In 1988, a published series of 33 Australian patients with abdominal pain and blood eosinophilia included a patient with ileal obstruction caused by an eosinophilic phlegmon to which a hookworm was attached [7]. Histologic sectioning precluded specific identification of the parasite. Because human hookworm infection, a persistent Third World problem, is not endemic in urban Australia and has virtually disappeared from the aboriginal communities of Queensland [8, 9], it was speculated that the cases had a common cause, possibly a dog hookworm. Shortly afterwards, two other cases of infection with a single adult A. caninum were reported, both detected at colonoscopy and one associated with eosinophilic ileitis [10, 11]. We developed an enzyme-linked immunosorbent assay (ELISA) and Western blot [12, 13] that are helpful in the diagnosis of abdominal pain with blood eosinophilia and indicate that the parasite is a common cause of this syndrome in eastern Australia.

We describe six additional cases of human enteric infection with A. caninum and summarize pertinent features, including serologic findings, of all nine Australian cases so far identified.

Methods

Top Methods Results Discussion Author & Article Info References

From 1985 to 1992, five patients were diagnosed by us and four others were referred with single hookworm infections. They lived either in Townsville (latitude, 19 degrees S) or Brisbane (latitude, 27 degrees^S). In each case, the medical records and biopsy tissue were provided by the attending clinician and pathologist. At least one serum sample collected within 4 weeks of diagnosis and the formalin- or alcohol-fixed worms were also available from all except one patient (patient 1).

Measurements

Clinical and demographic data were sought from the records and by interview. A complete blood examination, total serum IgE assay, and fecal microscopy were done in commercial or teaching hospital laboratories. Blood eosinophilia was ascribed to a count higher than 0.50 x 10⁹/L, and the upper normal value for IgE was 288 µg/L. Tissue biopsy specimens were examined and reported by histopathologists not associated with the study. Hookworms were examined microscopically by experienced parasitologists and were drawn and measured using standard techniques. Species identification was based on accepted criteria, which included overall length; esophageal length; body width; number of teeth in the buccal cavity; and, in male worms, length of spicules and structural details of the copulatory bursa [14, 15].

Sera were stored at –15°C and tested, among numerous control sera, for IgG and IgE antibodies by one of the investigators who was unaware of the patient histories. The ELISAs incorporated excretory-secretory antigens from adult A. caninum [12]. In the Western blots, these antigens were separated electrophoretically on polyacrylamide gels and transferred to nitrocellulose paper, which was then incubated with patient sera diluted 1:10 [13]. Controls were sera from blood donors in the state of Tasmania, where A. caninum does not exist. The cut-off absorbance levels for positivity in IgG and IgE ELISAs were 0.147 and 0.155 optical density units, respectively. Both the IgG and IgE Western blots were reported as positive when a protein band (Ac68) with a molecular weight of 68 kd was recognized.

Results

Top Methods Results Discussion Author & Article Info References

All patients were white; six were men and three were women, with a mean age of 48 years (range, 30 to 61 years). Patients 1 to 3 were diagnosed from 1985 to 1990; the others were diagnosed from 1991 to 1993. Each patient lived in a freestanding house with lawns and garden and kept one or more dogs as pets. By gardening or walking outside barefooted, all patients had been potentially exposed to soil contaminated with canine feces. Except for patient 7, none had recently travelled to an area endemic for human hookworm. None had had atopic or immunologic diseases, and none reported cutaneous or respiratory symptoms that may occur with human or zoonotic ancylostomiasis [1]. After removal of the worm, all patients recovered and have remained well (Table 1).

Patients 1 to 3

The clinical features of these patients have been published previously and are summarized in Table 1 (7, 10, and 11). A 61-year-old woman (patient 1) with a 4-week history of recurrent bowel obstructions had a laparotomy with resection of an inflamed ileal segment. A single hookworm was attached to the mucosa and was sectioned for histologic study, which precluded a species identification. Her case was reported to one of us several months after her recovery; serum had not been stored. A 50-year-old tradesman (patient 2) recurrent abdominal pain and a positive fecal occult blood test was diagnosed at colonoscopy. A feeding hookworm, species A. caninum, was removed from an inflamed and ulcerated terminal ileum. A 53-year-old male teacher (patient 3), having colonoscopy for vague abdominal pain and minor rectal bleeding thought to be caused by hemorrhoids, had a single feeding worm, species A. caninum, removed from his rectum. The colon and rectum appeared healthy.

