INTRODUCTION
Tuberculosis (TB) continues to represent a major health concern, with an incidence rate in 2018 in France of 8.9 (7.8–10)/100,000 persons. 1 In 2016, an increase of 15% in the incidence of TB was recorded in the Paris region, and this trend was confirmed the following year. 2 Extrapulmonary TB represented more than 28% of the declared cases, 2 half of them without any respiratory involvement. Abdominal TB (ATB) is reported as the sixth most common form of extrapulmonary TB 3–5 and is defined as the involvement of the gastrointestinal tract, peritoneum, lymph nodes, and/or solid organs (liver, spleen, and pancreas). Recent data in France show that extra-thoracic lymph nodes are the most frequently involved site among the extra-respiratory localizations (54.8%), whereas peritoneal, hepatic, and intestinal forms represented around 13% of cases. 6 Because of its nonspecific clinical presentation, ATB is usually diagnosed late and can be a source of significant morbidity and mortality, thus requiring a high index of clinical suspicion. Recent data regarding the epidemiology, clinical presentation, and outcome of ATB in developed countries are relatively scarce. This study describes the characteristics, evolution, and treatment of patients diagnosed with ATB in two tertiary-care hospitals in the Paris area and represents, to our knowledge, the largest French cohort of ATB described to date.
MATERIALS AND METHODS
We retrospectively reviewed the medical records of all patients diagnosed with ATB presenting over a 7-year period (January 2010–December 2016), in the infectious disease departments of two tertiary-care university hospitals in the Paris region (inner-city and suburbs), with a total bed-capacity of more than 2,000 places. Medical records were selected via Programme de Médicalisation des Systèmes d’Information (medicalized information system program), each medical file was verified individually, and data were collected based on the available clinical records. The research protocol was reviewed and approved by the Ethics Committee of the French Infectious Diseases Society (Comité d’Ethique de Recherche en Maladies Infectieuses et Tropicales) under the Institutional Review Board No IRB00011642. According to the French law (No 78-17 of January 6, 1978 on computers, files, and liberties), this study has been registered with the Commission Nationale de l’Informatique et des Libertés (CNIL) (French National Agency regulating Data Protection, CNIL 2085894) and was conducted in compliance with the reference methodology 004.
Abdominal TB was defined as a clinical presentation compatible with TB along with positive bacteriological examination and/or a histological sample of abdominal organ tissue compatible with TB (epithelioid gigantocellular granuloma—GCG—with or without necrosis) and/or imaging in favor of gastrointestinal (GI) tract, peritoneal, intraperitoneal solid organ, and/or lymph node TB. Suggestive radiological features were considered wall thickening or a mass in any part of the GI tract (mostly ileocecal); peritoneal or omental thickening; peritoneal/mesenteric nodules; liver, spleen or pancreas enlargement; liver and spleen nodules; and lymph nodes with central hypo-attenuation. Hence, two different diagnostic criteria (clinical presentation and signs and at least one among bacteriological, histological, or radiological criteria) and TB treatment initiation were always necessary to include a patient. Only patients for whom access to full medical records was possible were included. The genitourinary localization was not retained as part of the ATB case definition. We collected demographic data, clinical features, diagnostic procedures, medical and surgical management, safety, and disease outcome. Cure was defined as clinical improvement and satisfactory imaging 2 years after treatment initiation. Data regarding the incidence of paradoxical reactions (PR) and corticosteroid use were also recorded. Paradoxical reaction was defined as clinical or radiological worsening of preexisting tuberculous lesions or the development of new lesions, in patients receiving antituberculous medication who initially improved on treatment with a good adherence.
Mycobacterium tuberculosis PCR was performed using commercially available tests (FluoroType®MTB, Hain Lifescience GmbH, Nehren, Germany, and Xpert MTB/RIF, Cepheid, Maurens-Scopont, France) and culture obtained on the solid Löwenstein–Jensen medium or mycobacteria growth indicator tube or BACT-Alert (bioMérieux France, Craponne, France). Quantitative variables were expressed as median/IQR and qualitative variables as percentages.
