In the late 1970s, supported by the National Institutes of Health, we began a prospective study of a large cohort of persons with recently acquired symptomatic genital HSV infection in order to define more precisely the natural history of genital herpes. Clinical, demographic, and recurrence data were collected for 457 patients who presented with virologically, serologically, and clinically confirmed first-episode genital infection. More than half of these patients did not receive antiviral therapy during their primary episode, providing a population not likely to be replicated in the future. We summarize the natural history of symptomatic recurrences in the complete population and in the subset of untreated patients.

Methods

Top Methods Results Discussion Author & Article Info References

In 1974, a research clinic dedicated to the study of genital herpes infection was established at Harborview Medical Center, a King County-funded hospital affiliated with the University of Washington. Patients were referred by their private physicians or by the sexually transmitted disease clinic at Harborview, or responded to advertisements for participation in clinical studies of HSV infection. Only patients willing to participate in prospective studies with long-term follow-up or in therapeutic trials were enrolled.

Between 1974 and 1988, we registered 457 patients with serologically and virologically proven first-episode infection who were followed for at least 60 days from the onset of infection. At the initial clinic visit, genital lesions were cultured and described by anatomic site, stage, and area. Patients were then generally followed at 2- to 3-day intervals until their lesions had healed. The median number of visits and genital examinations during the initial disease episode was 5 (range, 3 to 14 visits). After resolution of the first clinical episode, patients were instructed to return to the clinic during each recurrence or for routine visits at least every 2 to 3 months. Patients who were unable to return for each recurrence were instructed to maintain a diary of onset and resolution dates for each recurrence. These data were collected at the following clinic visit. In general, we insisted that patients return for recurrences until they were able to recognize the signs and symptoms of genital herpes reactivation and to fill out patient diaries about onset and healing of reactivations.

We defined a recurrence of genital herpes as the presence of genital ulcerations. We defined duration of a recurrence as the number of days from the first appearance of genital lesions to the reepithelialization of all lesions. If new lesions appeared before complete healing of other lesions, all were considered part of the same episode. We report only recognized symptomatic (lesional) recurrences and do not address subclinical shedding of HSV in the genital tract.

Serum specimens from both the acute and convalescent phases were obtained from all patients and tested for HSV-specific antibodies by Western blot; all patients showed seroconversion to HSV-1 or HSV-2 [11, 12]. Some had antibodies to HSV-1 in their acute-phase specimen and antibodies to both HSV-1 and HSV-2 in their convalescent-phase specimens. Patients were categorized as having primary first-episode disease if their acute-phase serum specimen showed absence of antibodies to HSV by Western blot. Based on Western blot profiles of convalescent-phase specimens and the subtype of the HSV isolates from lesions, patients were categorized as having primary HSV-1 or primary HSV-2 infection. Patients who had HSV-1 antibodies in acute-phase specimens and antibodies to HSV-2 in convalescent-phase specimens were classified as having nonprimary first-episode HSV-2 disease [11, 13]. Patients with detectable antibodies to the homologous viral type isolated in initial specimens were classified as having recurrent disease [13]. Forty-two patients had serologic evidence of HSV-2 antibodies in acute- and convalescent-phase serum specimens, and 19 patients with HSV-1 antibodies in both their acute- and convalescent-phase specimens were classified as having reactivation infection and were not included in this study even though they claimed to be experiencing their first episode. We also excluded 64 patients seen between 1974 and 1985 whose serum specimens were analyzed by microneutralization but whose enrollment specimens could not be retrospectively retrieved for confirmation by Western blot [14].

In this study, we compared baseline patient characteristics and severity of primary infection with the subsequent recurrence experience of the patient. Patient characteristics included sex, race, age, and measures of past sexual activity, such as number of partners and history of sexually transmitted diseases. Clinical characteristics included pain, itch, discharge, fever, headache, photophobia, and stiff neck. Because many patients were participants in our early randomized trials of acyclovir, treatment status was also recorded [15-17].

Statistical Methods

Monthly recurrence rates for each patient were estimated by dividing the number of recorded recurrences by the number of months the patient was followed. Comparisons of recurrence rates were made using the Wilcoxon rank-sum test or, for comparisons between more than two groups, the Kruskal-Wallis test. The Kaplan-Meier estimate was used to compute time to first recurrence, and appropriate comparisons were made using the Cox model [18].

