INTRODUCTION

Top Abstract Introduction Materials and methods Results Discussion References

Human granulocytic ehrlichiosis (HGE) was first described in 1994 for a series of 12 patients residing in Minnesota or Wisconsin (4, 12). HGE is a febrile illness characterized by headache, myalgia, malaise, thrombocytopenia, leukopenia, and elevated levels of hepatic transaminases (5). HGE is clinically indistinguishable from human monocytic ehrlichiosis (HME), which is caused by Ehrlichia chaffeensis (2). As of May 1998, approximately 350 cases of HGE (four of which resulted in death) had been recognized in the United States (4, 21, 26, 30). The HGE agent is closely related to (or conspecific with) Ehrlichia equi, the agent of equine ehrlichiosis, and Ehrlichia phagocytophila, the agent of tick-borne fever in ruminants (15, 18, 47).

HGE is a zoonotic disease, and its natural history is still being defined. The blacklegged tick, Ixodes scapularis Say (including the species formerly known as Ixodes dammini Spielman, Clifford, Piesman, and Corwin [40]), is believed to be a principal biological vector of the HGE agent in the regions where this tick occurs (30, 41, 48, 49). Peromyscus leucopus, the white-footed mouse, is a competent reservoir host (48). Additional tick vectors or vertebrate host species maintain the HGE agent in some locations, such as northern California, where I. scapularis does not exist. Serologic evidence suggests that HGE-like agents occur in additional rodent species and in regions outside of the areas where HGE is currently recognized (39). It is possible that the agent is maintained in nature in a tick-rodent cycle similar to the Borrelia burgdorferi maintenance cycle, with humans being involved only as incidental dead-end hosts (31).

The Centers for Disease Control and Prevention (CDC) has made serologic testing for HGE available for state health departments since August 1995, following an investigation of 29 confirmed or probable cases of HGE in Westchester County, N.Y. (9). Before 1996, several strains of granulocytic ehrlichiae grown in horse neutrophils were used as antigens for testing for HGE by indirect immunofluorescence assay (IFA) at the CDC and elsewhere because the HGE agent had not yet been isolated and maintained in cell culture. The close genetic and antigenic similarities between these agents resulted in considerable cross-reactivity of human antibodies, which is sufficient to identify cases of HGE (15). Antigens produced in experimentally infected horses are still used for testing for HGE by some institutions and commercial laboratories. The HGE agent was recently isolated and adapted to cell culture (24, 38), and IFAs that use cell culture-derived antigens have been developed (38, 43). These assays offer several advantages over assays that use horse-derived antigens and are being increasingly used for testing for HGE. An assay developed at the CDC (38) was used to test serum samples from individuals with suspected rickettsial and ehrlichial illnesses for antibodies to the HGE agent.

	MATERIALS AND METHODS

Top Abstract Introduction Materials and methods Results Discussion References

Acquisition of samples. Serum samples from patients with suspected rickettsial and ehrlichial illnesses were submitted to the Viral and Rickettsial Zoonoses Branch, CDC, by physicians through their state health departments from throughout the United States. Serum samples were stored at 4°C or were retrieved from storage at 70°C prior to being tested for HGE. Retrospective testing back to 1987 was conducted on all available samples from seven states where the HGE agent is known or suspected to be endemic (California, Connecticut, Florida, Maryland, Minnesota, New York, and Wisconsin). Prospective testing for HGE began in August 1995 on samples submitted from any state for any rickettsial or ehrlichial antibody evaluation.

HGE agent antigens. Two sources of antigen were used for HGE testing. Commercially available antigen dotted onto Teflon-coated microscope slides (Spirochete and Rickettsia Laboratory, University of California School of Veterinary Medicine, Davis) was used until April 1996. These slides had been prepared with infected neutrophils harvested from a horse experimentally infected with the BDS strain of the HGE agent (36). This antigen was used to test 440 serum samples in this study.

