Human Ehrlichiosis in Adults after Tick Exposure: Diagnosis Using Polymerase Chain Reaction

Methods

Patients were sought by asking house officers to notify the infectious diseases division about patients who had the previously published symptoms or had signs of human ehrlichiosis [6]. However, with rare exception, the diagnosis was first suggested through our routine consultation service. Patients agreed to the study by oral consent as approved by the institutional review board. History and physical findings were reviewed and recorded. Blood specimens were obtained, and most patients started receiving doxycycline (Doryx, Vibramycin, and Monodox), 100 mg twice daily intravenously or orally, depending on their ability to tolerate oral intake. Hospitalized patients were followed on a daily basis until the illness resolved. Patients were urged to return to clinic in 2 to 4 weeks but not all returned because many lived long distances from our health care facility.

From 1989 to 1991, most serologic testing was done by the Centers for Disease Control and Prevention using an indirect fluorescent antibody technique and E. canis as the antigen [9]. During 1992, an indirect fluorescent antibody method was done using E. chaffeensis antigen. From 1990 through 1991, citrated blood was collected and frozen at −70°C in anticipation of developing a PCR technique. In 1992, blood was collected in ethylenediamine tetra-acetate-treated tubes and was prepared for PCR within 24 hours. Polymerase chain reaction was done on 200 µL of whole blood using the method of Anderson and colleagues [10]. Oligonucleotide primers HE1 and HE3 were reported by these authors to be homologous to sequences within the 16S rRNA gene of E. chaffeensis. The primer set (HE1, HE3) allowed the amplification of a 389-basepair product from E. chaffeensis DNA but did not amplify products from E. canis, E. ewingii, E. phagocytophila, E. sennetsu, or E. risticii[10].

After amplification, the PCR products were subjected to electrophoresis on 1% agarose gels, stained with ethidium bromide, and visualized directly with ultraviolet light. In some experiments, PCR products were Southern blotted onto GeneScreen nylon membranes (DuPont Company, NEN Research Products, Boston, Massachusetts [11]). The membrane-bound PCR products were prehybridized for 2 hours and then hybridized at 5 to 10 °C below the dissociation temperature for 2 to 16 hours with oligonucleotide HE1, which had been radiolabeled (by kinasing) with (Phosphorus-32)dATP [12]. After three washes for a few minutes at room temperature, the filters were washed for 5 minutes at 5 to 10 °C below the dissociation temperature and were visualized after autoradiography. We found that radioactive detection of positive samples provided the most accurate means of detection because positive PCR products were not always visible after staining with ethidium bromide. Another way in which PCR products were sometimes detected using radioactive label was facilitated by the addition of 35SdATP (DuPont Company) to the PCR reaction. The final specific activity was 0.2 microcuries/µL. The products were subjected to electrophoresis on 1% agarose gels, Southern blotted onto nylon membranes, and visualized after autoradiography.

Initially, to establish controls, we added the primers HE-1 and HE-3 to purified DNA from E. chaffeensis and Rickettsia rickettsiae, and we did PCR. Strong amplification was noted of a 389-basepair product to the E. chaffeensis DNA, but no amplification occurred to R. rickettsiae. Subsequently, we used whole blood from a patient with a highly positive band as a positive control for unknown samples. Negative controls consisted of whole blood samples from healthy persons.

For the purposes of this report, we considered a bonafide case of human ehrlichiosis to be one where the patient had a fourfold increase or decrease in antibody titers to E. canis or E. chaffeensis or a positive PCR result or both. Patients were excluded who failed to meet these laboratory criteria.

Results

General

We identified 30 patients with human ehrlichiosis between 1989 and 1992. Twenty-three patients were admitted to the hospital because of their degree of illness or occasionally for perplexing illness. Admitting diagnoses included influenza, rickettsial disease, urinary tract infections, renal colic, urosepsis, epididymitis, pneumonia, viral infection, gastroenteritis, herpes encephalitis, meningitis, and fever of unknown origin. Ehrlichiosis was not considered to be the admitting diagnosis for any patient.