Recursive partitioning was used as an exploratory technique to identify potential subsets of patients at higher risk for subsequent recurrence. Classification and regression trees were used for recursive partitioning of recurrence rates and a modification of this technique by LeBlanc and Crowley was used to establish time to first recurrence [19, 20]. These techniques identify the variable (and cutpoint, for continuous variables) most closely related to the recurrence pattern. Within each of these two splits, or nodes, of the data, the data are split again. This process is repeated within each node until no further splits appear to be important in predicting recurrences. This partitioning of the data can be represented as a "tree" that shows the splits of the variables into disjoint patient subsets.

Because our previous placebo-controlled treatment trials failed to detect an effect of antiviral therapy on subsequent recurrences, most of our analyses are reported for the entire patient population. However, because treatment is known to affect the duration of primary symptoms, analyses to assess the potential effect of symptoms on subsequent recurrences were also done in the subset of untreated patients.

Results

Top Methods Results Discussion Author & Article Info References

Viral Type and Clinical Classification of Persons with First-Episode Genital Herpes

Of the 457 patients presenting with first-episode genital herpes, 399 (87%) had primary genital HSV infection; 73 (16% of the total cohort) were infected with HSV-1 and 326 (71% of the total cohort) were infected with HSV-2. Fifty-eight (13%) patients were classified as having nonprimary HSV-2 infection. The median age of the patients was 24 years; 92% were single and 91% were white. Demographic characteristics were similar among those presenting with primary HSV-1, primary HSV-2, and nonprimary HSV-2 infection and were identical to those reported previously [10, 15-17, 21].

Frequencies of the major clinical signs and symptoms of the initial episode of genital herpes are shown in Figure 1. Patients with true primary genital herpes infection, regardless of infecting viral type, tended to have more severe disease than did patients with previous HSV-1 infection. This was most evident with respect to constitutional symptoms: Seventy-nine percent of those with primary HSV-1 or HSV-2 infection reported at least one constitutional symptom (fever, headache, photophobia, or stiff neck) during their first episode, compared with only 43% of those with nonprimary infection. Seventy-seven percent of those with primary episodes had inguinal adenopathy compared with 52% of those with nonprimary genital herpes (P = 0.001). Frequencies of symptoms and signs were similar between those with primary genital HSV-1 and primary genital HSV-2 infections, except in the case of nuchal rigidity, which was reported by 42% of patients with primary HSV-2 infection and 12% of those with primary HSV-1 infection (P = 0.005).

View larger version (53K): [in this window] [in a new window]   Figure 1. Frequency of clinical signs and symptoms of first-episode genital herpes by viral type and evidence of previous herpes simplex virus type 1 (HSV-1) exposure. P values are based on chi-square differences among the three groups.

Overall Recurrence Rates after Resolution of the Initial Episode

Follow-up was defined as the time from enrollment to the date of the last clinic visit or to the date that a patient initiated long-term suppressive oral acyclovir therapy. The median follow-up was 418 days (range, 61 to 4897 days) and was similar in all subsets of patients whether segregated according to viral type, sex, or severity of clinical episode. The median numbers of clinic visits per patient in the first 90, 91 to 180, 181 to 365, and 366 to 720 days of follow-up were 10, 2, 2, and 1, respectively. The large numbers of visits in the first 3 months were due not only to our request for frequent visits (which were necessary to teach patients to identify symptoms) but also to participation by many patients in studies of cell-mediated immunity after the primary episode [22, 23]. The median number of clinic visits for persons in this study was 18 (range, 11 to 29 visits). Seventy-eight percent of the patients were followed for at least 6 months, 56% for at least 1 year, 32% for at least 2 years, and 17% for 3 or more years. The predominant reason for dropping out, given by 68% of patients, was the time commitment required for the study; only 20 patients were censored during follow-up for initiating oral suppressive therapy with acyclovir.

Eighty-nine percent of patients with primary HSV-2 infection had a recurrence during follow-up; their median recurrence rate was 0.34 recurrences per month and 4 recurrences per year. Fifty-one percent of the recurrent episodes were seen in the clinic and 49% were self-reported in patient diaries. Seventy-five percent of the first recurrences were seen in the clinic. The percentage of patients who had recurrences and the median recurrence rates were similar for those with primary and nonprimary HSV-2 infections. Figure 2 shows the frequency of recurrences in the initial 12 months of follow-up among the 203 patients with HSV-2 infection who were followed for at least 1 year. Only 11% of patients did not have recurrences; seventy-seven (38%) had at least 6 recurrences during the first year and 40 (20%) had more than 10 recurrences in the first year of follow-up. These data appear to be representative of the entire cohort because first-year recurrence rates were similar among patients followed for at least 1 year and those who discontinued follow-up within a year of their primary attack; only 5 patients in the cohort of 457 patients were placed on suppressive therapy with acyclovir during the first year of follow-up. The median recurrence rate for patients with HSV-2 infection was much greater than that for patients with HSV-1 infection (0.34 per month compared with 0.08 per month [P < 0.001]).