Serologic assay. Serum samples were tested by an IFA that has been previously described (38). Prior to use, antigen slides were removed from storage at 70°C, placed into a desiccator, and allowed to warm to room temperature. To screen the samples, serial twofold dilutions were made and two dilutions (1/64 and 1/128) were placed onto 18- or 24-well, Teflon-coated microscope slides (Erie Scientific Co., Portsmouth, Maine). Slides were placed in a humidified plastic chamber and incubated for 30 min at 37°C and then washed three times for 5 min each time in phosphate-buffered saline, pH 7.4. An optimized dilution (1/100) of fluorescein isothiocyanate (FITC)-labeled goat anti-human conjugate specific to the heavy and light chains of human immunoglobulin G (IgG) (Kirkegaard & Perry Laboratories, Inc., Gaithersburg, Md.) was then applied to the slides. Slides were incubated and washed as before, except that 3 to 4 drops of a 1.65% solution of eriochrome black T in water was added to the middle wash as a counterstain. Glycerol-based antifade mounting medium was added to each well, a coverslip was applied, and the slides were read by using UV illumination. Any reactive samples were then titrated to endpoint with an optimized dilution (1/150) of FITC-labeled goat anti-human conjugate specific to the gamma chain of human IgG (Kirkegaard & Perry Laboratories, Inc.). Antibody titers were confirmed by a second microscopist reading coded slides. Titers are reported as the reciprocal of the highest dilution showing specific ehrlichial fluorescence. In this study, a titer of 64 or greater to the HGE agent was considered to be positive evidence of specific antibody (10).

Reactivity with other antigens. Serum samples usually had been tested against other rickettsial antigens, and titers to the HGE agent were compared with titers to Coxiella burnetii phase II (causative agent of Q fever), Ehrlichia canis (canine ehrlichiosis), E. chaffeensis (HME), Rickettsia rickettsii (Rocky Mountain spotted fever), and Rickettsia typhi (endemic typhus). E. canis had been used as an antigen for the diagnosis of HME until 1991, when E. chaffeensis was isolated and maintained in cell culture; the results were combined for the analysis (14). Titers to these additional antigens were obtained by review of CDC records. Occasionally, endpoint titration had not been done, and for these samples the highest titer obtained was used in the calculation of geometric mean titers (GMT). FITC-labeled goat anti-human conjugate specific to the heavy and light chains of human IgG was used for testing against these antigens by IFA.

Case definition. The surveillance definition for human ehrlichiosis requires the presence of an illness clinically compatible with human ehrlichiosis with laboratory confirmation (10). Laboratory confirmation of human ehrlichiosis requires the presence of at least one of the following: a fourfold or greater change in the titer of antibody to Ehrlichia spp. antigen by the IFA test in paired acute- and convalescent-phase serum samples (including a change in titer of <64 to 64); a positive PCR assay result; and identification of intracytoplasmic morulae in stained blood, bone marrow, or cerebrospinal fluid leukocytes together with an IFA titer of 64. Because of incomplete data on the forms accompanying the samples, it was often difficult to determine whether an illness compatible with human ehrlichiosis was present. Signs and symptoms for rickettsial and ehrlichial diseases are similar in many regards (e.g., headache, fever, malaise, arthralgia, thrombocytopenia, and elevated levels of transaminases). In this report, we have assumed that because the samples were submitted for testing against a panel of rickettsial and ehrlichial antigens, the patient was ill with symptoms that were clinically compatible with human ehrlichiosis.

Epidemiologic comparisons. Epidemiologic information was often incomplete or absent from the forms accompanying samples; however, most reports included demographic information (sex, age, state of origin). Clinical information (date of onset, signs and symptoms, laboratory findings) was reported less often. Because usually only the presence and not the absence of signs and symptoms was noted (i.e., the denominator was unspecified), we report these findings only on the basis of frequency of reporting and not on the basis of percentages of occurrence in all cases.

	RESULTS

Top Abstract Introduction Materials and methods Results Discussion References

Number of confirmed and probable cases of HGE. A total of 2,251 serum samples were obtained from 1,602 individuals for HGE testing. A total of 142 individuals (8.9%) had at least one serum sample that reacted with the HGE agent antigen at a titer of 64. Titers of 64 were obtained from 184 (78%) of the 235 samples that were available for testing from these 142 individuals. Nineteen confirmed and 59 probable cases of HGE were detected. Among the 19 confirmed cases, 15 (79%) had fourfold or greater increases in titer. Twelve of the first samples from these confirmed cases were seronegative (IFA titer of <64), and three were seropositive (titers of 64, 512, and 512). Titers obtained from the second samples of these 12 patients had a GMT of 706 (range, 64 to 4,096), and titers obtained from the second sample of the three individuals who were initially seropositive rose to 1,024, 2,048, and 4,096, respectively. The four remaining confirmed cases had fourfold or greater decreases in titer. The first samples of these individuals had a GMT of 1,448 (range, 128 to 65,536), and the second samples had titers of <64, <64, 512, and 1,024. It was uncommon for more than two samples to be submitted from the same individual, as only 52 third samples and 7 fourth samples were identified.