RESULTS
Patients’ characteristics and clinical presentation.
We included 80 patients who met the ATB case definition. Patients’ baseline characteristics are summarized in Table 1. The median (IQR) age was 39 (29–50) years, and 56.2% were male. Among the patients, 63.7% were born in African, mostly sub-Saharan countries, 15% and 11.2% were, respectively, of Asian and European descent (7/80 patients were Metropolitan France natives). For patients born outside France, the median (IQR) time between ATB diagnosis and arrival in France was 6 years (1–20). Six patients in our cohort were homeless.
Patients baseline characteristics
| Characteristic | n = 80 |
|---|---|
| Age (years), median (IQR) | 39 (29–50) |
| Male gender, n (%) | 45 (56.2) |
| Country/region of origin, n (%) | |
| Sub-Saharan Africa | 44 (55) |
| Maghreb | 6 (7.5) |
| Asia | 12 (15) |
| Europe | 9 (11.2) |
| Other | 9 (11.2) |
| Time interval between arrival in France/TB diagnostic (years), median (IQR) | 6 (1–20) |
| History of previous TB treatment, n (%) | 10 (12.5) |
| HIV coinfection, n (%) | 21 (26.2) |
| Patients with plasma viral load < 50 cp/Ml | 2 |
| TB localization | |
| Extra-abdominal localization present, n (%) | 65 (81.2) |
| Intra-abdominal lymph nodes, n (%) | 58 (72.5) |
| Pulmonary, n (%), of whom AFB+ in sputum | 50 (62.5), 31 |
| Peritoneal, n (%) | 50 (62.5) |
| Gastrointestinal tract, n (%) | 16 (20) |
| Solid organ (liver, spleen, and pancreas), n (%) | 20 (25) |
| Pleural, n (%) | 6 (7.5) |
| Bone, n (%) | 3 (3.7) |
| Neuromeningeal, n (%) | 3 (3.7) |
| Ascitis, n (%) | 43 (53.7) |
| Intra-abdominal abcess, n (%) | 15 (18.7) |
| Treatment duration (months), median (IQR) | 9 (9–12) |
| Patients with treatment adverse events grade 3–4 requiring the discontinuation of one or more molecules, n (%) | 4 (11.7) |
| Patients with a paradoxal reaction, n (%) | 12 (15) |
| Patients with a paradoxal reaction requiring corticosteroid treatment | 6 (7.5) |
| Bacteriological results, n (%) | N = 57 |
| Sensitive M. tuberculosis | 52 (65) |
| MDR/XDR M. tuberculosis | 3 (3.7) |
| Mycobacterium africanum | 2 (2.5) |
AFB = acid-fast bacilli; M. tuberculosis = Mycobacterium tuberculosis.
Twenty-eight patients had at least one cause of immunosuppression, including 21 patients (26.2%) HIV-infected patients with a median (IQR) CD4 count of 77.5 cells/mm3 (77–166), mostly with a detectable HIV plasma viral load (19/21 patients, of whom seven with a new HIV diagnosis) at TB diagnosis. Other causes of immunosuppression were biological treatment for systemic inflammatory disease (n = 3), hematological diseases (n = 2), cancer (n = 1), and heart transplant (n = 1). Five patients had diabetes mellitus and one was pregnant.
Ten patients of 80 (12.5%) had a previous history of TB treatment, whereas two patients had in their past records a positive Quantiferon test without prophylactic treatment. Regarding previous TB treatment, this was for pulmonary TB, and none of the patients had been treated for ATB. Exact delay for previous TB treatment was not available, but none of the patients had a treatment during the previous year.
Presenting symptoms were mostly nonspecific. Weight loss was a common finding in 85.2% of patients having lost a median (IQR) of 15% (12–20) of their normal weight. Ascitis was present in 43 (53.7%) patients. About half of patients had abdominal symptoms (pain and altered bowel transit), and a similar amount complained of fever and/or night sweats. Respiratory symptoms were recorded in one-third of patients.