View larger version (37K): [in this window] [in a new window]   Figure 2. Frequency of clinical recurrences of genital herpes simplex virus type 2 (HSV-2) infection during the first 12 months after acquisition of infection. Sample is based on 145 women and 58 men followed for 12 months after acquisition of initial HSV-2 infection.

Effects of Sex, Age at Acquisition, and Severity of the Initial Episode on Recurrence Rates

We did analyses of risk factors influencing recurrence rates. Univariate analyses showed that more women than men had either no or one recurrence in the first year of follow-up (26% of 145 women compared with 8% of 58 men; (Figure 2). Fourteen percent of women and 26% of men had 10 or more recurrences in the first year. Among the 326 patients with primary HSV-2 infection, men tended to have recurrences sooner than women (Figure 3). The median first-year recurrence rate among men who presented with primary HSV-2 infection was 0.43 recurrences per month, compared with 0.33 recurrences per month among women [P = 0.005] (Table 1).

View larger version (11K): [in this window] [in a new window]   Figure 3. Kaplan-Meier analysis of time to initial recurrence of genital herpes among male and female patients with primary herpes simplex virus type 2.

Patients who had prolonged first episodes ( 35 days in duration) had much shorter times to first recurrence (P = 0.013) and tended to have recurrences twice as frequently as patients with shorter initial episodes. Overall, among the 20 (6% of the 326) patients with primary HSV-2 infection who had episodes that lasted 35 days or more, the median recurrence rate during the first year was 0.66 recurrences per month, compared with 0.36 recurrences per month among patients with shorter durations (P = 0.001). A similar relation between duration of primary episodes and subsequent recurrences was obtained when we considered only the 198 patients who received no antiviral therapy during their primary episode of infection (Table 2). Again, the median monthly recurrence rate was 0.68 (approximately 8 recurrences per year) compared with a rate of 0.33 (approximately 4 recurrences per year) for those with healing times less than 35 days.

Younger age at acquisition was also associated with more frequent recurrences. Patients who were in their teens or twenties tended to have a shorter time to first recurrence. The median recurrence rate for those 30 years of age or younger was 0.41 compared with 0.34 for those patients older than 30 years (difference, 0.07 [95% CI, 0 to 0.17]; P = 0.11). We were unable to detect an effect of the presence or absence of previous HSV-1 infection on either recurrence rates or time to first recurrence.

Assessment of the Relations between Risk Factors

Because there may be interactions between variables that might not be detected in univariate or standard stepwise analyses, we did exploratory analyses using recursive partitioning on the time to first recurrence. For these analyses, we considered group, age, sex, duration of the primary episode, number of constitutional symptoms, presence or absence of symptoms, and duration of symptoms. These analyses were first done in the subset of patients not treated with acyclovir. As expected, the most important determinant of subsequent recurrences was viral type. Among the patients with HSV-1 infection, there were no further splits, indicating that we were unable to detect additional variables statistically related to the time to first recurrence. Among patients with HSV-2 infection, those with longer primary episodes of genital herpes tended to have recurrences sooner than those with shorter episodes of initial infection. Among the 20 patients in whom the duration of first-episode infection was 35 days or more, the median time to first recurrence was 27 days, compared with 50 days in the 211 patients in whom the duration of the initial infection was 34 days or less (P < 0.05). Among the 211 patients whose first episode infection lasted less than 35 days, men tended to have recurrences sooner than women (median time to recurrence, 47 and 53 days, respectively) (P < 0.001). No other measures of severity of the primary episode appeared predictive of the subsequent time to first recurrence. Similar results were seen using all 384 patients with first-episode HSV-2 infection (primary and nonprimary infection). In this larger subset, age and sex were also identified as factors related to the time to first recurrence.

