C. Laughlin (Project Officer, National Institute of Allergy and Infectious Diseases, National Institutes of Health, Bethesda, Maryland); R.J. Whitley, H. Weiss, J. Gnann, S.-J. Soong, L. Sherrill, P. Pappas (Central Unit, University of Alabama at Birmingham, Birmingham, Alabama); S. Greenberg, R. Couch, J. Wolfe, E. Basler, I. Orengo (Baylor College of Medicine, Houston, Texas); H.C. Liu (University of Kansas, Kansas City, Kansas); R. Pollard, S. Tyring (University of Texas Medical Branch, Galveston, Texas); F. Hayden (University of Virginia, Charlottesville, Virginia); C.J. Schleupner (Veterans Affairs Medical Center, Salem, Virginia); G.J. Mertz (University of New Mexico, Albuquerque, New Mexico); M. Savoia, M.N. Oxman (University of California, San Diego, California); D. Fine (University of Oklahoma, Oklahoma City, Oklahoma); R. Mangi (Hospital of St. Raphael, New Haven, Connecticut); H. Kessler (Rush-Presbyterian-St. Luke's Hospital, Chicago, Illinois); J. Peacock (Wake Forest University, Winston-Salem, North Carolina); J. Tilles (University of California, Irvine, California); F. Flowers (University of Florida, Gainesville, Florida); and K. Beutner (Solano Dermatology, Solano, California).

Study medications and matching placebos were provided by L. Smiley and G. Davis (Burroughs Wellcome Co., Research Triangle Park, North Carolina); L. Eron (Infectious Diseases Research, Fairfax, Virginia); and C. Vance (Hennepin County Medical Center, Minneapolis, Minnesota).

Drs. Weiss and Soong: University of Alabama at Birmingham, Comprehensive Cancer Center, 1824 6th Avenue South, WTI 220, Birmingham, AL 35294-3300.

Dr. Gnann: University of Alabama at Birmingham, 845 South 19th Street, BBRB 220, Birmingham, AL 35294.

Dr. Tyring: Department of Microbiology, University of Texas Medical Branch, 300 University Boulevard, RT J-19, Galveston, TX 77550.

Dr. Mertz: Infectious Diseases, University of New Mexico School of Medicine, 915 Camino de Salud, BRF-323, Albuquerque, NM 87131.

Dr. Pappas: University of Alabama at Birmingham, 1900 University Boulevard, THT 229, Birmingham, AL 35294.

Dr. Schleupner: University of Virginia and Veterans Affairs Medical Center, 1970 Roanoke Boulevard, Salem, VA 24153.

Dr. Hayden: University of Virginia Medical Center, Hospital Drive, Barringer 1, Room 183, Charlottesville, VA 22908.

Dr. Wolfe: Baylor College of Medicine, 6560 Fannin, Suite 802, Houston, TX 77030.