There were 20 men and 10 women ranging in age from 22 to 79 years. Twenty-six had a history of attached ticks, 1 had a history of removing ticks from her dogs, and 3 recalled no tick exposure. Most of our patients received numerous tick bites in the preceding weeks; therefore, we were unable to define an incubation period for the illness. The median time from onset of illness to the seeking of medical care was 3 days (range, 1 to 6 days). The time elapsed for consideration of the diagnosis by a physician ranged from 0 to 30 days.

Many patients received antimicrobial agents before diagnosis, including cephalosporins, fluoroquinolones, aminoglycosides, penicillins, lincomycins, sulfonamides, antituberculous drugs, or combinations of these drugs. We could not discern a beneficial effect in any of the patients using these classes of drugs. However, a comparative trial would be necessary to conclusively confirm this observation. Two patients who received chloramphenicol improved clinically within 24 to 48 hours but were subsequently treated with doxycycline.

Clinical Symptoms and Signs

The clinical symptoms and signs of the patients are shown in Table 1. The patients had essentially the same findings on their first contact with a physician. Physical findings other than fever and rash were uncommon. The rashes were pleomorphic, consisting of either a transient mottled or diffuse erythema in 5 patients, a fine petechial eruption on the ankles in 2 patients, macular rash in 1 patient, papular rash occurring on the second or third day of treatment (which resolved despite continuing drugs) in 1 patient, and urticarial rash in 1 patient after sulfonamide treatment. Both patients with hepatomegaly or splenomegaly or both had a history of heavy ethanol intake. Mental status changes were mainly confusion or hallucinations or both and lethargy.

Laboratory Results

The abnormal laboratory findings are shown in Table 2. In addition to these findings, minimal hypocalcemia, trace to 1+ proteinuria, and mild anemia were common. Serologic and PCR results are shown in Table 3. To summarize the serologic and PCR findings, three patients with a fourfold or greater increase or decrease in antibody titer to E. canis had negative PCR determinations. These were not tested with E. chaffeensis because this antigen was not available at the time. Eight patients with no increase in antibody titers had positive PCR results, eight had a fourfold or greater increase or decrease in antibody titers to E. chaffeensis, and one to E. canis had a positive PCR result; three patients without acute and convalescent serologic results had positive PCR results. In seven patients no PCR was done.

Participants tested who had negative PCR determinations included those who had Rocky Mountain spotted fever, Q-fever, the myelodysplastic syndrome, Candida fungemia, alcoholic hepatitis, the acquired immunodeficiency syndrome, or an acute febrile illness for which no diagnosis was firmly established and healthy control persons. Two of the 30 patients (patients 17 and 19) had serologic titers done against E. canis and E. chaffeensis that showed no increase in titer to E. canis but showed an increase in titer to E. chaffeensis; both had positive PCR results to E. chaffeensis sequences.

Eight patients had cerebrospinal fluid examinations because of mental status changes, photophobia, or intense headache. Five had results compatible with meningeal inflammation. Cell counts varied from 10 to 110 × 10⁹/L and, with one exception, showed a lymphocytic pleocytosis. All had increased protein values ranging from 0.078 to 0.206 g/L. Glucose levels in cerebrospinal fluid ranged from 2.28 mmol/L to 3.55 mmol/L, with two patients having minimal hypoglycorrhachia. Spinal fluid that was not bloody was available from two of the five patients and was positive by PCR for E. chaffeensis sequences.

Twenty-five patients received chest radiographs. All except 2 had normal or unchanged radiographs. One patient, an asthmatic, had an atelectatic segment that cleared the following day, and 1 elderly patient had an exudative pleural effusion. It was unclear whether the effusion was related to her Ehrlichia infection. Peripheral blood smears from 10 patients were examined microscopically using Wright stain, and 2 had leukocyte inclusions compatible with Ehrlichia morulae. Three patients had bone marrow examinations as part of an evaluation for leukopenia and thrombocytopenia. The morphologic results were essentially normal except for multiple small noncaseating granulomas in 1 patient. In 1 patient, morulae were seen in the cytoplasm of mononuclear cells in the bone marrow aspirate.