An extra-abdominal localization was present in the majority of patients (n = 65, 81%), with CT scan or X-ray showing in more than half of patients pulmonary lesions (n = 50, 63%), affecting both lungs in 33/50 patients and compatible with a hematogenous miliary distribution in 11/50 patients. Excavating lesions were found in 13/50 patients. Concomitant genitourinary involvement was present in 13 patients (16%). Few patients presented with pleural (n = 6, 8%), bone (n = 3, 4%), or meningeal involvement (n = 3, 4%).
Diagnostic methods and investigations.
Tuberculosis diagnosis was established based on a set of clinical, bacteriological, histological, and/or radiological findings. Tuberculin skin test was available in 16 patients (20%) and positive in 14 patients, with a phlyctenular reaction in four patients.
Abdominal imaging was available in all patients, and it was always abnormal (Figures 1–3). Findings were compatible with an involvement of intra-abdominal mesenteric lymph nodes (n = 58, 73%), peritoneum (n = 50, 62.5%), GI tract (n = 16, 20%), and solid organs (n = 20, 25%). Peritoneal involvement was suggested by the presence of ascitis (n = 43, 53.7%), omental/peritoneal or mesenteric thickening, and/or omental/peritoneal nodules. An intra-abdominal abcess was found in 15 patients (19%).
Example of liver and spleen involvement in a 35-year-old male patient. Both the spleen and liver have an increased size (23 cm and 13 cm, respectively). There are multiple hypo-attenuated nodules in the hepatic and splenic parenchyma.
Citation: The American Journal of Tropical Medicine and Hygiene 104, 1; 10.4269/ajtmh.20-0023
Example of omental nodular thickening in a 30-year-old male patient. There are multiple nodules on the greater omentum with a pattern of omental cake.
Citation: The American Journal of Tropical Medicine and Hygiene 104, 1; 10.4269/ajtmh.20-0023
Example of enlarged nodes in a 31-year-old male patient. There are multiple enlarged nodes in the coeliac and para-aortic areas. Some of the nodes show a central hypo-attenuation. The spleen is also enlarged.
Citation: The American Journal of Tropical Medicine and Hygiene 104, 1; 10.4269/ajtmh.20-0023
A histological sample was obtained in more than half of patients (n = 42, 52%). An abdominal tissue sample (peritoneal, gastric, epiploic, or lymph node biopsy) was obtained in 26 patients (32%). Ascitis punction was performed in 18 patients, with a positive acid-fast bacilli (AFB) in one patient and a positive culture in 10 patients.
When a pulmonary form of TB was concomitant with the abdominal involvement, TB diagnosis was established by a positive AFB sputum staining in 31 patients (39%) (Figure 4). In the absence of respiratory involvement or in the remaining patients with an initial AFB negative sputum, diagnosis was confirmed by culture (N = 26, 12 sputum), histology presenting GCG with or without necrosis (N = 30), or culture and histology (n = 14). PCR testing was available in 30 patients and positive in 19, of whom 7/19 patients had negative AFB staining (ascitic fluid, lymph node, or peritoneal biopsy), and all 19 patients had a positive culture. In 17 patients with a compatible histology, a bacteriological TB confirmation was never obtained (negative AFB staining and culture). Of note, among the 42 patients with a histological sample examined, GCG were present in all, except two patients. Diagnosis was made solely on clinical (weight loss, fever and/or night sweats, and abdominal pain) and compatible radiological examination in eight patients, of whom four also had a phlyctenular tuberculin skin test.
Abdominal TB diagnosis. *Histology samples obtained (n = 22). **Histology samples obtained (n = 20). ***Type of histology sample obtained: peritoneal (n = 9), lymphadenopathy (n = 5), caecum (n = 2), and omentum (n = 1). AFB = acid-fast bacilli; TB = tuberculosis.
Citation: The American Journal of Tropical Medicine and Hygiene 104, 1; 10.4269/ajtmh.20-0023
Management and treatment outcome.
