One of the most serious complications of typhoid fever is intestinal perforation. Of 27 patients admitted to a provincial hospital in the Mekong Delta region of Vietnam who had gastrointestinal perforation secondary to suspected typhoid fever, 67% were male, with a median age of 23 years and a median duration of illness of 10 days. Salmonella enterica subspecies enterica serotype Typhi (S. Typhi) was isolated from 11 (41%) of 27 patients; of 27 patients, only 4 (15%) had positive cultures from gut biopsies. S. Typhi DNA was detected by polymerase chain reaction for all perforation biopsy samples. Detailed histological examination of the gastrointestinal mucosa at the site of perforation in all cases showed a combination of discrete acute and chronic inflammation. Acute inflammation at the serosal surface indicated additional tissue damage after perforation. Immunohistochemical results showed that the predominant infiltrating cell types at the site of perforation were CD68+ leukocytes (macrophages) or CD3+ leukocytes (T lymphocytes).
Typhoid fever is a systemic infection caused by the bacterium Salmonella enterica subspecies enterica serotype Typhi (S. Typhi). The disease causes much morbidity and mortality in developing counties and is characterized by prolonged fever, bacterial growth in cells of the reticuloendothelial system, and significant inflammation of the lymphoid organs of the small intestine. There are an estimated 33 million cases of typhoid fever per year worldwide, most of which occur in the developing countries of Southeast Asia and Africa . Mortality rates associated with typhoid fever vary from region to region, with the highest (up to 12%–30%) reported from Indonesia, Nigeria, and India . The case-fatality rate is highest in patients >30 years or <1 year old [3, 4]. With early, appropriate, antimicrobial therapy, most cases resolve without complications. However, a minority of patients develop complications, the most serious of which is perforation of the gastrointestinal (GI) tract [5–7]. This usually occurs in the terminal ileum, near the ileocecal junction, in the center of an ulcer . This complication is more common in male and older patients . Although perforation can occur early in the febrile phase of typhoid fever, it is more commonly seen during or after the second week of illness . Patients with typhoid perforation usually require surgical intervention and vigorous antimicrobial therapy. In cases of perforation, laboratory confirmation of a clinical diagnosis of typhoid fever is difficult, because blood and bone marrow cultures often show no growth [10, 11].
The pathogenesis of typhoid perforation is poorly understood, including both bacterial and host factors, although it has been suggested that its pathology could be mediated by the immune system [6, 12]. In an attempt to elucidate the mechanisms behind typhoid perforation, we conducted a prospective study to describe the clinical, microbiological, and pathological features of patients with suspected typhoid perforation. The study was conducted in Dong Thap Provincial Hospital, in the Mekong Delta region of southern Vietnam. Typhoid fever is endemic locally, with an incidence of ∼193 cases per 100,000 population . Many patients are admitted to this hospital with suspected typhoid perforation, but the clinical diagnosis is often not confirmed microbiologically. The failure to grow S. Typhi may be due, in part, to treatment with antibiotics before blood samples are obtained for culture. Our findings provide information about the diagnosis of perforation in typhoid fever and offer insight into the pathogenesis of the condition.
Patients and Methods
Clinical study. The present study was approved by the scientific and ethical committees of The Hospital for Tropical Disease (Ho Chi Minh City) and the Dong Thap Provincial Hospital, Vietnam. Patients admitted to Dong Thap Provincial Hospital during the period from April 1997 through February 1998 who had a clinical diagnosis of GI perforation secondary to suspected typhoid fever were enrolled into the study. A standard clinical pro forma was completed for each patient at admission to the hospital, and specimens were obtained for microbiological, biochemical, and hematological analysis. All patients had detailed clinical descriptions of illness history, the findings of a full physical evaluation, and demographic information recorded at hospital presentation. Soon after admission to the hospital, blood samples were obtained, and a full blood count, the determination of urea and electrolyte levels, and liver function tests were performed. The Widal test was performed to detect serum antibody against O and H S. Typhi antigens, and samples of blood, bone marrow, urine, and stool were cultured (including 3 urine and 3 stool samples for each patient). GI biopsy samples from the site of perforation were obtained, cut into 4 equal parts along the length of the tissue with a disposable scalpel, and either stored directly in liquid nitrogen or formaldehyde or sent to the hospital's microbiology laboratory for culture. Patients were empirically treated for peritonitis with cefotaxime or ceftriaxone and gentamicin. Investigations were repeated later as clinically indicated. Informed consent was obtained from patients or from a member of their immediate family.
