Colon Tuberculosis: Endoscopic Features and Prospective Endoscopic Follow-Up After Anti-Tuberculosis Treatment

INTRODUCTION

Mycobacterium tuberculosis (TB) infection is still common in many countries. According to a World Health Organization report, global annual incidence of TB is estimated to be 9.4 million cases, of which 1.98 million cases are from India,¹ and close to 500,000 per year will die of the disease in India.² There has also been a resurgence of TB in many developed countries secondary to migrant population, deteriorating social conditions, cutbacks in public health services and increasing prevalence of immuno-suppressed individuals.^{3, 4, 5, 6, 7, 8, 9} In one study, coinfection of TB and HIV was reported to be as high as 73%.⁶ TB is also a major concern in those receiving biological agents for conditions like rheumatoid arthritis or inflammatory bowel disease.^{3, 10} Immuno-suppressants given on the mistaken diagnosis of inflammatory bowel disease can result in systemic spread of TB with fatal consequences.

Although TB of the gastrointestinal tract is not as common as pulmonary TB, terminal ileum is the most frequently affected organ. This is attributed to the high prevalence of lymphoid tissue in the terminal ileum and a longer contact time of luminal contents with the small bowel. Ulcerations and luminal narrowing are the predominant lesions found in this region of the small bowel. Besides the terminal ileum, several small case series have described TB affecting the colon.^{11, 12, 13, 14, 15, 16, 17, 18, 19, 20, 21} TB of the colon is rare and hence it can be misdiagnosed as colon cancer, inflammatory bowel diseases, ischemic colitis or infectious colitis (also endemic in areas with high prevalence of TB).

In view of the anatomic (lymphoid tissue) and physiological (contact time with luminal contents) differences between the small bowel and the colon, the resolution pattern of lesions in the colon after anti-TB treatment is not well known as routine colonoscopy is not performed after completion of treatment. The aim of the present study was to evaluate the endoscopic features of TB of the colon and prospectively evaluate the healing of TB lesions after anti-TB treatment.

METHODS

We conducted a prospective cohort study enrolling consecutive patients diagnosed with colon TB from October 2004 to December 2010 at one tertiary care center in India. Informed consent was obtained from all the patients. The study was approved by the local Institutional Review Board.

Patients ≥18 years of age who were suspected or confirmed to have colon TB based on clinical evaluation and endoscopic biopsies were enrolled. Those with ulcerative colitis, Crohn's disease, microscopic colitis, or indeterminate colitis were excluded. Patients presenting with symptoms suggestive of colitis such as bloody diarrhea, abdominal pain, and fever routinely underwent stool studies, and where indicated additional tests were performed to rule out infectious etiology like salmonellosis, shigellosis, and amebiasis.

Demographic information, clinical history and physical examination findings were noted. Information collected included previous history of TB, history of TB in close contacts, history of HIV, the presence or absence of abdominal pain, fever, bleeding per rectum, weight loss, loss of appetite and altered bowel movements. Clinical examination findings including vitals, pallor, abdominal tenderness, abdominal mass and lymphadenopathy were recorded. Results of routine laboratory tests were also noted. Medication list was reviewed for immunosuppressive agents. Additional investigations including chest X-ray, abdominal ultrasound and abdominal computerized tomography scan were done when clinically indicated.

RESULTS

DISCUSSION

M. tuberculosis infection is endemic in many countries and there has been a resurgence of TB in the Western world.⁹ The annual global incidence of TB is estimated to be 9.4 million cases.¹ A report from a panel of TB experts and epidemiologists from more than 40 countries showed that an estimated 1.87 million (1.4 million–2.8 million) people died of TB in 1999.³⁰ The global case fatality rate was 23% and exceeded 50% in some African countries with high HIV prevalence rates.³⁰ The resurgence of TB in many developed countries has been attributed to higher migrating population, deteriorating social conditions, cutbacks in public health services and increasing prevalence of immuno suppressed/HIV-positive individuals.^{3, 4, 5, 6, 7, 8, 9} In the United States, this escalation peaked in 1992 and since then there has been a steady decline, with TB now being diagnosed mostly in the migrant population.^{8, 9}

Gastrointestinal TB, although not as commonly seen as pulmonary TB, can be a source of significant morbidity and mortality as it can mimic other conditions and tends to be diagnosed late. Less than 25% of patients with gastrointestinal TB have concomitant pulmonary TB.⁹ Hence diagnosing gastrointestinal TB requires a high index of suspicion, especially in countries where TB is not endemic. Some of these patients are misdiagnosed as having inflammatory bowel disease and may get immunosuppressant medications resulting in dissemination of TB.