The variables were then considered using the Cox proportional-hazards model. As before, viral type and duration of the primary episode were the first variables to influence the model evaluating the population and time to initial recurrence. Age and sex, when added at subsequent stages, also influenced the model. Thus, using several methods of analyses, we found that viral type, severity of the primary first episodes, male sex, and younger age were variables associated with a shorter time to first recurrence. No other measures of severity of the primary episode, such as photophobia, nuchal rigidity, or duration of local pain or itching, emerged as a predictor of recurrence.

We examined the first-year recurrence rates in a similar fashion. Among patients with primary HSV-2 infection, duration of the primary episode was again selected in the recursive partitioning model, with the corresponding split at 35.5 days (rate, 0.33 recurrences per month for 35 days compared with 0.68 recurrences per month for >35 days). Year of enrollment (reflective of changing patient populations, due in part to the availability of acyclovir trials), younger age, and male sex were also selected by the model as factors influencing recurrence rates. These variables were also identified using multiple regression. Thus, using two different measures of recurrence and two analytic techniques, we found that viral type and duration of the initial episode of infection emerged as the main variables associated with subsequent recurrences; male sex and younger age were also identified as variables influencing reactivations.

Because each of the above analyses was exploratory in nature, we reassessed these models using an independent data set obtained from patients enrolled at the same clinic in the 3 subsequent years. From 1989 to 1992, 84 additional patients with first-episode disease were followed in a similar manner; 31 (37%) had primary HSV-1 infection, 47 (56%) had primary HSV-2 infection, and 6 (7%) had first-episode nonprimary HSV-2 infection. Again, viral type and duration of the primary episode were identified as the main variables related to time to first recurrence and recurrence rates. The parameter estimates for the Cox models from the two data sets are of similar magnitude, suggesting that the models reflect a real relation among these variables rather than a chance relation based on a single data set (Table 3). However, the analysis did not show a similar effect for sex. Although the median recurrence rate for men (0.47 per month) was similar to that for the men in the original data, the women had a median rate of 0.49 per month, roughly one third greater than that of the women in the earlier data. These differences in recurrence rates between the original data file and the validation set cannot be explained by the more frequent use of acyclovir in the latter patient group. Our original placebo-controlled trials failed to see any difference in recurrence pattern. Moreover, a comparison between patients treated with acyclovir in the 1974-1988 data set and those in the 1989-1992 data file also indicated a more rapid time to first recurrence in those in the latter time period. As reflected by the changing distribution of viral and serologic type, we feel that these data are related to changes in the type of patient referred to our clinic during those years when routine acyclovir therapy for genital HSV infection had become widely accessible.

Discussion

Top Methods Results Discussion Author & Article Info References

This study extends our previous work on the natural history of genital herpes infection in several important areas. First, patient classification was done using the latest available viral typing techniques. All viral isolates were definitively typed using monoclonal antibodies, and all acute- and convalescent-phase serum specimens were assayed using our recently developed type-specific Western blot serologic test [11, 12]. Thus, all patients were accurately classified as having first-episode primary HSV-1 infection, first-episode primary HSV-2 infection, or first-episode nonprimary HSV-2 infection. Patients were seen frequently during their initial episode, allowing us to objectively define the severity and duration of each patient's initial attack of genital herpes. In addition, each patient was given exhaustive instruction on the signs and symptoms of recurrent genital herpes, urged to return to the clinic during the first recurrence before historical and patient-observed information was used, and followed for a much longer time than were patients in similar studies [24].

Recurrence rates in patients with symptomatic genital herpes are high. Eighty-nine percent of patients with HSV-2 infection had recurrences during follow-up (median number of recurrences, 4 per year). Thirty-eight percent of patients with HSV-2 infection had 6 or more recurrences yearly and 18% had 10 or more recurrences. Among patients followed for 2 or more years, recurrence rates in the first and second years were similar.

The strongest correlate of subsequent recurrences of HSV-2 infection was the severity of the first episode of genital herpes. Approximately 10% of the untreated patients with first-episode primary genital herpes had an initial episode of prolonged duration ( 35 days). Among these patients, the median time to first recurrence was nearly twice as short and the overall recurrence rate was twice as high as in those with first episodes of shorter duration. Further support for the relation between initial severity and subsequent recurrence pattern comes from 42 patients who presented with apparent symptomatic first-episode disease but for whom previous HSV-2 antibody was detected, suggesting a subclinical initial infection. In these patients, the overall median recurrence rate during follow-up was 0.22 recurrences per month, far less than the rate for patients with documented first-episode disease. Although prolonged duration ( 35 days) appeared to be an indicator of subsequent recurrences, we were unable to show a monotone (dose-response) relation between the duration of the initial primary episode and subsequent recurrence. As such, persons with a lesion duration of 20 to 30 days did not differ in recurrence pattern from those with a lesion duration of 10 to 20 days.