Geographic distribution of cases of HGE. Confirmed or probable cases of HGE were detected in samples submitted from 21 states and from the District of Columbia (Table 1). Most of the patients (47 of 78, 60%) resided in three states where HGE is endemic: Connecticut (n = 14), New York (n = 18), and Wisconsin (n = 15). However, samples from patients with confirmed or probable HGE originated from several states where HGE has infrequently or never before been reported (Georgia, Kentucky, Hawaii, Missouri, Montana, Oklahoma, Tennessee, Washington, and West Virginia) (Table 1). The overall prevalence of antibodies to the HGE agent among individuals with suspected rickettsial and ehrlichial illnesses from the seven states selected for retrospective surveillance was 7.7% (62 individuals with confirmed or probable cases of 809 individuals tested) and varied from 1.6% in California and Maryland to 23.7% in Connecticut (Table 1).

Approximate interval for antibody patency of HGE cases. The time of disease onset was reported for 46 of the 78 patients with confirmed or probable HGE. Eleven (65%) of 17 individuals who had serum samples submitted during the first week following onset of illness had positive samples, as did 10 (91%) of 11 individuals who had samples submitted during the second week (Fig. 1). These individuals represented 14 and 13% of the confirmed or probable cases of HGE, respectively. Every sample from patients with confirmed or probable HGE that was submitted after 2 weeks of illness was seropositive, with the exception of two samples from individuals who became seronegative by day 21 (doxycycline given on the day of onset) and day 178 (no data for antibiotic therapy reported) following the onset of illness. Overall, 91% (n = 42) of the confirmed or probable cases of HGE for which the date of onset of symptoms was reported were diagnosed by the identification of one or more positive serum samples within the first 3 months following the onset of the disease. However, in four cases, the first serum samples received at the CDC were positive and were obtained at 92, 254, 277, and 344 days after disease onset (Fig. 1).

View larger version (15K): [in this window] [in a new window]   FIG. 1.   Development of titers of antibody to the HGE agent in 46 patients with confirmed or probable HGE for whom data were available. The solid bars represent the GMT for the indicated periods.

Seasonal distribution of HGE cases. The earliest onset of illness for a patient with probable or confirmed HGE was recorded in April, and a strong summer peak followed, as 21 (46%) of the 46 patients for whom the date of onset was reported occurred in June and July (Fig. 2). Onset of illness was reported throughout the fall and through December; no cases were identified from January through March.

View larger version (15K): [in this window] [in a new window]   FIG. 2.   Distribution of confirmed and probable cases of HGE in the United States by month of onset (n = 47).

Demographic, clinical, and exposure histories for HGE patients. Among the 78 patients with confirmed or probable HGE, the average age was 56.8 years (n = 65; median, 61 years; range, 3 to 88 years). Males accounted for 53 (67.9%) of the 78 patients with HGE.

Reactivity to other antigens among HGE-positive samples. Reactivity to one or more rickettsial and ehrlichial antigens was noted among the 184 samples that were positive for the HGE antigen (Table 2). The greatest reactivity among those samples that were positive for the HGE antigen occurred with the E. chaffeensis antigen, both in percentage of samples positive at titers of 64 (53.8%) and in GMT (694). Reactivity with both the HGE and E. chaffeensis antigens occurred in samples from 74 individuals, 10 (13.5%) of whom had confirmed or probable HGE. The remaining 64 individuals were classified as having either confirmed or probable HME (n = 34) or human ehrlichiosis (agent undetermined) (n = 30) (see below).

Possible dual infections with other agents. Reactivity to the HGE agent and to either C. burnetii, R. rickettsii, or R. typhi antigens was less frequent (7 to 9% of the HGE-positive samples) than was reactivity to both the HGE agent and the E. chaffeensis antigens (54 of the HGE-positive samples); the GMT to these antigens were also lower (Table 2). However, in some instances, the titers to these agents approached or exceeded those to the HGE agent. Under different study criteria, these individuals would be interpreted as having evidence of current or past infection with a different agent. The HGE patient from West Virginia had an antibody titer of 4,096 to the HGE agent and an antibody titer of 512 to C. burnetii (phase II) in the same sample. Two HGE patients from New York had identical maximum titers of antibodies to both the HGE agent and C. burnetii of 512, and a third patient from New York had a titer of antibody to the HGE agent of 128 and a titer of antibody to C. burnetii (all phase II) of 512. The HGE patient from New Jersey had a titer of antibody to the HGE agent of 64 but also had a titer of antibody to R. rickettsii of 2,048. The patient from Pennsylvania had maximum titers of antibody to the HGE agent and to R. rickettsii of 128 and 512, respectively, while another patient from New York had corresponding titers of 256 and 2,048, respectively.