The intestinal tract is usually affected by swallowed bacterium, spread from adjacent organs or via hematogenous spread. Secondary to abundant lymphoid tissue and the relatively longer contact time of intestinal contents with the mucosa, the terminal ileum has a higher predilection for TB infection.^{9, 11, 14} Hence, in an appropriate clinical setting (for example, an immunosuppressed patient or immigrant population), TB is always in the differential diagnosis if there is an ulcer, stricture or an inflammatory mass in the terminal ileum. However, diagnosing TB of the colon requires a high index of suspicion as colon is not the most commonly affected site and symptoms are non-specific. In a study by Nagi et al.,¹⁷ only 10.8% of the 684 cases with abdominal TB had TB affecting the colon. Case reports and small retrospective case series have been published describing the endoscopic features of TB of the colon.^{11, 12, 13, 14, 15, 16, 17, 18, 20, 21, 31, 32, 33, 34, 35, 36, 37, 38} These features can be non-specific and mimic inflammatory bowel disease or malignancy. As follow-up colonoscopy is not routinely performed after completion of anti-TB treatment, there are limited data on predicting the outcomes of TB lesions, namely, resolution of, lack thereof, or scarring or stricturing of the lumen after healing.

In this study, we report on one of the largest cohorts of patients with colon TB who were followed prospectively with colonoscopy examination after completing anti-TB treatment. Similar to previously reported case series,^{11, 21, 35, 39} majority of the patients had non-specific symptoms, with weight loss and abdominal pain being the predominant symptoms in over 80% of the patients. Less than half of the patients in this study had fever. Abdominal mass on physical examination was noted in only 6.4% of the patients. The number is significantly less compared with the study by Palmer et al.⁴⁰ in which 43% patients developed abdominal mass comprising either inflammatory intestinal lesions or enlarged mesenteric lymph nodes. The difference is likely due to the fact that their series had predominantly small bowel involvement (40 of 42) compared with only 4.7% showing isolated TB affecting the colon. Bleeding per rectum was observed in 20% of patients in the present series, but none of these was severe enough to require surgical or angiographic intervention. In the series by Misra et al.³⁵ 10% of patients required surgery for massive lower gastrointestinal bleeding. Similar to previously published series,^{11, 21, 35} in our study, the predominant lesions identified on colonoscopy were ulceration and nodularity. The right colon (ileocecal valve, cecum, and ascending colon) was significantly more affected compared with the left. Nagi et al.¹⁷ on the other hand found the transverse colon to be the most commonly affected site and strictures the predominant lesion. However, their study was based on radiological evaluation and not colonoscopy.

On the basis of the above, the symptoms and the endoscopic findings of colon TB can be non-specific and hence diagnosing TB of the colon can be difficult. We had six patients with previous pulmonary TB as seen on chest X-ray and one patient with ascites had positive adenosine deaminase test. Histology was supportive of TB in 73% of the patients. Three patients showed positive acid-fast bacillus on histology; granulomas were the predominant finding in other patients, but as reported previously,³⁵ caseating granulomas were rarely seen. In the current series, caseating granulomas were present in two patients and in all the five resected specimens of those who underwent surgery. Majority of the patients presented with suspected TB and diagnosis was confirmed by response to anti-TB treatment. Response to anti-TB treatment as a criterion to confirm TB has also been used in other studies.²³ Although it will be difficult to justify empiric anti-TB treatment for patients in developed countries, this has been an acceptable approach in countries where TB is endemic. Park et al.⁴¹ compared the outcomes of anti-TB treatment in patients with non-specific ileocecal ulcers. They showed that 3 months of trial of anti-TB treatment can be sufficient to assess response to treatment and thus differentiate between colon TB and Crohn's disease. Our study results are consistent with this, as majority of the patients had significant clinical improvement within 4–6 weeks after initiation of anti-TB treatment. Skin Purified Protein Derivative (PPD) and QuantiFeron tests can be used as supportive evidence in regions non-endemic for TB, but their utility in endemic areas is questionable. Hence Purified Protein Derivative (PPD) test is not recommended by the Revised National Tuberculosis Control Program guidelines from India for diagnosis of TB.²⁴ QuantiFeron test is relatively contraindicated in the evaluation of active TB, as active TB can be associated with suppressed interferon-γ responses.^{25, 26}M. tuberculosis culture from intestinal biopsy specimen is expensive and previous studies have shown variable yield.^{12, 27, 35, 42} In a study by Singh et al.,²⁷ none of the 62 patients showed positive cultures and only 3 out of 50 intestinal biopsy cultures in a study by Shah et al²⁰ showed M. tuberculosis. Hence, we elected not to culture the biopsy specimens.