All patients received standard four-drug therapy (except 3/80 patients with multidrug resistant/extensively drug resistant [MDR/XDR] TB), for a median (IQR) duration of nine (9–12) months. The outcome was favorable, leading to cure, defined as clinical improvement and satisfactory imaging in 68 (83.7%) patients. Twelve patients were lost to follow-up. Adverse events related to treatment and requiring a temporary discontinuation were recorded in 14 patients (liver toxicity [n = 9], ophthalmological toxicity [n = 2], fluoroquinolone-related tendinopathy [n = 1], rifampin-related thrombocytopenia [n = 1], and cardiac toxicity [n = 1]), but with all these patients reaching cure after treatment adjustment. A PR occurred in 12 patients (15%). Among them, six were uncontrolled HIV infection and developed TB-associated immune-reconstitution inflammatory syndrome (TB-IRIS) after antiretroviral treatment initiation, with an appearance of same signs or aggravation in the initial symptomatology (abdominal pain, fever, and lymph node enlargement) or new lesions. Overall, corticosteroid therapy was used in 15 cases (four HIV-infected patients), either after the onset of a PR/IRIS (n = 6, of whom three HIV) or to prevent one because of the risk for complications (neurological TB [n = 3], pericarditis [n = 4], peritoneal TB [n = 1], and hemophagocytic syndrome in an HIV patient [n = 1]), and the recorded final outcome was favorable in at least 12/15 patients (n = 3, lost to follow-up). Eight patients underwent surgery for either obstruction or perforation, and they all had peritoneal involvement.
DISCUSSION
France is considered as a low-endemic country, with an estimated TB burden in 2018 of 5,800 (5.1–6.5) cases (eight, nine cases per 105 inhabitants) in 2018. 1 Paris region has an incidence roughly double in size compared with the national average. After reaching in 2015 the lowest number of declared active TB cases since 2002, starting with 2016, the curve shifted, and TB incidence climbed again. An increase of 15% was observed in the Paris region, and the number of TB cases rose from 323 in 2015 to 372 in 2016. This trend was confirmed in 2017 (403 cases). 2 This change of pattern is assumed to be at least in part related to an increase in the number of TB cases among people from sub-Saharan Africa, recently arrived in France, some living in precarious conditions. 2
In line with these data and consistent with other studies, 7 in our cohort, most of the patients were either of African descent or born in other endemic regions. More than 50% of patients were in France for more than 6 years, a finding in concordance with other reports 7 and French epidemiology data showing that 33% of active TB cases occur within 2 years and half to the 5 years after arrival in France. 2 The relatively young age and the nearly equal distribution between genders both reflect the profile of this mostly migrant, mobile population, in line with previous data. 7,8 Extreme precarity was observed in only six patients who were homeless, but this variable might have been underestimated as detailed information about the living conditions of each patient was not collected.
Extrapulmonary TB in general, and ATB in particular, is more frequent in immunocompromized patients, 9,10 but several reports show a high incidence of ATB in patients without an immunocompromized health status, 7,11 emphasizing the need for a low threshold of suspicion in appropriate clinical circumstances. 7 In our cohort, 35% of patients had a clear factor of immunosuppression, with 26% HIV-infected patients, mostly with a detectable HIV plasma viral load and/or newly diagnosed HIV infection. Uncontrolled HIV infection, a known risk factor for extrapulmonary TB, should elicit a systematic assessment of extra-respiratory sites. Compared with other cohorts, 7 we report a relatively high number of HIV-infected patients, possibly because of the profile of the two sites, also HIV reference centers.
Presenting symptoms were nonspecific, but well described with TB (weight loss, night sweats, and fatigue) and dependent of other concomitant localizations. More than half of the patients had some abdominal symptoms (abdominal pain or altered bowel habit). Even discreet but persisting abdominal symptoms in a context of suspected TB should lead to investigating for an abdominal localization.
Consistent with previous reports, 11 most of our patients had an extra-abdominal site involved. Pulmonary TB is generally reported in 15–50% of patients with concomitant ATB, 8,12 but it was present in almost two-thirds of our patients (63%). We also reported a high incidence of genitourinary TB (16%), next most frequent localization after the lungs. This emphasizes the need to investigate this specific localization, as it may also provide a more accessible site to establish the diagnosis. Mesenteric lymph-node involvement (73%) overpassed the peritoneum (62.5%) described by several reports 7,11–13 as the most frequently affected abdominal site, reflecting potentially an improvement in the precision of the diagnostic methods used.