	DISCUSSION

Top Abstract Introduction Materials and methods Results Discussion References

We have previously shown that the lack of a serologic assay caused cases of HGE to be missed (38). The present study identified 78 confirmed or probable cases of HGE from 21 states. Most of these cases were identified by retrospective analysis of samples submitted from four states where HGE is known to be endemic: Connecticut, Minnesota, New York, and Wisconsin (4, 5, 9, 11, 26). However, patients with antibodies reactive to the HGE agent were identified among specimens received from several states where this disease has not previously been identified or has been infrequently reported (Georgia, Hawaii, Kentucky, Maine, Missouri, Montana, Oklahoma, Tennessee, Texas, Washington, and West Virginia). Because travel histories were not provided for most of the patients from the latter states, it was not possible from the report forms to determine where all exposures might have occurred and therefore whether HGE is actually endemic to the state from which the sample was submitted. In two instances, travel and tick exposure histories were noted on the submitted forms. The single patient with HGE from Texas reported an exposure to ticks in Minnesota, and the patient from Maine experienced tick exposure in Massachusetts, both states of known endemicity for HGE (4, 48). Similarly, an HME patient from Hawaii had been bitten by ticks while travelling in Texas. These cases indicate that the determination of HGE and HME endemicity will require more detailed case investigations and clearly illustrate the need for health providers to be alert to travel and exposure histories. As the human ehrlichioses are becoming notifiable in many states (11), it is hoped that the collection of information of this kind will become routine.

Of interest, seven cases of HGE were identified among Florida residents, and cases of HGE were found to have occurred there as early as 1988. Serologic evidence of granulocytic ehrlichial infections in rodents has been reported from Florida (39). E. equi was reported from a Florida horse infested with I. scapularis in 1984 (8). The occurrence of additional HGE cases in Florida suggests that the HGE agent may be endemic to that state and is possibly transmitted by local populations of I. scapularis. HME also occurs in Florida, and we identified eight additional cases of HME there. Amblyomma americanum is thought to be the vector of E. chaffeensis (3, 19, 32, 33).

The extensive dual reactivity to the HGE agent and E. chaffeensis antigens and the occasional reactivity to other antigens suggest that some infected individuals might have been exposed to more than a single rickettsial agent. Reactivity to the HGE agent likely represents cross-reactive antibodies resulting from infection with the other agents (37, 50). As previously mentioned, the presence of HGE in certain states is not accurately documented in the present study because of the lack of travel and exposure histories. As an example, the "case" of HGE from New Jersey was in all probability a case of Rocky Mountain spotted fever (titers of antibody to R. rickettsii of 2,048 and 256 in the first and second samples, respectively) with the presence of low titers of antibodies (titer of 64 in both samples) to the HGE agent. A single, probable case of HGE, based on a single titer of 64, was detected in California, where HGE has been previously reported from the northern part of the state (22, 23) and where E. equi is endemic (25, 34, 35). Probable cases of HGE in Hawaii, Kentucky, Missouri, Montana, Oklahoma, Tennessee, and Washington were based on maximum titers of 64 or 128. The problem of potential cross-reactivity of antibodies, the possibility of dual or past infections, the clinical and epidemiologic similarities of the illnesses, and the difficulty in obtaining accurate travel and exposure histories from some patients indicate that additional testing will be required to confirm the presence of HGE in these states.

There are limitations in ascribing etiology based on the comparison of titers obtained by using different IFAs and different antigens (E. chaffeensis, the HGE agent, and previously E. canis and E. equi). However, when varied reactivity to multiple ehrlichial antigens is observed, etiologic determination based on fourfold differences in titers of antibodies to different antigens has been supported by PCR studies (13) and serologic data (28, 38). Taking these limitations into account, the majority of the data generated on HGE in this report are consistent with what is currently understood about its geographic distribution and clinical presentation (5, 11).

Although clinical histories were missing or incomplete for many of the patients, the most commonly observed clinical signs, symptoms, and clinical laboratory findings for HGE in the present study (fever, headache, myalgia, thrombocytopenia, and leukopenia) were similar to those reported previously (1, 4, 5). Cases identified as HGE occurred more commonly in males (67.9%) than in females, a finding that is also seen with patients who have been diagnosed with HME (17). The median age of the HGE patients in this study was 61 years (mean, 57), a finding consistent with reports from Connecticut and New York describing the median and mean ages of patients with HGE as 50 years (1, 9, 11). The median age for cases of HME has been younger (44 years) (20). Why HGE occurs or is recognized in older individuals more commonly than HME is currently not understood, but differences in the levels of severity of the disease caused is a probable explanation.