Two patients who did not respond to anti-TB treatment were eventually diagnosed to have inflammatory bowel disease. None of the patients received any empiric immunosuppressant medications before anti-TB treatment on the assumption that they have inflammatory bowel disease, for fear of dissemination in case they had TB. Differentiating TB ulcer from Crohn's disease ulcer has always been challenging.^{19, 34, 43, 44, 45, 46, 47, 48, 49} In this series, TB ulcers were linear, fissured, transverse, or circumferential. Similar to as reported by others,^{11, 20, 50} we also observed that the mucosa surrounding TB ulcers was abnormal, showing features like erythema, edema, mucosal irregularity or nodularity. In Crohn's disease, the ulcers are usually surrounded by normal-looking mucosa. Moreover, aphthous ulcers as seen with Crohn's disease were not observed, except in the patient who subsequently was diagnosed to have Crohn's disease. Crohn's patients can present with significant gastrointestinal bleeding.⁵¹ None of the patients in the current series had any significant colonic bleeding. However, this feature cannot be used as suggestive of TB as in a previous case series 10% of patients with colon TB required interventions for massive gastrointestinal bleeding.³⁵ Multiple studies have attempted to identify features to differentiate between TB and Crohn's disease. A study by Makharia et al.²³ attempted to calculate a scoring system based on the following features—weight loss, absence of sigmoid colon involvement, absence of blood in stool and absence of focal colitis to differentiate between Crohn's and TB. Another study by Lee et al.⁴⁹ evaluated various other endoscopic features to differentiate between Crohn's disease and colon TB. Four parameters—anorectal lesions, longitudinal ulcers aphthous ulcers and cobblestone appearance—were significantly more common in patients with Crohn's disease than in those with intestinal TB. Four other parameters—involvement of fewer than four segments, a patulous ileocecal valve, transverse ulcers, and scars or pseudopolyps—were more frequently observed in patients with intestinal TB than in those with Crohn's disease. However, there have not been any large population-based studies to validate these findings. A study by Gan et al.⁵² showed that PCR can be useful in distinguishing between Crohn's and TB. Pulimood et al.⁵³ described histological features—type and frequency of granulomas, the presence or absence of ulcers lined by epithelioid histiocytes and microgranulomas and the distribution of chronic inflammation being useful to distinguish between the two diseases.

Luminal narrowing was noted in 30 patients in the current series. Four patients (13.3%) with luminal narrowing on index colonoscopy and one patient with an obstructing polypoidal mass mimicking colon cancer required surgery. We were unable to assess the efficacy of endoscopic dilatation in these patients as they were taken to surgery before endoscopic intervention could be applied. In a previous series, endoscopic dilatation of colon strictures was performed and was shown to be effective and safe.^{35, 54} In the remaining patients, the narrowing resolved with anti-TB treatment, thereby suggesting that these strictures were inflammatory in nature. Majority of the other types of lesions (ulcers, nodularity, and polyps) also resolved with anti-TB treatment, as confirmed on follow-up colonoscopy.

The findings of our study have important implications. As majority of the patients had resolution of the colon lesions with anti-TB medications, follow-up colonoscopy may not be required in patients with colon TB if they are asymptomatic after treatment. Majority of the luminal narrowing as noted on index colonoscopy also resolve with anti-TB treatment and hence these strictures appear to be inflammatory rather than fibrotic in nature.

This study had several limitations. First, few tests were performed owing to patients concerns regarding the cost involved and hence small bowel enterography and computerized tomography/US of the abdomen were not performed in all the patients. Second, we did not perform culture for TB as previous studies have shown variable yield.^{27, 35, 41, 42, 55} We could have performed Mycobacterium culture in the two patients who did not respond to conventional anti-TB treatment to evaluate for drug-resistant TB. However, follow-up colonoscopy features were suggestive of Crohn's disease and these patients responded to prednisone. Although the prevalence of Crohn's disease is low in India, there is always a chance of falsely labeling a Crohn's disease patient as having colon TB with resolution of lesions secondary to the anti-inflammatory effect of anti-TB medications. However, a study by Park et al.⁴¹ demonstrated that patients with suspected inflammatory bowel disease treated with anti-TB medications did not have any significant lasting improvement in symptoms or resolution of lesions seen on colonoscopy. Eventually, while on anti-TB medications these patients may present with ongoing active disease, as was seen in the two patients with Crohn's disease in our study. The findings of our study are not applicable to HIV-positive or immuno-compromised patients as none of the patients in our study were HIV-positive or immuno-suppressed.

In summary, patients with TB of the colon present with non-specific symptoms and hence diagnosis in countries where TB is non-endemic involves a high index of suspicion (migrant population, immuno-suppressed individuals). Abdominal pain and weight loss are the predominant symptoms. Ulceration, nodularity, and luminal narrowing are the prominent endoscopic findings mostly affecting the right colon. Histology is helpful in ruling out other conditions, but TB-specific findings like caseasting granuloma and acid-fast bacilli are rarely seen. Tuberculous ulcers are morphologically different form Crohn's ulcers in that they are predominantly linear, fissured, transverse, or circumferential and the mucosa surrounding these ulcers is abnormal. Majority of the colon lesions resolve with anti-TB treatment, including strictures, thereby suggesting that strictures are more inflammatory rather than fibrotic in nature. Follow-up colonoscopy is not required in those who have symptomatic improvement after anti-TB treatment.

Study Highlights