Blood and bone marrow culture. Venous blood samples (10 mL each) and bone marrow samples (1 mL each, obtained from the ileac crest) were cultured at dilutions of 1:5 and 1:50, respectively, in brain-heart infusion broth (Oxoid). Antibiotic susceptibilities to ampicillin, cotrimoxazole, chloramphenicol, ofloxacin, nalidixic acid, ciprofloxacin, cefotaxime, and gentamicin were determined by modified Kirby-Bauer disk diffusion technique. Stool samples (3 per patient) were cultured directly onto MacConkey agar, desoxycholate citrate agar (Difco), and xylose lysine deoxycholate (XLD; Oxoid) agar, and immunoglobulin was placed into selenite F broth (Oxoid) and subcultured onto XLD agar after overnight incubation at 37°C. A urine sample (10 mL) was obtained aseptically and centrifuged at 4000 rpm for 10 min, and the deposit was placed into 10 mL of selenite F enrichment broth and cultured as described above. Agglutination of specific antisera and a panel of biochemical tests were used (Murex) to identify salmonella isolates. Identity was confirmed by API20E test strips (BioMérieux).
Culture of lymph nodes, intestinal tissue, peritoneal fluid, and perforation tissue samples. Intestinal tissue was cut into small pieces with use of a sterile technique, cultured directly, and placed into 10 mL of selenite F broth. Peritoneal fluid samples were assessed for the presence of WBCs, centrifuged, and directly cultured onto selective media, and selenite F was poured over the centrifuged deposit. Selenite F was cultured on XLD after overnight incubation.
PCR for S.Typhi DNA. For intestinal specimens that contained a sufficient amount of tissue from the perforation site (25 of 27 specimens), a nested PCR with unique primer sequences for the S. Typhi flagellin gene  was used to detect S. Typhi DNA. Whole DNA extracts were prepared with use of a Qiamp tissue kit (Qiagen), according to the manufacturer's instructions, with a final sample volume of 30 µL. One microliter of this was used for the first round of PCR amplification. For the second reaction, 5 µL of the first reaction mixture was used. The primers for the initial reaction were 5′-TTAACGCAGTAAAGAGAG-3′ (ST1) and 5′-ACTGCTAAAACCACTACT-3′ (ST2). The primers for the second reaction were 5′-AGATGGTACTGGCGTTGCTC-3′ (ST3) and 5′-TGGAGACTTCGGTCGCGTAG-3′ (ST4). A standard reaction mix was used: Taq polymerase, dNTPs, and buffer (Appligene), with a reaction mixture 50 µL. For reaction 1 (primers ST1 and ST2), the cycle conditions were 1 cycle at 94°C for 5 min and 30 cycles at 94°C for 30 s, 58°C for 30 s, and 74°C for 30 s. For reaction 2 (primers ST3 and ST4), the cycle conditions were 1 cycle at 94°C for 5 min and 30 cycles at 94°C for 30 s, 68°C for 30 s, and 74°C for 30 s. PCR products were visualized on a 1% agarose gel that contained ethidium bromide (0.5 µg/mL).
Widal tube agglutination test. The Widal tube agglutination test (Sanofi Diagnostics) was performed according to the manufacturer's instructions. Widal serological testing was performed on serum samples from 19 of 27 patients, 12 of whom had tests performed on both acute and convalescent serum samples and 7 of whom had tests performed on acute serum samples only.