Clinical Course

All patients except one were eventually treated with a 10- to 14-day course of doxycycline, 100 mg twice daily. The exception was a patient who spontaneously recovered after 30 days of a relapsing febrile illness. All patients recovered after therapy, and complications were few. Two elderly patients developed respiratory distress with evidence of pulmonary edema. One patient became hypotensive but remained asymptomatic, and another became afebrile but then confused on the third day of therapy. His cerebrospinal fluid examination was compatible with meningeal inflammation.

One patient was of particular interest because she was not diagnosed for 8 days. She developed signs of delirium, and her cerebrospinal spinal fluid showed evidence of meningitis. Doxycycline therapy was begun and her mental status waxed and waned during the next 7 days, after which she steadily improved and was discharged after a 14-day course of treatment. Six days later, she returned with a left central facial nerve palsy. Repeated testing of her cerebrospinal fluid showed continued pleocytosis and increased protein levels. An enzyme-linked immunospecific assay on blood and cerebrospinal fluid for Borrelia burgdorferi antibodies was not diagnostic. Nevertheless, she was treated with a 14-day course of ceftriaxone (Rocephin; Hoffmann-LaRoche Incorporated, Nutley, New Jersey). Her facial nerve palsy improved after 2 weeks and completely resolved 3 months after its onset.

Most patients responded quickly to doxycycline therapy both subjectively and objectively. Fourteen of 23 patients with serially recorded temperatures became afebrile in 2 to 3 days (range, 1 to 8 days). In those patients who were leukopenic on admission, the median time to an increase in leukocyte count after initiation of doxycycline therapy was 3 days (range, 1 to 6 days). The time to an increase in the platelet count was essentially the same as that for the leukocyte response. Fifteen patients developed lymphocytosis of 43% to 83% of the total leukocyte count between day 2 and day 19 after starting therapy.

Twelve of the patients with an abnormal liver profile had follow-up studies between days 7 and 31 after therapy. All had returned to normal or to near normal. Of the 15 patients with hyponatremia, all returned to normal within 3 to 7 days. Only 1 patient had transiently increased blood urea nitrogen and serum creatinine values, and none developed renal failure.

Discussion

Since the sentinel case of human ehrlichiosis was reported in 1987, 299 patients were diagnosed from 24 states through 1992. Such reports came from all over the United States (with the exception of the desert Southwest) with Oklahoma, Missouri, Virginia, Texas, Georgia, and Arkansas accounting for more than 200 of the cases (Dawson JE. Personal communication).

With rare exception, the clinical and laboratory data relative to human ehrlichiosis were gathered from retrospective serologic testing along with chart reviews and isolated case reports [2-6, 13-17]. The prospective study was primarily a serologic survey of febrile patients admitted to a Georgia hospital and a recording of their symptoms and laboratory data. This study resulted in the diagnosis of ehrlichiosis in 8 of the 75 patients enrolled. No statistically significant differences in clinical presentation could be found between case and noncase patients [13]. To our knowledge, our investigation included the largest group of patients to be studied prospectively.

Although not conclusively shown, human ehrlichiosis is thought to be transmitted by ticks. In our study and others [6], a strong association was noted between tick bites or exposure to ticks and the development of illness (P = 0.0001, Fisher exact test). Epidemiologically, the disease tends to occur in the geographical distribution of the tick Amblyomma americanum[6]. In a survey of 490 ticks from Fort Chaffee, Arkansas (using a dog anti-E. canis conjugate and a direct immunofluorescent technique), 4 ticks (2 A. americanum and 2 Dermacentor variabilis) yielded positive test results. Only 1 of the 4 ticks, a D. variabilis, contained E. chaffeensis sequences by PCR [9]. In our laboratory, using a conjugated polyclonal rabbit antibody to E. chaffeensis, 0.3% of 318 A. americanum ticks and 5.9% of 372 D. variabilis ticks were positive by indirect fluorescent antibody technique. Studies are under way to confirm these findings by more specific testing.