Patient 4

A 41-year-old woman from Townsville described three attacks during the preceding 5 weeks of mid-abdominal pain, which aroused her from sleep, lasted a few hours, and did not require urgent care. Her stools were loose and more frequent after each episode but she felt well between attacks. Results of a physical examination, abdominal ultrasonography, and upper gastrointestinal endoscopic examinations were normal. Colonoscopy, completed 20 cm into the ileum, showed two ulcers, 2 mm in diameter, in the cecum. A feeding nematode, 5 cm proximal to the ileocecal valve, was removed with biopsy forceps (Figure 1). Damage to the buccal cavity prevented specific identification of this 8-mm long, male hookworm, with spicules 750 and 785 microns long. Tissue obtained from near the worm and the cecal ulcers was excessively infiltrated with eosinophils and plasma cells.

View larger version (134K): [in this window] [in a new window]   Figure 1. Feeding hookworm in ileum (patient 4). The worm, seen at colonoscopy, was feeding vigorously with blood present in and expelled from its gut. Localized hemorrhage into the mucosa appears at the point of attachment. The buccal capsule, necessary for morphologic identification, was damaged when extracted with biopsy forceps.

Patient 5

A 45-year-old Brisbane man, with mild chronic dyspepsia that had not been investigated, was hospitalized after 24 hours of increasingly severe colic. He was afebrile and his abdomen was distended and tender; radiologic findings showed dilated loops of small bowel with fluid levels. A complete blood examination showed initial neutrophilia (10.25 x 10⁹/L), with a normal eosinophil count of 0.37 x 10⁹/L. After early improvement, the signs worsened; a laparotomy was done 36 hours later and showed increased serous peritoneal fluid and a 24-mm, inflamed ileal stricture, which was resected. An 8-mm long, female hookworm, without eggs, was attached to the lumen of the opened segment, but damage to the buccal capsule precluded its specific identification. Histologic findings showed intense eosinophilic ileitis. Mild blood eosinophilia (0.68 x 10⁹/L and 0.76 x 10⁹/L) was recorded 4 and 14 days after admission and resolved (0.15 x 10⁹/L) by 4 weeks.

Patient 6

A 30-year-old Townsville woman complained of anorexia, abdominal pains, diarrhea, and a 6-kg weight loss during a period of 4 weeks. She described the pain as a persistent, dull ache with severe exacerbations lasting 5 minutes, often provoked by eating. Her normal bowel habit, three formed stools daily, had increased to 10 liquid movements, often at night; passing stool gave temporary pain relief. A clinical examination was normal, but the blood eosinophil count was 2.96 x 10⁹/L. The results of a fecal culture and microscopic examination were normal. The colonoscope was passed approximately 20 cm into the ileum. The only abnormal finding was a hookworm feeding in the cecum; it was removed with biopsy forceps and identified as a 10-mm long, adult male A. caninum, with spicules 785 and 800 microns long (Figure 2). Random ileal and cecal biopsy samples showed intense eosinophilic inflammation. Three weeks later, her blood eosinophil count (0.26 x 10⁹/L) was normal.

View larger version (155K): [in this window] [in a new window]   Figure 2. Scanning electronµgraph (patient 6). Anterior extremity of formalin-fixed, 10-mm hookworm, clearly showing a pair of three ventral teeth in the buccal cavity; these plus spicule size clearly identify the specimen as A. caninum (original magnification, x 700).

Patient 7

A 51-year-old Brisbane man had a laparotomy for suspected acute appendicitis. He had presented 3 days after developing generalized abdominal pain, which improved initially but became severe again just before assessment. Six weeks earlier, he had returned from a 3-week vacation in Vanuatu, in the South Pacific, where he had walked barefooted and had observed many freely roaming dogs and extensive environmental fouling with their feces. He bred and trained greyhound dogs professionally, treating his animals regularly with anthelminthic agents. A clinical examination showed maximum tenderness over the right iliac fossa, with absence of bowel sounds. The operative findings were peritonitis, with slightly turbid fluid (culture sterile; cells not identified) and an intensely inflamed ileal segment 60 cm proximal to the ileocecal junction, surrounding a 2-cm thickened area that at first appeared to be a tumor. The ileum was opened to show a small ulcer that, histologically, was intensely inflamed and infiltrated with eosinophils. An 8-mm long, female A. caninum, without eggs, was attached to the center of the ulcer.