Histological and immunohistochemical testing. Intestinal tissue samples obtained during surgery for l perforation were fixed in 10% neutral buffered formalin, embedded in paraffin, and processed routinely. Microscopic examination of 3-µm hematoxylin-eosin—stained samples was performed. Immunohistochemical testing was performed with use of a peroxidase-labeled streptavidin-biotin conjugate. Primary antibodies were anti-CD3, anti-CD20, and anti-CD68/PGM1 (Dako). Matrix glycosaminoglycan (GAG) distribution was assessed by use of monoclonal antibody 10E4 (Seikagaku) that recognizes heparan sulfate chains and visualized with avidin-biotin (Vectastain Elite; Vector). Histochemical detection of sulphated GAGs was performed with use of a 1/100 concentration of cationic colloidal gold in PBS at pH 1.2 (polylysine gold; Biocell International) together with silver enhancer, as reported elsewhere [15, 16]. Specific enzymatic digestions have confirmed that this technique detects heparin sulfate in epithelial and endothelial cells and chondroitin and dermatan sulfates within the mucosal tissues [15, 17].
Clinical findings, gross histological findings, and routine laboratory test results. Twenty-seven patients (18 male and 9 female) with clinical typhoid perforation were enrolled into the study. Patient details, clinical findings, and laboratory results are shown in tables 1 and 2. An unexpected finding was that the median length of illness was 9 days (range, 1–28 days), with 13 of 27 patients presenting during the first week of illness. The following findings are possibly of use in diagnosis: the most common symptoms at presentation were abdominal pain with signs of peritonitis (in 100% of patients), fever (92%), headache (81%), and vomiting (66%); aspartate aminotransferase and alanine aminotransferase levels were elevated above reference ranges (males subjects, 7–46 IU/L; female subjects, 4–35 IU/L); total WBC counts were >11 × 109 cells/L in 18 (67%) of 27 patients; and neutrophil leukocytosis (i.e., >75% neutrophils in a peripheral blood count) was apparent in 22 (81%) of 27 patients.
At surgery, 20 of 27 patients were found to have a single perforation, and the remaining 7 patients had multiple small intestinal perforations (range, 2–5 perforations) (table 2). The size of the perforations was variable (median diameter, 5.0 mm; range, 1–25 mm). The median distance of the perforations from the ileocecal junction was 419 mm (range, 20–1800 mm). The GI tract was inflamed, and the mucosa around the perforation site was erythematous and swollen in all patients. Increased tissue fragility was also observed in 26 of 27 patients. Free fluid with fecal staining was found in the peritoneum in all cases.
Several surgical interventions were performed, including repair by simple closure (in which only suturing was necessary; in 2 of 27 patients); debridement, wedge excision, and suturing (in 13); drainage (in 6); and temporary ileostomy (in 4). One patient died after developing Escherichia coli septicemia secondary to multiple bowel perforations and peritonitis. Operative details were not available for 2 patients.
Microbiological findings. A clinical diagnosis of typhoid fever was confirmed by isolation of S. Typhi from ⩾1 specimen in 11 (41%) of 27 patients (table 3). Blood culture gave yields similar to those of bone marrow culture (15%), but S. Typhi was isolated from samples of both blood and bone marrow in only 2 of 27 patients. Culture of stool samples gave the highest diagnostic yield, being positive for S. Typhi in 5 of 27 patients; both blood and bone marrow cultures showed no growth in 4 of those 5 patients. Culture of peritoneal fluid samples yielded positive results for 3 of 27 patients; all 3 of these patients had blood and bone marrow cultures that showed no growth but had positive stool culture results, which confirmed fecal peritonitis. Of the 7 mesenteric lymph node homogenates cultured, S. Typhi was isolated from 3. Two of these were from patients with bone marrow cultures that were positive for S. Typhi but blood cultures that showed no growth. S. Typhi was isolated from the perforation sites of only 4 of 27 patients, 3 of whom had blood and bone marrow cultures that showed no growth. For 2 patients, S. Typhi was isolated from samples of peritoneal fluid, stool, and perforation site tissue. These results highlight the need for multiple cultures and demonstrate the value of cultures of stool, peritoneal fluid, and bowel tissue samples, all of which may yield S. Typhi in patients with bone marrow and blood cultures that show no growth. Isolates were tested for antibiotic sensitivity, and all were found to be resistant to ampicillin, cotrimoxazole, and chloramphenicol. Sixty percent of the isolates were also resistant to nalidixic acid, but all were sensitive to ofloxacin and ceftriaxone.