As in the retrospective studies, we found the presenting signs and symptoms to be nonspecific, with fever, chills, and headache being the most common along with many other symptoms that could lead the clinician to suspect other disease processes [3, 5, 6]. This is shown by the array of admitting diagnoses in our patients. In contrast to the retrospective studies, we did not find pharyngitis, respiratory symptoms, or lymphadenopathy to be prominent features of the illness [6]. We did find dysgeusia (impairment of the sense of taste) and photophobia, not previously highlighted, in a few patients. The most important historical finding as noted by others was a history of tick exposure [13]. Ninety percent of our patients had a clearcut exposure to ticks, usually multiple bites.

The most consistent laboratory findings reflected a disease process involving the hematologic, hepatic, and central nervous systems. Thrombocytopenia, leukopenia, and abnormal liver profiles were common and consistent with previous observations [3, 6, 13]. Other hematologic findings included lymphopenia and prolongation of activated partial thromboplastin times, increased fibrin split products, normal fibrinogen levels but rarely prolonged prothrombin times. No clinical evidence of excessive bleeding was noted. Further, the incidence of the abnormal clotting variables is probably biased in our observations because only the more seriously ill patients were tested.

Three of our patients had bone marrow examinations. Although bone marrow hypocellularity has been reported [2, 13], our patients showed essentially normal morphologic characteristics of the bone marrow. One patient did have noncaseating granulomas in the marrow, which has been previously reported [2]. Morulae were seen in the bone marrow of one of our patients.

The liver profiles tended to show only minimal-to-modest increases in transaminase levels, although they were occasionally in the range found in viral hepatitis. Alkaline phosphatase levels were less frequently increased as were total bilirubin levels, which were mainly direct bilirubin. Approximately one fourth of patients tested had hyperbilirubinemia. None of our patients had liver biopsies done; therefore, the nature of the hepatic disturbances remains largely unknown. In a previous case [18], a patient had severe jaundice, and the histologic findings showed a portal mixed cellular infiltrate, bile plugging, lymphocytosis in the sinusoids, and a neutrophilic cholangitis. No hepatic cell necrosis was seen [18].

Not previously well documented was our finding of frequent increases in lactate dehydrogenase levels. Only two patients had lactate dehydrogenase isoenzyme determinations; neither showed a predominant source for increased levels of lactate dehydrogenase. Hyponatremia occurred in more than 50% of patients without an identifiable reason other than acute infection. Urinary electrolyte levels and serum and urine osmolalities were tested in two patients; the results did not suggest a syndrome of inappropriate antidiuretic hormone secretion.

Eight of our 30 patients had sufficient symptoms to result in a cerebrospinal fluid examination by the primary physicians. Five of these 8 patients had abnormal findings of pleocytosis, increased protein, and, in two, minimal hypoglycorrhachia. Four of the 5 had a lymphocytic pleocytosis, whereas 1 had a slight neutrophilic pleocytosis. Three other cases have been reported [17, 19, 20] with details on spinal fluid findings; one of the patients had a neutrophilic pleocytosis. Glucose concentrations in cerebrospinal fluid were normal [17, 19, 20]. In one of these patients, morulae were seen in the cytoplasm of mononuclear cells in the spinal fluid and PCR showed the presence of E. chaffeensis sequences [20]. Two of our patients with meningitis had PCR done on spinal fluid that was not bloody; both of them were positive.

In contrast to a previous review [6], we did not find respiratory tract, lymphatic system, or renal involvement to be common in our patients. Only 2 of 25 patients had abnormal chest radiographic findings; in neither case could these findings be clearly associated with the Ehrlichia infections. Two patients did develop pulmonary edema after therapy was begun, probably because of excessive intravenous fluids. Although it is clear that renal failure may occur in some patients with ehrlichiosis [1, 6, 18, 20], only 1 of our patients had increased levels of blood urea nitrogen and serum creatinine and these levels returned to normal. No patient developed renal failure. The mechanism of renal failure in ehrlichiosis is unknown at present.