Patient 8

A 50-year-old Townsville man with atherosclerosis was hospitalized with a 6-hour history of severe abdominal pain coming in waves lasting several seconds. He had noticed a mild increase in stool frequency during that day and had had bright rectal bleeding occasionally during the preceding 12 months. A clinical examination showed abdominal distension but little tenderness. Ascites, unexplained by liver, cardiac, or renal disease, was later confirmed by computed tomography, and abdominal radiography showed multiple fluid levels but no bowel dilation. A complete blood examination showed neutrophilia (19.1 x 10⁹/L). His symptoms resolved completely within 12 hours, and he was discharged, feeling well, after 2 days. At colonoscopy 3 weeks later, a feeding hookworm was removed with biopsy forceps from the mid-transverse colon. The terminal ileal and colonic mucosa looked normal, except for 7- and 12-mm dysplastic polyps, which were removed from the sigmoid colon. Biopsy specimens from the feeding site showed normal histologic findings. The worm was a 7-mm long, female A. caninum without eggs. A repeat complete blood examination, including an eosinophil count of 0.16 x 10⁹/L, was normal.

Patient 9

A 54-year-old Townsville man, asymptomatic except for anxiety about bowel cancer, had a colonoscopy. Sixteen days earlier, he had self-medicated with mebendazole, 200 mg daily for 3 days. His mother, assessed by one of us because of abdominal pain associated with blood eosinophilia, had responded to treatment with mebendazole and had advised her family to treat themselves regardless of symptoms. The colonoscopic findings were a few small ulcers in the terminal ileum, the largest being 5 mm in diameter, and a feeding hookworm 5 cm proximal to the ileocecal valve. The worm was a female A. caninum, measuring 10 mm; eggs were not seen. Ileal biopsy samples showed mild inflammatory change with 15 eosinophils per high-power field.

Serologic Findings

The serologic results from each patient are shown in Table 1. Both ELISAs were positive in four patients (patients 2, 3, 8, and 9), and four patients were negative by both tests. One or both Western blots tested positive in seven of eight patients (Figure 3). One patient (patient 6), who tested negative by both IgG and IgE Western blots, also had negative ELISA results. The negative IgE Western blots (patients 6 and 7) had faint 68Ac bands, but the intensity was insufficient to grade other than borderline. The intensity of the Ac68 bands produced by IgG and IgE Western blots varied inconsistently within individuals.

View larger version (68K): [in this window] [in a new window]   Figure 3. Immunoglobulin G and E Western blots comparing cases with controls. Molecular weight markers in the left lane. Numerals represent patient numbers. C = control (blood donor from Tasmania) serum. Reaction with a band at 68 kd signifies a positive reaction; other bands are ignored.

Discussion

Top Methods Results Discussion Author & Article Info References

The major single morphologic feature distinguishing human from zoonotic hookworms is the tooth pattern. Three pairs of teeth, sometimes difficult to identify, characterize the two most common animal hookworm species in northeastern Australia, A. caninum of dogs and A. tubaeforme of cats [10], but are not found in human hookworm species. Although they were originally assumed to be the one species, they can be distinguished by careful morphologic studies [14, 15]. Even if the damaged specimens from our patients were not A. caninum, the circumstantial evidence suggests they were not anthropophilic species, and the significance of our finding is not diminished. The patients originated from populations free of human hookworms; patent hookworm infections (where eggs are produced) are now rarely seen in either Brisbane or Townsville, and then only in travellers from overseas (unpublished observation). Living conditions are among the most sophisticated in the world, with highly developed and efficient water supply and sanitation facilities. It is clear that pets sharing our domestic environment are the reservoir of infection, and all our patients had been in contact with soil or grass potentially contaminated with dog feces. Detection of these nonpatent, zoonotic hookworm infections has been greatly facilitated by the eradication of previously endemic human ancylostomiasis, coupled with the increasing use of colonoscopy.

We have not been able to identify any feature of the worms retrieved from our patients or from local dog populations, to suggest a novel species. All worms were immature: they had not attained the full adult length (14 to 20 mm for females, 11 to 13 mm for males), and the females were not producing eggs [15]. Further, these worms were all unusually low in the gastrointestinal tract, even in the patients who had surgery; in host-specific hookworm infections, the parasites aggregate in the jejunum [1]. Failure to develop fully may reflect poor adaptation to the human host and premature expulsion from the bowel by the intense inflammatory response (along with the preparation for colonoscopy).

Infection probably caused illness. Although their symptoms ranged in severity, our patients had common clinical features including pain (sometimes associated with obstruction or peritonitis), diarrhea, abdominal distension, weight loss, and rectal bleeding. Blood eosinophilia was usual but not universal, and it resolved with symptoms. Along with the elevated IgE levels, these findings are consistent with other enteric parasitic infections, including the human-specific hookworms A. duodenale and N. americanus [16, 17]. Subclinical infection, confirmed by serendipity in one patient (patient 9), may be common.