PCR of tissue homogenates. S. Typhi DNA was detected in all 25 (100%) of the intestinal tissue samples tested by nested PCR for amplification of the fliC sequence. This product has been shown to be specific for S. Typhi . The PCR product from the final round of a single reaction was sequenced. Over a 300 bp region, complete identity at the nucleotide level was found with the S. Typhi CT18 fliC sequence (Sanger). We therefore assumed that specific amplification occurred.
Serological test results. A single acute serum sample was obtained from 7 of 19 patients, and paired samples were obtained from 12 patients. A Widal titer of TO ⩾1:200 or TH ⩾1:100 is considered to be suggestive of typhoid fever in this region . Of these 19 patients, 10 had an elevated single TO titer (⩾1:100); only 2 had a titer of ⩾1:200. Three of 19 patients had a Widal TH titer of 1:100, and the remaining 16 had negative results. Only 1 patient demonstrated a 4-fold increase in titer, with TO antibodies increasing from 0 to 1:400. This patient had culture results that were negative for S. Typhi.
Histological and immunohistochemiical test results. Intestinal tissue specimens obtained from 25 of 27 patients consisted of 1 lymph node specimen, 5 full-thickness specimens, 14 partial-thickness samples, and 5 samples that consisted of inflamed tissue or serosa only. Evidence of severe acute inflammation, with fibrinopurulent exudates extending along the serosal surface, was observed by either macroscopic or microscopic examination at the site of perforation in all samples. A combination of acute and chronic inflammation was observed immediately adjacent to the site of perforation. The histological status further away from the perforation consisted of chronic inflammation with edematous granulation of tissue and infiltration of neutrophils, lymphocytes, plasma cells, and histiocytes (macrophages). In the majority of samples investigated, the inflammation was confined to the tissue adjacent to the perforation, with a gradual transition to normal tissue proximally or distally.
Evidence of villous blunting and elongation of the crypts was observed, which suggests chronic mucosal injury. Increased numbers of lymphocytes and plasma cells were present at the crypt bases. In the majority of specimens, these changes occurred only immediately adjacent to the perforation. Lymphoid follicles (i.e., Peyer patch) were observed in 1 specimen only.
In contrast to the mild-to-moderate inflammatory infiltrate in the lamina propria, the submucosa and muscularis showed signs of severe inflammation. A histiocytic infiltration with abundant lymphocytes was a prominent feature in many specimens. The histiocytes were frequently seen as confluent sheets, displaying abundant pink cytoplasm, and some of them contained phagocytic debris and indented nuclei. Occasional multinucleated giant cells were present, but well-formed granulomas were not observed. Microabsesses that consisted of small collections of neutrophils and apoptotic cells were observed within the histiocytic sheets, although bacteria were not visible. Large areas of lymphocytic infiltrate were present in a number of the specimens, usually deep within the submucosa. Gaseous necrosis was not observed, although hemorrhagic necrosis of the muscularis externa was observed in 1 specimen.
Immunoperoxidase staining of representative tissues demonstrated that the predominant cells in the areas of inflammation were CD68/PGM1+ macrophages. Most small lymphocytes within the areas of cellular infiltrate were CD3+ T cells, with infrequent CD20+ B cells.
For specimens with preserved external muscle layers, a pattern of relative normality of the external longitudinal muscle with vacuolation and disruption of the inner circular muscle layer was seen. In these cases, abnormalities were seen in the ganglia within the myenteric plexus, with a loss of normal structural integrity characterized by vacuolation and indistinct appearances of neural and glial cells. There was evidence of infiltration of the ganglia by inflammatory cells, including CD3+ T cells. The study of matrix composition showed a similar pattern of predominant abnormality within the deeper layers of the bowel wall. Thus, in those specimens with identifiable lamina propria and epithelium, mucosal GAGs were largely preserved, and the deeper layers of the same specimens showed evidence of GAG disruption, as has been reported elsewhere in cases of active intestinal inflammation . Staining for GAGs emphasized the focal abnormalities surrounding the myenteric plexus ganglia between circular and longitudinal muscle layers, with particular evidence of focal neovascularisation.