Indeed, histology and radiological investigations were key in establishing the diagnosis in this cohort. All patients had an abnormal abdominal imaging, and half of the patients had some histological sample (abdominal or other) consistent with TB. The majority of patients (90%) had an objective confirmation of TB, and cultures were positive in about 70% of patients, which is one of highest rates reported in the literature, 8,12 reflecting also the high prevalence of pulmonary TB in our cohort. As reported by others, the importance of histology seems paramount to support the diagnosis. 7,11,14,15
The use of PCR was relatively low in our patients (37%), but allowed a more rapid diagnosis in some patients with a negative AFB staining, which all finally had positive cultures (N = 19). Before the release of the standard operating procedure in 2013 for the GeneXpert MTB/RIF assay 16 by the WHO, various PCR techniques were used for processing extrapulmonary samples. A recent evaluation of the use of GeneXpert on extra-respiratory and respiratory samples other than sputum in a low-incidence setting 17 concluded that although this technique could provide rapid diagnostic results, the overall sensitivity to rule out the disease may be suboptimal for some specimen types. Performance varied according to specimen type and AFB smear status. Acid-fast bacilli–negative samples had a sensitivity of 77% and specificity of 96%, detecting in this report 27% more cases than microscopy. 17 Results from lymph node tissues should be interpreted with caution as sensitivity and specificity were of 63% and 33%, respectively. 17 Independent of the PCR result, it is highly important for all samples to be cultured.
Despite a recent randomized-controlled trail showing that 6-month therapy was as effective as 9-month therapy in patients with intestinal TB, 18 treatment duration in our cohort was often longer than 6 months, similar to other reports, 7,12,18 and in more than 50% of cases longer than 9 months. Evolving practices may lead to a shorter treatment duration of ATB, as foremost of extrapulmonary localizations.
Corticosteroids were used in our study in 19% of patients, either for a PR or to prevent complications. The use of steroids in the context of ATB is poorly reported in the literature. A recent systematic review and meta-analysis on the use of adjunctive steroids in tuberculous peritonitis 19 has shown that this strategy was more effective than antituberculous treatment alone for the prevention of symptomatic stricture (RR: 0.15 [0.04–0.62] P = 0.008) and intestinal obstruction (RR: 0.18 [0.03–0.99] P = 0.05). However, the poor quality of the available studies limits the generalizability of these findings addressing tuberculous peritonitis alone.
The need for surgery was very limited in our patients, contrary to previous data, 12 but we reported only cases that needed laparotomy to address complications, as most of the histology samples were obtained by less invasive methods.
Our study has several limitations. It is retrospective, involving only two centers, both from the Paris region, and purely descriptive, data not allowing correlation analysis of variables. Despite these limitations, this report represents to our knowledge the largest French cohort of ATB described to date.
In conclusion, our cohort of patients with classical risk factors for TB, both demographical (immigration) and medical (immunosuppression), the main abdominal localizations, were lymphatic and peritoneal, and extra-abdominal involvement was present in more than three quarters of cases. In the context of a recent increase of TB cases in the Paris region, a high index of suspicion needs to be maintained for abdominal involvement. Patients of this cohort had in their majority bacteriologically proven TB. Treatment duration was variable, but usually longer than 6 months, with a favorable outcome. The use of steroids remains controversial and should be considered as it seems to allow a reduction in some complications, but it does not seem systematically needed.
REFERENCES
- 1.↑
WHO , 2019. Global Tuberculosis Report 2019. Geneva, Switzerland: WHO. Available at: http://www.who.int/tb/publications/global_report/en/. Accessed December 19, 2019.
- 2.↑
Antoun F , 2018. La tuberculose à Paris en 2016: un tournant?/tuberculosis in Paris region in 2016: a turning point? BEH Bull Épidémiol Hebd 8: 113–120.