Among the 142 individuals in the present study, dual reactivity occurred in cases of HME (34 of 207 confirmed or probable cases of HME, 16.4%) slightly more often than did dual reactivity among confirmed or probable cases of HGE (10 of 78 confirmed or probable cases of HGE, 12.8%). This observation was also reflected in the higher GMT of antibodies to E. chaffeensis antigen (694) compared with that of antibodies to the HGE agent (306) among the dually reactive samples. Assessment of the proportion of dually reactive serum samples among cases of HGE and HME in this study should be interpreted with caution because of potential biases in the selection of samples that were included. Oversampling of specimens from states where HGE is endemic may have been offset by the inclusion of samples from all states in prospective testing, but we made no effort to correct for this sampling scheme. Regardless of this potential bias, significant dual reactivity occurred in cases of both HGE and HME in the present study.

The 37 serum samples that were less than fourfold different in titers of antibodies to the HGE agent and to E. chaffeensis antigens resulted in our inability to ascribe etiology for 30 (21%) of the 142 individuals positive for HGE in this study. Thus, dual reactivity among the serum samples caused significant diagnostic confusion in determining exact etiology in some suspected cases of ehrlichiosis. Although this limitation is important for elucidating the epidemiologic distinctions between HGE and HME, it has no impact on the practical treatment of ehrlichiosis, as doxycycline is the drug of choice for both diseases (11). One potential issue for surveillance purposes is that the current Council of State and Territorial Epidemiologists' case definition for ehrlichiosis does not directly address the issue of serum samples that react to both the HGE agent and E. chaffeensis (10). The potential for simultaneous seroconversion to both agents, as well as the unique reactivity to a single antigen among many specimens, underscores the need to test serum samples against both ehrlichial agents when suspected ehrlichiosis cases are confirmed by IFA. This requirement may be essential in states where both agents and their respective tick vectors occur (e.g., Connecticut, Florida, Maryland, and New York). As previously illustrated by case histories (including travel), a general awareness of the limitations of serologic testing is important.

Although tick exposure or bite was indicated in only one-fourth of the probable and confirmed HGE cases, the peak in seasonal distribution of cases coincides with the peak in activity of host-seeking nymphs of I. scapularis in the northeastern United States (45). These data are consistent with the hypothesis that the nymphal stage of I. scapularis is the most important life stage in the transmission of the HGE agent in areas where this vector occurs (5). Tick bites being unreported for patients with a tick-borne disease is a widespread phenomenon and is well documented in studies of Lyme disease (46) and Rocky Mountain spotted fever (27). Cases of HGE were also identified in the late fall and early winter into December, which coincides with the onset of the activity of adult I. scapularis in the Northeast (45) and upper Midwest (29). The late fall and early winter transmission of the HGE agent is in contrast to the transmission cycle of E. chaffeensis. Reports of illness associated with E. chaffeensis infections begin in April, peak in midsummer (May through July), and generally cease by November (16, 20). This pattern reflects the seasonal activity of A. americanum (44).

In addition to the traditional means of exposure to the HGE agent (via a tick), exposure to deer blood was a possible mechanism of infection for one individual who sustained a cut while dressing a deer. Exposure to deer blood has been hypothesized to be the route of exposure for three men who each butchered >250 deer carcasses in the upper Midwest in the 2 weeks prior to their illnesses (6). Although granulocytic Ehrlichia infection can be common among white-tailed deer in disease-endemic areas (7), the role of deer as competent reservoir hosts of the HGE agent is still undefined.

Over 400 cases of HGE have been described from the United States. Most cases of HGE have occurred in the upper midwestern states of Minnesota and Wisconsin and in several states in New England (Connecticut, Massachusetts, New York, and Rhode Island). HGE has been less commonly reported from other regions, such as the West and Southeast (e.g., northern California and Florida). The HGE agent has been present and presumably responsible for disease in at least two widely separated locales in the United States (Florida and the upper Midwest) for at least a decade. Because the disease is easily treatable, yet potentially fatal, an increased index of suspicion on the part of physicians is of paramount importance in efforts to reduce morbidity caused by this agent. Increased surveillance for ehrlichiosis, as indicated by trends toward making these diseases nationally notifiable, should provide needed information on the geographic extent of affected areas and insights into variations in the epidemiologies of these diseases across ecologically diverse regions.