The majority of patients enrolled into the study were male, with an average age (23 years) similar to that in earlier reports . However, the average illness length of 9 days was less than expected and was similar to that of patients with noncomplicated typhoid fever at this hospital . It has been suggested elsewhere that it is during the second or third infection with S. Typhi that perforation occurs . We could not confirm or refute this. Only 3 patients gave a history consistent with a previous episode of typhoid fever, but some patients had been previously vaccinated against the disease.
The results of our microbiological study revealed that S. Typhi could be cultured from nearly one-half of the patients with suspected typhoid perforation if multiple samples were obtained. Serological testing was not useful. The presence of S. Typhi DNA was demonstrated in all patients according to the results of PCR; however, the lack of control specimens prevented the calculation of specificity. This technique deserves further evaluation.
The histological study of the perforation specimens was hampered by the extensive autolytic changes that had occurred in many of the samples, and the study was essentially limited to granulation tissue in some cases. Specimens were processed by standard protocols; therefore, the autolytic changes may themselves have been due to the pathological processes. Overall, the histological picture of typhoid perforation was found to be one of a chronic, but discrete, inflammation around the perforation site, with relatively mild-to-moderate mucosal changes. This was overshadowed by clear pathology within the deeper layers. In this regard, the changes seen were more reminiscent of diseases such as neonatal necrotizing enterocolitis than they were of classic inflammatory bowel disease [20, 21]. The findings of staining for sulphated GAGs confirmed the preferential involvement of the deeper layers and also demonstrated dense focal neovascularization. Thus, the question arises as to whether S. Typhi is able to trigger ischemic necrosis through effects on the deeper layers of the intestine. Both the vasculature and the neural supply of the gut are potential candidates for inducing bowel infarction in this situation. The loss of GAGs observed in the present study is consistent with the induction of matrix degrading metalloproteases by TNF-α . Indeed, TNF-α and the lipid mediator platelet-activating factor have been shown to synergize to induce fulminant bowel necrosis in rodents , and TNF-α has been associated with tissue necrosis in infections with Mycobacterium tuberculosis . Because bacteria were rarely seen in the histological sections and S. Typhi were cultured from only 4 perforation biopsy samples, it is possible that an exaggerated host response to a limited number of bacteria and/or bacterial products within the Peyer patch contributes to the development of perforation. This inappropriate or exaggerated host response may be due to immunological priming of the Peyer patches as a result of prior exposure to S. Typhi, much like that seen with bacterial endotoxin in the Shwarzman reaction . It is of interest that the Shwartzman reaction involves the clumping of reactive macrophages and lymphocytes around vascular tissues, resulting in intravascular thrombi and necrosis of venules—changes similar to those observed in the present study. These effects occur because bacterial products prepare tissue sites so that they become extremely sensitive to cytokine-mediated tissue damage on reexposure to a cytokine-triggering stimulus .
The pathogenesis of intestinal perforation in patients with typhoid fever is poorly understood with respect to the host and bacterial factors involved. It is generally believed that perforation occurs in the Peyer patches of the distal ileum. We were unable to confirm this in the present study, because histological evidence of the involvement of lymphoid tissue was present in only 1 specimen. However, the tissue samples that we used were obtained after the specific events of tissue damage had occurred; thus, some of the specimens consisted of dead tissue and inflamed mucosa, and the underlying architecture could not be ascertained. Our data confirm the inflammatory nature of typhoid perforation, showing that the inflammatory infiltrate consists predominantly of macrophages and T lymphocytes and that it is most severe in the deeper tissues. The loss of GAGs and the vascular changes are indicative of TNF-α—mediated pathology.
We thank the technical and medial staff of the microbiology laboratories, the clinical staff, and the directors of The Centre for Tropical Diseases and The Dong Thap Provincial Hospital, Dong Thap, Vietnam.
Financial support. The Wellcome Trust of Great Britain.
- polymerase chain reaction
- enteric fever
- intestinal perforation
- salmonella enterica
- inflammation, acute
- inflammation, chronic
- cd68 antigen, human
- gastrointestinal perforation
- gastrointestinal mucosa
- illness length
- tissue damage
- intestinal biopsy