After the diagnosis was made and therapy was begun with doxycycline, patients rapidly responded, with subjective and objective improvement within 24 to 72 hours. Complications were few and included development of pulmonary edema in two patients, mental confusion on day 3 of therapy in one patient, and development of facial palsy in one patient. Although fatalities have been described in ehrlichiosis, all of our patients recovered [6]. We found no long-term sequelae in our patients; hematologic and liver variables returned to normal or to almost normal within 31 days. Lymphocytosis developed in 15 of 23 patients, usually on the third to fifth day of treatment (range, 2 to 19 days), which was as high as 83% of the total leukocyte count. The nature of this observation is currently under investigation.

We found that tetracyclines are effective treatment, which is consistent with previous observations [6]. Our patients were uniformly treated with doxycycline, 100 mg twice daily for 10 to 14 days. Although we only used chloramphenicol for a brief time in two patients, others [6] have found this to be suitable therapy, particularly in children. One isolate of E. chaffeensis was tested in vitro (using tissue culture) for susceptibility to antimicrobial agents. The results indicated susceptibility to doxycycline and rifampin. Conversely, using this technique, the isolate was resistant to chloramphenicol (an agent that appears effective in vivo) and to ciprofloxacin, erythromycin, trimethoprim-sulfamethoxazole, penicillin, and gentamicin [21].

During the time encompassed by this study, the methods used to confirm the diagnosis of ehrlichiosis changed. Initially, patients were tested only with E. canis antigen; later, they were tested with E. chaffeensis antigen and then ultimately with PCR. Two patients did not have increased antibodies to E. canis, but both had substantial increases to E. chaffeensis using the indirect fluorescent antibody assay. Twenty-three of the 30 patients had PCR done on whole blood samples, and 20 had positive results. Three patients who had substantial titer increases or decreases to E. canis antigen (not tested for E. chaffeensis) did not have amplified products to E. chaffeensis sequences. The reason for this is not entirely clear. Several possibilities exist: First, these patients had their specimens frozen for 2 years before testing, which could have produced a technical error. However, similarly stored samples were positive from this same period. In addition, these patients did not receive therapy before the specimens were obtained. Second, other undefined Ehrlichia species may exist that could result in serologic conversion but that have DNA sequences not detected by E. chaffeensis primers. Third, the PCR assay is not 100% sensitive for the diagnosis of all patients.

Conversely, eight patients with a clinical syndrome compatible with ehrlichiosis did not have a serologic conversion but had positive PCR results. In two of these patients, convalescent sera were obtained less than 2 weeks after the onset of the illness. Six had convalescent sera obtained 14 to 35 days after onset of their illness. In some patients, serologic conversion may take 4 weeks or longer.

Based on our data, we believe that PCR is a rapid and accurate method for diagnosing most patients with human ehrlichiosis. During this period of active surveillance, we only diagnosed three patients with Rocky Mountain spotted fever; two blood samples from these patients were studied with PCR and were negative for E. chaffeensis but were positive using nucleotide sequences from a strain of Rickettsia rickettsiae. Additionally, four patients who did not have serologic titer increases to other rickettsial antigens had a fourfold or greater increase to Q-fever antigen. One of the four studied by PCR was negative for E. chaffeensis.

Human ehrlichiosis is a more common disease than has been previously recognized. Many physicians are still unaware of or fail to recognize the disease even in geographic areas where it appears to be relatively common. No specific clinical or routine laboratory findings exist that are diagnostic; however, patients with febrile illnesses after tick exposure in certain geographic areas who have thrombocytopenia, leukopenia, or lymphocytopenia or abnormal liver profiles, or both, should be strongly suspected of having ehrlichiosis. The PCR assay is a rapid method for confirming the diagnosis of most patients with human ehrlichiosis. Until PCR techniques are more widely available, serologic confirmation is useful. To optimize serologic results, convalescent sera should probably be obtained at least 4 weeks after onset of the illness. However, the provisional diagnosis of ehrlichiosis, like several other tick-borne diseases, is a clinical diagnosis and we advocate early intervention with tetracyclines.