Hookworm detection, when infection is by a single, tiny organism, is difficult and approaches a chance finding. Despite our practice of routinely inspecting the ileum, the colonoscopic detection rate is likely to be low because the ileal segments found inflamed at laparotomy were 15 cm or more proximal to the ileocecal junction, beyond the reach of a conventional examination instrument. The clinical and laboratory features of the patients who had surgery, along with the operative findings of segmental inflammation with free intraperitoneal fluid, are virtually identical with those described in a series from Townsville and Brisbane, in which hookworms were not found but were suspected [18]. Obscure abdominal pain, often associated with blood eosinophilia and responsive to treatment with mebendazole, is common in northeastern Queensland where we have investigated over 200 cases during the last 10 years (unpublished observation). We have linked this syndrome epidemiologically [7, 10] and serologically [12, 13] to occult infection by canine hookworms.

The histologic finding is identical with eosinophilic enteritis. This entity, often with gastric involvement, differs from our cases where inflammation was confined to the distal small bowel and colon. By convention, eosinophilic enteritis is idiopathic, specifically not caused by parasitic infection [19]. Whereas classic enteric helminthiases associated with eosinophilic infiltration of the gut, such as ancylostomiasis, schistosomiasis, acute trichinelliasis, anisakiasis, capillariasis, angiostrongyliasis, and, sometimes, strongyloidiasis, can be readily diagnosed by fecal microscopy, routine endoscopy, or serologic tests, infection by A. caninum does not yet fit into this category. Fecal microscopy is likely to be negative, the worm is usually beyond the reach of endoscopy, and serologic tests are still being developed. This parasite is now implicated as the cause of the variant of eosinophilic enteritis commonly diagnosed in Queensland.

The heterogeneous clinical manifestations suggest a marked degree of host idiosyncrasy ranging from immunologic tolerance to a type 1 hypersensitive reaction, which even appears to vary within individuals. One patient (patient 8), presenting with acute obstruction and ascites, had been free of symptoms for 3 weeks before his colonoscopy. Presumably, the initial symptoms had resulted from localized bowel inflammation in response to worm secretions, either by the worm found at colonoscopy or, more likely, a predecessor, which had been expelled as a consequence of the immune reaction. The specific allergen might be secreted only by more advanced worms; this one (7 mm) was the smallest in the series. Alternatively, secretions of individual worms may differ qualitatively, so that not all induce an inflammatory reaction. Another possibility is that different parts of the gastrointestinal tract vary in their allergic reactivity.

The positive rate for the ELISA in this series was a surprisingly low 50%. In the Townsville population, 70% of patients with suspected A. caninum infection (presenting with obscure abdominal pain associated with blood eosinophilia) compared with 8% of healthy controls had positive ELISA values; we have observed a positive correlation with chronicity of symptoms and falling values during convalescence (unpublished observation). Patients with primary and acute symptoms have lower values, a situation that might be more representative of the persons with proven infection reported. Although the ELISA incorporates a wide range of antigens, only one of which has been found to be diagnostic by Western blot (Ac68), it is specific for hookworm as opposed to other nematode infections, but it does not discriminate among different hookworm species [12]. The positivity rate of the combined Western blots was 87.5% compared with 10% in Townsville patients with nonparasitic gastrointestinal diseases (unpublished observation), higher than the ELISA. The ELISA readings did not consistently correlate with band intensities on Western blots. Although IgG and IgE ELISA readings correlated within individuals, this association was less evident in the Western blots. It is not clear how these detectable antibodies relate to the gut inflammation and why a variation in the dominance of the different classes of immunoglobulin should exist among patients. In its current form, serologic testing has only limited diagnostic application.

Presumably, human exposure to infective larvae of A. caninum is related to its prevalence in local dog populations. In Brisbane, hookworm infection in impounded suburban dogs was 22% in winter and 38% in summer and in Townsville, the rates (around 50%) are higher (unpublished observation). These rates, however, are not exceptional; even in the more temperate climate of New Jersey, a 6-year survey found 24% of dogs to be infected [15]. Perhaps the differences in canine infection rates are less clinically significant than human behavioral adaptations to climate in determining exposure to infection, such as contact with contaminated soil and grass. Nevertheless, as most of the world's people share their environment with A. caninum-infected dogs, canine hookworms could be encountered widely as a human enteric parasite [20].

Infection by A. caninum, proven and otherwise, undoubtedly explains some cases of ileal eosinophilic enteritis. As more clinicians and pathologists become aware of this phenomenon and its manifestations, enteric infections by hookworms acquired from pets will become recognized as a major zoonosis.