- 3.↑
Mehta JB , Dutt A , Harvill L , Mathews KM , 1991. Epidemiology of extrapulmonary tuberculosis. A comparative analysis with pre-AIDS era. Chest 99: 1134–1138.
- 4.
Vaid U , Kane GC , 2017. Tuberculous peritonitis. Microbiol Spectr 5: 1–6, doi: 10.1128/microbiolspec.TNMI7-0006-2016.
- 6.↑
Santé Publique France , 2018. Données Epidémiologiques/Tuberculose/Maladies à Déclaration Obligatoire/Maladies Infectieuses/Dossiers Thématiques/Accueil. Available at: http://invs.santepubliquefrance.fr/Dossiers-thematiques/Maladies-infectieuses/Maladies-a-declaration-obligatoire/Tuberculose/Donnees-epidemiologiques. Accessed January 1, 2019.
- 7.↑
Chien K , Seemangal J , Batt J , Vozoris NT , 2018. Abdominal tuberculosis: a descriptive case series of the experience in a Canadian tuberculosis clinic. Int J Tuberc Lung 22: 681–685.
- 8.↑
Gill P , Coatsworth NR , Gundara JS , Hugh TJ , Samra JS , 2013. Tuberculosis: experience in a low endemic area Australian tertiary hospital. World J Surg 37: 984–990.
- 9.↑
Tan KK , Chen K , Sim R , 2009. The spectrum of abdominal tuberculosis in a developed country: a single institution’s experience over 7 years. J Gastrointest Surg 13: 142–147.
- 10.↑
Rosengart TK , Coppa GF , 1990. Abdominal mycobacterial infections in immunocompromised patients. Am J Surg 159: 125–131.
- 11.↑
Chen HL , Wu MS , Chang WH , Shih SC , Chi H , Bair MJ , 2009. Abdominal tuberculosis in southeastern Taiwan: 20 years of experience. J Formos Med Assoc 108: 195–201.
- 12.↑
Fillion A et al. 2016. Abdominal tuberculosis in a low prevalence country. Med Mal Infect 46: 140–145.
- 13.↑
Jehangir W , Khan R , Gil C , Baruiz-Creel M , Bandel G , Middleton JR , Sen P , 2015. Abdominal tuberculosis: an immigrant’s disease in the United States. N Am J Med Sci 7: 247–252.
- 14.↑
Ramesh J , Banait GS , Ormerod LP , 2008. Abdominal tuberculosis in a district general hospital: a retrospective review of 86 cases. QJM Mon J Assoc Phys 101: 189–195.
- 15.↑
Chou CH , Ho MW , Ho CM , Lin PC , Weng CY , Chen TC , Chi CY , Wang JH , 2010. Abdominal tuberculosis in adult: 10-year experience in a teaching hospital in central Taiwan. J Microbiol Immunol Infect 43: 395–400.
- 16.↑
WHO , 2014. Xpert MTB/RIF Implementation Manual: Technical and Operational ‘How-To’; Practical Considerations. Geneva, Switzerland: World Health Organization. Available at: http://www.ncbi.nlm.nih.gov/books/NBK254323/. Accessed January 2, 2019.
- 17.↑
Pandey S , Congdon J , McInnes B , Pop A , Coulter C , 2017. Evaluation of the GeneXpert MTB/RIF assay on extrapulmonary and respiratory samples other than sputum: a low burden country experience. Pathology 49: 70–74.
- 18.↑
Park SH , Yang SK , Yang DH , Kim KJ , Yoon SM , Choe JW , Ye BD , Byeon JS , Myung SJ , Kim JH , 2009. Prospective randomized trial of six-month versus nine-month therapy for intestinal tuberculosis. Antimicrob Agents Chemother 53: 4167–4171.
- 19.↑
Soni H , Bellam BL , Rao RK , Kumar PM , Mandavdhare HS , Singh H , Dutta U , Sharma V , 2018. Use of steroids for abdominal tuberculosis: a systematic review and meta-analysis. Infection 47: 387–394.