This is the first time we have been able to link the severity of the initial attack with the subsequent recurrence pattern of genital HSV infection. In previous studies, the number of patients with HSV-2 infection who had extended initial episodes was insufficient to discern a relation between severity of the primary episode and recurrence pattern [15-17, 24-27]. Recently, Burke and colleagues [28] showed that in guinea pigs the number of ganglia expressing the HSV latency transcripts correlates with subsequent recurrences of reactivation. These findings in animal models and our data suggest that virologic and host events that occur with the initial infection influence the subsequent long-term natural history of this infection. We have seen that sexual partners with the same strain of HSV-2 can have markedly different recurrence rates [29, 30], which suggests that the association between initial infection and subsequent recurrence rate is multifactorial and is not related solely to virus strain or subtype.

Previous studies of antiviral therapy have not shown a discernible effect of treatment on subsequent recurrence rates. One explanation for this may be that too few placebo recipients with extended primary HSV-2 infections were included in these studies to allow for detection of differences [24-27]. The high morbidity of first-episode genital herpes precludes withholding antiviral therapy from patients. However, it is of interest that in the period from 1989 to 1993, during which 79 of the 84 patients referred to our clinic for primary genital herpes were given oral acyclovir (1 g/d for 7 to 10 days), 4 patients had initial episodes lasting more than 4 weeks (1 patient had a 75-day primary episode).

The higher recurrence rate for men compared with women was one of our most surprising findings. Previous studies of acute symptomatic first-episode genital herpes infection have shown that women have more severe episodes as indicated by frequency and magnitude of systemic symptoms and prolonged duration of symptoms [10, 17, 31, 32]. The prevalence and severity of reported pain in reactivation HSV infection is also greater in women than in men. Our study is the first to detect sex differences in recurrence rates (0.44 recurrences per month for men compared with 0.33 per month for women). This difference between men and women is unlikely to be due to health care behavior because all care in our clinic was provided without charge and the frequency of visits and follow-up were relatively similar in men and women. Whether these recurrence rates are affected by differences in reactivation of HSV infection between men and women, differences in local mucosal responses in containing infection, or other factors is unclear.

These differences in recurrence rates have obvious implications for sex stratification in trials of therapies that attempt to influence genital herpes recurrence rates. In addition, the higher recurrence rate in men compared with women may explain why male-to-female transmission of HSV-2 is more frequent than female-to-male transmission [33].

Although these data on genital herpes are the most extensive and thorough ever assembled, some cautions must be noted regarding the interpretation and generalizability of our findings. Despite diligent efforts by the clinical research team, nearly half of the patients failed to complete a full year of follow-up. We have been unable to detect any differences in severity of initial disease, time to first recurrence, or early recurrence rates among patients who dropped out early compared with those who remained in follow-up. Nonetheless, we recognize that an unexplained potential bias might exist. In addition, our analyses must be viewed as exploratory. Although we were able to verify that most risk factors identified in the initial analyses were also related to recurrence patterns in a second data set, both patient groups come from the same referral population. Because our clinic is referral-based, our patient population may, in general, have more severe initial episodes of genital herpes infection. Assessment of these risk factors in other patient populations is needed.

Clinical Implications

Our data indicate that when patients are educated about the clinical manifestations of recurrent genital herpes infection, almost all those with HSV-2 infection, whether male or female, recognize that they have symptomatic recurrences. Viral typing is an important predictor of the subsequent likelihood of recurrences and hence of the potential for subsequently transmitting genital herpes. In addition, recurrences occur nearly twice as often in patients with primary episodes of extended duration, and men tend to have recurrences more frequently than do women. Perhaps most important, more than 35% of patients with symptomatic first-episode genital HSV-2 infection have frequent recurrences (> 6 per year). Among men, 45% have at least 6 recurrences within the first year after acquisition of infection, and the overall recurrence pattern does not appear to decrease in the following year of follow-up. Our data suggest that such persons might benefit from the early use of suppressive antiviral therapy [34, 35]. Whether more frequent use of suppressive antiviral therapy can reduce transmission of infection is unclear but warrants study.